Spatial Data Intelligence and City Metaverse - A Review

JID: FMRE
ARTICLE IN PRESS [m5GeSdc;February 2, 2024;14:31]
Fundamental Research xxx (xxxx) xxx
Contents lists available at ScienceDirect
Fundamental Research
journal homepage: http://www.keaipublishing.com/en/journals/fundamental-research/
Review
Spatial data intelligence and city metaverse: A review

Xiaofeng Meng a, Yong Li b, Ke Liu c, Yu Liu d, Bin Yang e, Xuan Song f, Guoqiong Liao g,
Senzhang Wang h, Ziqiang Yu i, Longbiao Chen j, Xiao Pan k, Yuming Lin b,∗
a
School of Information, Renmin University of China, Beijing 100872, China
b
Department of Electronic Engineering, Tsinghua University, Beijing 100084, China
c
Department of Information Sciences, National Natural Science Foundation of China, Beijing 100085, China
d
Institute of Remote Sensing and Geographic Information Systems, School of Earth and Space Sciences, Peking University, Beijing 100871, China
e
School of Data Science and Engineering, East China Normal University, Shanghai 200062, China
f
Department of Computer Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, China
g School of Information Management, Jiangxi University of Finance and Economics, Nanchang 330013, China
h School of Computer Science and Engineering, Central South University, Changsha 410017, China
i School of Computer and Control Engineering, Yantai University, Yantai 264005, China
j
Department of Computer Science, Xiamen University, Xiamen 361000, China
k
College of Information Science and Technology, Shijiazhuang Tiedao University, Shijiazhuang 050043, China
a r t i c l e i n f o a b s t r a c t
Article history: Spatial Data Intelligence (SDI) encompasses acquiring, storing, analyzing, mining, and visualizing spatial data
Received 17 June 2023 to gain insights into the physical world and uncover valuable knowledge. These understandings and knowledge
Received in revised form 31 August 2023 play a crucial role in connecting physical and virtual realms, such as in developing a City Metaverse (CM) aimed
Accepted 20 October 2023
at enhancing and optimizing modern urban environments. The advancement of CM holds immense potential to
Available online xxx
benefit urban dwellers, making research on SDI an increasingly prominent area of focus. This paper contributes
significantly by organizing the relevant research and technologies within a coherent framework. Firstly, we iden-
Keywords:
tify SDI technologies capable of collecting real-world information to construct a virtual CM. Subsequently, we
Spatial data intelligence
City metaverse delve into the technologies that can be compositely integrated with SDI to facilitate interaction with and man-
Virtual-real interaction agement of actual cities from the virtual perspective. Additionally, we emphasize the effectiveness and potential
Application prospect of these methods in practical applications. Lastly, we conclude our survey by discussing emerging challenges
Smart city associated with technological progress, the industrial chain, legal and regulatory aspects, and ethical and moral
considerations.
1. Introduction to meteorological, population, and socioeconomic data, have amplified

the significance of SDI in comprehending the world we inhabit.
Spatial data [1] refers to the information that characterizes individu- On the other hand, the City Metaverse (CM) encompasses apply-
als, objects, and events within natural geographical and human activity ing the Metaverse concept specifically within urban contexts [2]. The
spaces. By primarily considering spatial data, the advancement of ar- CM can be described as a virtual and digital urban space that incor-
tificial intelligence (AI) technology across algorithms, data processing, porates urban data, models, and algorithms, enabling the simulation
and computing has contributed to spatial data collection and analysis, of diverse scenarios and synchronous interaction, also leverages vari-
culminating in the emergence of Spatial Data Intelligence (SDI). SDI ous technologies including AI, cloud computing, big data, blockchain,
involves collecting, storing, analyzing, mining, and visualizing spatial and virtual reality (VR). CM exhibits distinctive characteristics such as
data to grasp spatial information and reveal essential insights. Conse- being data-driven, facilitating authentic simulations, enabling intelli-
quently, the fundamental technologies of SDI encompass spatial sensing, gent decision-making, combining virtual and real elements, supporting
data storage, mining, and computing. The accumulation and digital human-computer interaction, spanning multiple domains, and prioritiz-
storage of the spatial data, ranging from geo-location and terrain data ing data security [3]. With such multidimensional capabilities, the CM
∗
Corresponding author.
E-mail addresses: xfmeng@ruc.edu.cn (X. Meng), liyong07@tsinghua.edu.cn (Y. Li), liuke@nsfc.gov.cn (K. Liu), liuyu@urban.pku.edu.cn (Y. Liu),
byang@dase.ecnu.edu.cn (B. Yang), songx@sustech.edu.cn (X. Song), liaoguoqiong@163.com (G. Liao), szwang@csu.edu.cn (S. Wang), zqyu@ytu.edu.cn (Z. Yu),
longbiaochen@xmu.edu.cn (L. Chen), smallpx@163.com (X. Pan), linyuming9@mail.tsinghua.edu.cn (Y. Lin).
https://doi.org/10.1016/j.fmre.2023.10.014
2667-3258/© 2023 The Authors. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY
license (http://creativecommons.org/licenses/by/4.0/)
Please cite this article as: X. Meng, Y. Li, K. Liu et al., Spatial data intelligence and city metaverse: A review, Fundamental Research, https:
//doi.org/10.1016/j.fmre.2023.10.014
JID: FMRE
X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx
demonstrates the vast potential in diverse applications such as urban • Metaverse Space Construction, involving metaverse model construc-
planning, traffic management, cultural tourism, urban management, an- tion, multi-source model fusion, and metaverse model verification.
ticipating smart cities, optimizing urban resource utilization, and bol-
These technologies are systematically organized within the frame-
stering the sustainability of urban life.
work of “sensing-calculation-mining-construction”. Such a framework
The rapid advancement in AI has sparked a notable collaboration
can be traced back to the early vision of smart cities [4], integrated
and interdependence between SDI and CM, resulting in a synergistic
sensors and electronics with databases, tracking, and decision-making
relationship between the virtual and real domains. On one hand, SDI
algorithms. With respect to future trends and the current state [5], we
catalyzes various technologies, establishing a pipeline to utilize spatio-
further extend the four-layer framework by Tong et al. [6] which con-
temporal data through the “sensing-calculation-mining-construction.”
sists of Data collection, Data transmission, Data processing and Applica-
In essence, this process can be perceived as a continuous abstraction
tion, to cover more related technologies that help SDI harness data from
from the “real” to the “virtual” domain. On the other hand, the CM origi-
real cities and effectively applies it within the context of the CM.
nates from virtual city simulations and digital properties using technolo-
Simultaneously, the technologies encompassing the CM are system-
gies such as VR, augmented reality (AR), and other synchronous inter-
atically organized in a reverse “virtual-to-real” order, exploring the key
actions. These technologies facilitate the connection of physical space
components involved and their respective functionalities. The organiza-
components through distributed systems and ultimately influence the
tion is as follows:
physical world through intelligent decision-making support. This repre-
sents a process that moves from the “virtual” to the “real” domain. • Smart City Simulation, which encompasses spatio-temporal data visu-
Despite the significant volume of research conducted in SDI, CM, alization, spatio-temporal dynamic simulation, and digital twin city;
and associated technologies, there remains a notable gap in compre- • Human-computer Interaction involves various modalities such as vir-
hensive surveys that effectively bridge the “virtual-real” connection be- tual reality, augmented reality, mixed reality, and brain-computer
tween these domains. Consequently, this review aims to fill this void by interfaces;
establishing a theoretical framework that elucidates the interaction be- • Distributed Technology includes blockchain, Internet of Things (IoT),
tween “virtual” and “real”. Additionally, this review seeks to summarize and Non-fungible tokens (NFTs);
the current status of core technologies in both SDI and CM, providing a • Smart City Decision Support incorporates concepts such as City Infor-
holistic understanding of recent advancements. Moreover, the practical mation Modeling (CIM), Virtual Geographical Environment (VGE),
application prospects and future challenges associated will be analyzed, and urban middle platforms.
offering insights into their potential implementation and identifying ar-
All these technologies are arranged in the order of “simulation-
eas that require further investigation. This review will provide valuable
interaction-control-application,” emphasizing the progressive process of
guidance for researchers and practitioners to develop SDI and CM tech-
influencing the real world. As a newly developed concept, there lacks a
nologies in real-world scenarios.
widely accepted framework of the city metaverse. Therefore, we drew
This review will be organized as follows: Section 2 will introduce
extensively on recent reviews. Among them, the inspiration to inte-
the framework and the method used in the survey. Section 3 would
grate digital twins with advanced technologies like IoT and blockchain
summarize key SDI technologies to fulfill the “real-to-virtual”, while
serves as the backbone [7]. Valuable surveys in related fields like Hu-
Section 4 would review the key technologies to support CM for “virtual-
manComputer Interaction (HCI) [8] and VGE [9,10] are used to fulfill
to-real” invention. After reviewing the key technologies, Section 5 will
the framework, and the very recent survey on the broad idea of meta-
illustrate the current and potential collaboration of SDI and CM, and
verse [11] also inspired a lot. Ultimately, we believe it is appropriate
Section 6 will look into the future direction and risks. Section 7 will be
to organize these related technologies in an order of increasing interac-
a brief conclusion.
tion degrees with the real world. By presenting the related technologies
under this framework, we highlight capabilities in manipulating urban
environments within the CM to ultimately impact real-world cities.
2. Framework and method
2.2. Relationship between spatial data intelligence and city metaverse
2.1. Theoretical framework and content
The proposed framework not only establishes a systematic order
As previously discussed, SDI is rooted in the physical world, while
(ǣreal-to-virtualǥ and ǣvirtual-to-realǥ) for organizing the technolo-
CM is centered on the digital space. Together, they create a synergis-
gies, but also introduces a novel perspective for cross-domain compar-
tic relationship that merges the virtual and real domains. This synergy
ison. For example, the 5G communication technology that facilitates
has inspired the development of a theoretical framework, as depicted
high-speed data transmission in “Data Transmission and Computing”,
in Fig. 1. Within this framework, SDI plays a crucial role as a key tech-
also serves as the technology infrastructure for “Distributed Technol-
nology and bridge, empowering the construction of the CM within the
ogy” including IoT [12] and blockchains [13]. Here, we adopt the con-
digital space.
text of “virtual-real integration” as the overarching framework, delving
In this review, our focus revolves around the fundamental concept
into the profound relationship between SDI and CM.
of integrating the virtual and real realms. Firstly, we will delve into SDI,
examining its distinct features, advantages, research status, and future • Smart City Decision Support & Spatial Intelligent Sensing: Both areas are
developments in four pivotal aspects, following the trajectory of “real- intricately linked to the physical urban space, representing the most
to-virtual” integration. These aspects include: tangible and “real” applications. Smart city decision support systems
like urban middle platforms directly build upon the data collected by
• Spatial Intelligent Sensing, encompassing spatial sensing, spatio- spatial sensing technologies [14], reflecting the direct collaboration
temporal database management, and high-precision spatio-temporal between SDI and CM.
mapping. • Distributed Technology & Data Transmission and Computing: Both areas
• Data Transmission and Computing, comprising 5G mobile communi- focus on the digital representations of physical space but with differ-
cation, spatial high-performance computing, edge computing, and ent emphases. “Data Transmission and Computing” cares about how
fog computing; data acquired in the real world can flow to the virtual realm, while
• Intelligent Data Mining, encompassing spatio-temporal rule mining, “Distributed Technology” cares about how virtual information and
abnormality analysis, correlation analysis, prediction and decision- instructions flow between physical devices [15]. Notably, the tech-
making; nologies involved in “Data Transmission and Computing”, such as
2
JID: FMRE
Fig. 1. The conceptual connection between SDI and SM.
5G and cloud computing, provide the foundation of distributed pro- In conclusion, SDI and the CM complement and reinforce one an-
cessing technologies like IoT [12]. other. The construction of CM relies on the data and technical support
• Intelligent Data Mining & Human-Computer Interaction: Both areas are provided by SDI, while the advancement of SDI benefits from the de-
intimately tied to human behaviors, respectively exhibited in physi- mands generated by the CM. Through their coordinated development,
cal and virtual spaces. Relying on informatization and virtualization, we can truly achieve the digital transformation and intelligent evolution
behavioral patterns observed in the virtual and real domains can mu- of cities, enriching people’s lives with greater convenience and aesthetic
tually inspire and complement one another [16]. appeal.
• Metaverse Space Construction & Smart City Simulation: Both areas are
directly associated with virtual city representations, showcasing the 2.3. Method and data source
most distinct “virtual” characteristics. The digital twin city con-
structed through SDI serves as the foundational platform for CM Despite significant advancements in relevant technologies, there is a
simulations [17], and the outcomes of simulations can be utilized noticeable dearth of literature and comprehensive reviews linking SDI
to calibrate the digital twin model. with CM. In the field of spatial data intelligence, there are a lot of re-
lated reviews concerning spatial data. But they are either too early to
The idea of linking SDI and CM stems from their interdependence cover the recent development of AI, such as Kopersk et al.’s survey on
and mutual promotion. We gain deeper insights into the intricate re- spatial data mining in 1994 [20] and Wang’s in 2005 [21], or not com-
lationship within the framework by dissecting and comparing these prehensive enough, such as Du et al.’s review that focused only on ma-
four focus areas. This analysis provides an orderly framework and of- chine learning algorithms, or Zhou et al.’s review [22] that targeted on
fers a novel perspective for cross-domain comparisons. Through this 3D spatial data. Alsaedi et al. [23] provided a comprehensive review
lens, we can examine the symbiotic relationship between SDI and of the fundamental components and characteristics of big spatial data,
CM, fostering a deeper understanding of their interplay and potential which inspired our framework but still failed to bridge the SDI with the
synergies. emerging metaverse. On the other hand, the CM field has some impor-
The extensive data derived from SDI proves invaluable to the CM. tant surveys on the whole picture of metaverse technologies like Ning
Firstly, SDI supplies static statistical and dynamic application data of ur- et al. [11], but have just started to examine their impact to the city [24].
ban life, including population, buildings, traffic, and the environment, Kusuma and Supangkat [25] reviewed the information technology that
providing a foundation for constructing CM. Secondly, SDI facilitates in- CM can use but did not provide a clear classification framework. Yaqoob
telligent management within the CM through analysis and processing. et al. [26] reviewed the benefits, technologies and future opportunities
Thirdly, SDI aids in building refined models of CM with precise spatial of metaverse in smart cities, but did not include the role of smart-city-
data from real-world structures. Finally, SDI enables intelligent interac- related technologies in promoting CM. It is very noteworthy that SE
tion within the CM through spatio-temporal visualization on terminal Bibri and Z Allam et al. carried out a series of studies [2,27–29] that
devices. Studies have sought to utilize SDI as a catalyst for advancing view metaverse as a virtual form of smart cities and analyzed the possi-
CM construction [204]. ble impact on platformization, governance, ethics and sustainability of
Conversely, the CM represents the possibilities offered by the virtual cities, which strongly support our idea to bridge SDI and CM together.
realm. Virtual representation of real space demonstrates the “virtual- However, their series of work paid more attention to the influence of
real integration” and provides a promising approach to timely adopting the social level and did not introduce the technology deeply enough.
urban changes for inclusivity and sustainability. Moreover, the CM fur- It can be seen that the existing literature surveys fail to fully reflect
nishes SDI with a new and in-depth data source through refined urban the synergy between the SDI and CM. To address this gap, this paper
models and vast virtual data for human behavior, simulation or genera- adopts a critical literature review approach to provide a comprehensive
tion. Virtual data facilitates the discovery of spatial data rules and allows overview of the field and various technological aspects. The primary
for intuitive simulation and decision support of diverse urban scenarios, focus of this review is to present an encompassing and informative
whose outputs can serve as inputs for iteration and evaluation of ur- perspective on the overall landscape and key research directions. The
ban planning, transportation planning, and architectural design [18]. data utilized for this review primarily consists of journal papers and
Furthermore, transformative technologies like VR and AR revolutionize conference papers about the key technologies discussed in Section 2.1.
the interaction between individuals and the city. Urban residents can Given the strong connection to industrial applications, grey literature
be immersed in virtual experiences, transcending spatial barriers and such as white papers, technical reports, and government documents
fostering inclusive urban living [19]. are also included. Notably, considering the emphasis placed on smart
3
JID: FMRE
Fig. 2. Key technology involved in the review on Spatial intelligent sensing.
city development as a national strategy in China and the significant crowd-sensing are focused on the integration and analysis of the
progress made in this domain, the literature surveyed encompasses digital footprints left by large-scale crowds, to establish reliable and
publications in both English and Chinese. semantic-rich representations of group behavior across spatial do-
mains [41]. The objective is to realize a mutually beneficial scenario
3. Spatial data intelligence boosts city metaverse for users, data providers, and application developers within the city
metaverse [42,43]. Addressing these challenges and advancing re-
3.1. Spatial intelligent sensing: data sources for virtual spaces search in these areas will facilitate the full potential of crowd-sensing
as a valuable source of spatio-temporal data in urban contexts.
• Intelligent Interpretation of Remote Sensing has gained significant at-
Served as the initial step, spatial intelligent sensing encompasses
the acquisition of spatial data using diverse technologies, subsequent tention due to the continuous advancements in higher temporal and
data processing and storage for initial analysis, and the visualization of spatial resolutions of captured imagery through satellites and drones.
data on maps. Therefore, the subsequent sections will provide a detailed As a result, the demand for efficient processing and understanding
exploration of the three fundamental aspects: spatial sensing technol- of massive volumes of remote sensing data prompts extensive explo-
ogy, spatio-temporal database technology, and high-precision spatio- ration of AI technologies [44]. Researchers have investigated into
temporal mapping (See Fig. 2). These technologies provide the basic the mining of multi-source, multi-resolution, and multi-scale remote
data and storage management methods for subsequent SDI and CM sensing data, trying to establish hierarchical scene analysis mod-
work, emphasizing the extensive and efficient update of data sources. els [45] and interpretation techniques for ground and object infor-
Their applications may be mentioned repeatedly in the other sections, mation [46]. Besides, research on AI processing has been conducted
especially in the space modeling of the metaverse, demonstrating com- in various fields including oceanography, land management, and
plex synergies between different fields. transportation, to provide technical support for applications such as
land and resources management, environmental protection, climate
change analysis, and national security considerations [47]. These
3.1.1. Spatial sensing
endeavors underscore the importance of intelligent interpretation
The very first step of spatial intelligent sensing is to gather large-
methods and unlock their potential for diverse real-world applica-
scale spatial data. Spatial sensing technology leverages a range of sen-
tions.
sors, including aerospace satellites [30], aircraft and drones [31], smart-
phones and mobile terminals [32], smart wearable devices [33], indus-
trial and household monitoring equipment [34], and wireless sensing 3.1.2. Spatio-temporal database
devices [35]. Considering its application in urban scenarios, it is also In the context of the three sensing methods discussed in
known as “urban sensing”. As spatial sensing is a collection of broad Section 3.1.1, spatio-temporal databases are needed for storage, query,
technologies, we will mainly focus on several aspects that have made and optimization of collected data, which should effectively support the
progress in recent years: 3D vision-based sensing, spatio-temporal data requirements of spatial and temporal data processing and analysis.
crowd-sensing, and remote sensing with intelligent interpretation. The spatio-temporal database is a technology designed to handle
data with both spatial and temporal characteristics, including geo-
• 3D Vision (3DV) is a multidisciplinary field encompassing computer graphic information, meteorological data, traffic data, and more. In re-
vision, computer graphics, and artificial intelligence. Its primary fo- cent years, there has been significant progress in developing storage and
cus is utilizing vision sensors to efficiently capture and analyze three- query technologies [48], particularly in creating new spatio-temporal
dimensional information from the real world. In recent years, there index structures that enable support for complex queries and fast re-
have been significant advancements in 3D sensor technologies such trieval [49]. Therefore, we will examine the advancement in index,
as Lidar [36] and depth cameras [37] with the help of AI. Recent ex- query and storage technologies of a comprehensive survey.
plorations in intelligent sensing systems [38,39] have covered theo-
retical frameworks and important techniques, including 3D Simulta- • Index optimization for 3D vision data: Index optimization is crucial in
neous Localization and Mapping (SLAM), point cloud processing, 3D achieving efficient data storage and querying due to the high spa-
target detection and tracking, 3D scene reconstruction, and dynamic tial dimensions and complex geometric structures inherent in 3DV
scene understanding. datasets. To address these challenges, spatio-temporal databases
• Crowd-sensing of Spatio-temporal Data refers to a technology that combine 3D index structures such as Octree, 4D R-tree, or KD tree
leverages distributed smart devices, such as smartphones, wearable with multi-scale representation and hierarchical storage techniques.
devices, and IoT sensors, within a crowd to collect and share data These index structures consider the spatial and temporal attributes
about the urban environment and social phenomena. However, no- of 3D vision data, enabling the hierarchical description of spatial
table challenges exist from the distinctive characteristics of the data relationships [50], facilitating applications such as 3D point cloud
and sensing methods employed, including data quality, privacy pro- data processing, spatial collision detection, moving object manage-
tection, user incentive mechanisms, energy consumption, and data ment, and trajectory data analysis. Furthermore, the spatio-temporal
fusion paradigms [40]. As a result, current research endeavors in database incorporates specific topological data structures, such as
4
JID: FMRE
the Half-Edge Data Structure, to support complex spatial queries • Map information update: As the environment is constantly evolving,
related to the topological and spatial relations of 3D models [51]. the ability to capture and reflect real-time changes in the map be-
Recent research has also focused on index optimization techniques comes crucial. The delay in updating map information can lead to
based on deep learning methodologies [52]. By leveraging deep neu- discrepancies between the actual environment and the representa-
ral networks, high-level representations can be extracted, enabling tion in the HST map. Ensuring timely updates to maintain the cur-
the construction of more compact and efficient index structures. rency of the map information is an ongoing challenge.
• Query optimization and quality management for crowd-sensing: Op- • Data privacy protection: Protecting user privacy while maintaining
timization is critical for spatio-temporal databases. The dynamic data quality and timeliness is a complex task. Striking a balance be-
nature of crowd-sensing data from many devices necessitates dis- tween data privacy and the need for accurate and up-to-date maps
tributed storage and computing frameworks, such as Hadoop and requires robust privacy protection mechanisms.
Spark. To enhance query performance and reduce computational
Fortunately, technological advancements such as crowd-sensing,
complexity, spatio-temporal databases employ techniques like grid
spatio-temporal data mining, and reinforcement learning have con-
indexing and trajectory clustering to reduce the query scope and
tributed to significant progress in addressing these challenges.
streamline the computational workload. Given the potential for
• Spatial scene modeling: The indoor-outdoor integrated scene mod-
errors, noise, anomalies, redundancy, and invalid data in crowd-
sensing datasets, data quality management is of paramount impor- eling [63] focuses on the seamless integration of indoor and out-
tance. Spatio-temporal databases implement quality inspection to door environments, involving techniques such as Multi-laser scan-
identify potential issues [53], error correction to rectify inaccura- ning simultaneous Localization and mapping (SLAM) techniques,
cies [54], data reconstruction to enhance completeness and consis- extracting and matching the structures in both indoor and outdoor
tency [55], and quality evaluation to assess the overall reliability scenes [64].
• Point cloud extraction: Point cloud extraction aims to extract relevant
and usefulness [56].
• Storage optimization for remote sensing images: Storage optimization information from the point cloud data. Efficient feature description
is another critical concern. Remote sensing images typically exhibit algorithms can improve the effectiveness of feature extraction [65],
high resolution and contain abundant spectral information, necessi- while independent object extraction algorithms [66] and feature
tating specialized techniques for efficient storage and rapid access. screening libraries automatically extract independent objects within
Spatio-temporal databases employ block storage and pyramid struc- the environment. Contextual feature extraction technology further
tures to organize the image data into blocks and utilize hierarchi- enhances the accuracy and efficiency of point cloud processing [67].
• Multi-platform large-scene fusion modeling: This technology integrates
cal structures [57]. Similarly, compression and indexing technolo-
gies, such as band selection or PCA, are employed for spectral data. data from multiple platforms to create a cohesive and detailed scene
These techniques reduce storage space while improving query by se- representation. Low-cost image sensors address blind spots and as-
lecting relevant spectral bands [58] or transforming the data into a sist in scene reconstruction. Urban appearance modeling is then
more compact representation [59]. On this basis, performance evalu- achieved through large-scale point cloud scene classification theory
ation tests encompass storage access, data division, index connection and the analysis of spatial topology relationships [68].
• Map information update: Begin with data collection using sensors, di-
structures, and query algorithms to identify system bottlenecks and
optimize efficiency. Recent advancements leverage machine learn- verse local multi-agents data must be effectively fused to create a
ing to automate the optimization of system parameters [60]. By high-precision global map [69]. This requires establishing protocols
extracting data features such as frequency, basic unit count, and and communication methods for data interaction between the data
spatio-temporal distribution, a mapping matrix is established to rep- center and the agents [70]. Additionally, coordination and task as-
resent the relationship between features and configuration param- signment methods are essential to ensure efficient cooperation be-
eters. This facilitates the creation of a trained system performance tween agents and the timely completion of tasks [71].
• Data privacy protection: Data privacy protection involves several key
model for automatic optimization.
technologies, namely differential privacy [72], encrypted comput-
ing [73], and anonymization processing [74]. Differential privacy
3.1.3. High-precision spatio temporal map
safeguards individual privacy while preserving statistical charac-
Considering the particularity of spatio-temporal database, a very nat-
teristics by introducing noise during data release and queries. En-
ural idea is to illustrate it on a precise map before further analysis,
cryption computing, including homogeneous encryption and secure
namely the High-precision Spatio Temporal map (HST map).
multi-party computing, ensures secure data processing and transmis-
HST map refers to a digital map that provides an accurate represen-
sion by performing computations on encrypted data. Anonymization
tation of the earth’s surface environment and its temporal changes [61].
processing protects privacy during data publishing and sharing by
Distinguished from conventional maps, HST maps encompass informa-
generalizing and suppressing sensitive information.
tion about the ground, underground, underwater, and air domains, en-
abling precise depiction of spatial alterations across different time in- Despite the challenges faced by HST maps, ongoing advancements
tervals. Diverse sensor technologies including GPS, lidar, cameras, and in technology are leading to the emergence of solutions and methods.
inertial measurement units are used to capture relevant data, and ad- These developments promise to enhance the quality, timeliness, and pri-
vanced algorithms such as computer vision and machine learning are vacy of HST maps, thereby enabling the delivery of convenient, safe, and
employed to process the collected information in time. The versatility efficient services.
of HST maps enables their utilization across various application scenar-
ios like autonomous driving, traffic simulation and automatic parking. 3.2. Data transmission and computing: operation guarantee of virtual
While HST maps offer immense application potential, they also spaces
present significant challenges [62]. These challenges can be categorized
into three main areas: Upon acquiring a large amount of urban data, the rapid transmission
and computation of such data has arisen as a pivotal concern. Address-
• Spatial scene modeling: The sheer volume of data makes efficient storing data transmission encompasses not only the refinement of conven-
age and retrieval challenging, and the processing complexity affects tional communication technologies but also underscores the importance
the performance of matching queries and semantic interpretation. of 5G communication technology. On the other hand, the advances in
Ensuring data quality and accuracy while managing the scale of the data computation are characterized by two major efforts: the facilita-
data remains a crucial challenge. tion of high-performance computations for spatio-temporal data, and
5
JID: FMRE
the architectural optimization through edge computing and fog com- and based on road networks. For moving object queries in Eu-
puting paradigms. Henceforth, we take these three aspects as examples, clidean space, such as range query, k-nearest neighbor query, and
considering their recent advancement and potential development in the reverse k-nearest neighbor query, the key issue is determining the
future. search area containing the target moving object. Scholars have pro-
posed various spatio-temporal index structures based on R-tree [85],
3.2.1. 5G mobile communication technology Quard-tree [86], KD-tree [87], Voronoi, and Grid. On the other
Spatial sensing in Section 3.1 will continuously send back a huge hand, the query of moving objects based on the road network is
amount of data from all city corners, which need fast, reliable and low- more complicated due to the need to calculate distance on the road
latency transmission. The emergence of fifth-generation mobile commu- network. In recent years, many works have studied the k-nearest
nication technology (5G) [75] has played a pivotal role in achieving it. neighbor query problem based on road networks, like SILC [88],
The new architecture and technology of 5G offer higher transmission ROAD [89], S-GRID [90], V-tree [91], G-tree [92], TOAIN [93],
rates (with peak rates up to 20 Gbps), lower latency (in the millisecond GLAD [94], TD-H2H [95], G∗ -tree [96], focusing on determining
range), improved reliability, increased network capacity, and broader the target moving object under the premise of traversing as few
coverage. The high speed and low latency of 5G provide support for road network vertices as possible. Besides, massive mobile objects
data transmission, while the stability and reliability of the network en- and high concurrent queries challenge the storage and computing
sure the seamless execution of data-related tasks. resources of a single computing node. Yu et al. [97] proposed a
distributed index structure DSI and a distributed query algorithm
• Spatial Data Transmission: Firstly, the high transmission rate of 5G DkNN, which can continuously split and merge according to changes
enables rapid data transfer, facilitating real-time and high-precision in moving object density. MPR [98] proposed a concurrent execution
applications within the city [76]. Secondly, the low latency of 5G mechanism for porting single-threaded query algorithms to multi-
is crucial for the timely collection, update, and analysis of real- core servers. Furthermore, some work proposed GPU-based paral-
time spatial data, particularly for applications like autonomous driv- lel query algorithms [99], which use GPU to construct the index
ing [77]. Thirdly, the enhanced network capacity of 5G supports a and search the target area in parallel to generate a candidate result
greater number of device connections, catering to large-scale CM set.
systems that require continuous data collection from millions of ve- • Route Query in Road Network Space. In real life, road network-
hicles and individuals in urban areas [78]. Fourthly, 5G incorporates oriented route query technology is an important demand of spatial
robust security measures, including authentication, encryption, and data high-performance computing, which is essentially the shortest
secure communication protocols, ensuring enhanced data transmis- route query of the graph. Early typical work employed a heuristic in-
sion security and privacy protection [79]. cremental expansion like the Dijkstra algorithm [100] and A∗ [101],
• Spatial Data Computing: Firstly, the substantial bandwidth offered by which use the greedy strategy to expand and visit the vertex. To
5G technology facilitates the collaborative processing of extensive further improve the efficiency, some work calculates and indexes
spatial data from multiple devices [77]. Secondly, the high comput- the shortest distance between some vertices in advance like Con-
ing performance and low latency of 5G enable real-time spatial data traction Hierarchies (CH) [102] which calculates the shortest dis-
analysis, facilitating rapid processing within the CM environment. tance bottom-up or top-down to form a hierarchical index. Based
Thirdly, 5G supports cutting-edge technologies like edge computing on the CH algorithm, relevant scholars have proposed the short-
and fog computing [80,81], allowing spatial data to be processed est route query algorithm based on the hub point label, such as
and computed at the data source. This approach reduces data trans- HL (Hub Labeling) [103], DHP [104], PHL(Pruned Highway Label-
mission costs and delays, alleviates the workload on centralized data ing) [105], BHP [106], and SHP (Significant path based Hub Push-
centers, and enhances overall computing efficiency. ing) [107]. When facing huge graphs and high concurrent queries,
In conclusion, 5G technology significantly contributes to the transit is more effective to use a parallel strategy. Pregel [108] is a dis-
mission and computing of spatial data, offering essential capabilities for tributed graph computing framework that supports graph parallel
the operation of virtual space. Its characteristics provide ample assur- computing. Li et al. [109] proposed a distributed multi-modal route
ance for the CM to deliver intelligent and efficient urban planning and query algorithm on large transportation networks with three dif-
management services. ferent categories and task instructions according to the query start
point or end point. Aridhi et al. [110] used the MapReduce model
to iterative divide the sub-graph for intermediate results. For the
3.2.2. Spatial high-performance computing
dynamic road network with ever-changing road transit time cor-
The efficient computation of spatio-temporal representations is cru-
responding to edge weights, CANDS [111] created a distributed
cial for dealing with large-scale spatial data and its dynamic evolution
stream processing platform, adopted a divide-and-conquer strategy
over time. Through deducing the time-space state of people, objects, and
to divide the dynamic graph into disjoint sub-graphs, and proposed
things of the physical world in the large-scale cross-modal framework
rules to minimize update operation caused by weight changes. Yu
(see Fig. 3), spatial high-performance computing can accelerate many
et al. [112] proposed a dynamic graph-oriented distributed top-k
applications in the real world. Here we summarized the most common
shortest route query algorithm, using the lower bound for the short-
computational needs, including object search that searches nearest ob-
est distance between the sub-graph, and decomposing the original
ject in spatial databases [82], route query that searches shortest paths
queries into local top-k query problems in multiple sub-graphs. Ped-
for navigation [83], and trajectory similarity that can be used in com-
ersen et.al [113] proposed a Time-dependent and Uncertain Con-
putational acceleration and clustering [84]. We selected these aspects
traction Hierarchies (TUCH) to support stochastic routing where the
because they are common but require computing resources and fast re-
travel time of roads is not only time-varying but also uncertain.
sponse, which motivate the algorithm to progress. There are also syn-
• Spatial Trajectory Similarity Calculation. The trajectory data of peo-
ergies between them since the index structure of objects can accelerate
ple and vehicles is important for clustering the information from the
subsequent route queries, and the similarity calculation often relies on
physical world. For centralized calculation, directly computing on
previous trajectory partition.
any two trajectories will incur a high computational cost. Therefore,
• Moving Objects Search in Two-Dimensional Space. How to realize pruning the search space by building an index structure is promis-
searches for large-scale moving objects like cars and humans is ing. Some work uses locally sensitive hashing technology to establish
an important topic of spatial high-performance computing. It can trajectory index and focus on the nearest k results [114,115]. Grid
be roughly divided into moving object queries in Euclidean space index can also speed up the process through clustering grid cells
6
JID: FMRE
Fig. 3. Schematic diagram of high-performance computing framework for spatial data.
for representative trajectories [116], or pruning in the time dimen- Shang et al. [125] proposed a Spark-based system with a global in-
sion with hierarchical grid index [117]. Besides, signature-based index deployed on multiple computing nodes, which is similar to the
dexing like Strain-join [118] by Ta et al., dynamic space divisions work of Yuan and Li (Table 1).
like GeoSAX [119], global and local indexing scheme [117,120] can
also help to filter pairs of trajectories before distance computation, 3.2.3. Edge computing and fog computing
thereby reduce global transfer cost and local computation cost. To Besides the algorithms in Section 3.2.2 that accelerate computing,
further improve the efficiency, distributed computing mode for al- the optimization of computing frameworks, like edge computing and
gorithms is also important. The k-means algorithm can be optimized fog computing, is of special importance to issues such as high latency,
by Par3PKM [121] on the Hadoop platform, by Cui et al. [122] on low throughput due to uneven load distribution, and bandwidth limita-
MapReduce, and under the coarse-grained Dynamic Time Warp- tions. As decentralized distributed computing technologies, edge com-
ing [123]. Similar to k-means, DPDBSCAN [124] proposed a dis- puting integrates network, computing, storage, and application capa-
tributed parallel clustering db-scan based on trajectory density parti- bilities on a platform situated close to the data source, enabling the
tioning. For similarity on the road network, DISON [120] was imple- provision of services at the edge segment. On the other hand, fog com-
mented on Spark with a two-level tree-structured global index, while puting is similar to edge computing but focuses on processing data in
7
JID: FMRE
Table 1
Spatial high-performance computing algorithm.
Algorithm Task Year Key features
R-trees [85] general index 1984 fast nearest neighbours queries

Quad-tress [86] general index 1984 variable resolution
KD-tree [87] general index 1975 fast queries and insert
SILC [88] object query 2005 almost linear precomputing and storing
ROAD [89] object query 2006 predetermined tree path to avoid costly network expansion
S-GRID [90] object query 2007 pre-computed data independent of the data points
V-tree [91] object query 2017 balanced search tree support dynamical update
G-tree [92] object query 2015 assembly-based method for queries
TOAIN [93] object query 2018 auto-tune shortcut-based index
GLAD [94] object query 2019 scheduling algorithms to avoid conflicts and improve throughput
TD-H2H [95] object query 2022 pre-computed weight functions
G∗ -Tree [96] object query 2019 shortcuts between selected leaf nodes
DSI/DkNN [97] object query 2014 distributed processing
MPR [98] object query 2019 schedule query and update on the cores
G_grid [99] object query 2018 GPU-accelerated with lazy update
Dijkstra [100] route query 1956 breadth-first search for shortest path
A∗ [101] route query 1968 heuristic of cost for shortest path
CH [102] route query 2008 hierarchy node contraction and bidirectional search
HB [103] route query 2002 distributed 2-hop covers of the shortest paths
DHP [104] route query 2013 breadth-first search with pruning
PHL [105] route query 2014 highway-based labeling with pruned
BHP [106] route query 2014 ordering and compression of hub label
SHP [107] route query 2017 heuristic path based ordering
PREGEL [108] distributed route query 2010 graph parallel computing
CANDS [111] distributed route query 2014 asynchronous answering and update
DTLP/KSP-DG [112] distributed route query 2020 distributed for dynamic graph and insensitive virtual path
Li et al. [109] distributed route query 2020 distributed on multimodal path
Aridhi et al. [110] distributed route query 2015 parallel solve on subgraph
TUCH [113] distributed route query 2020 time-varying, uncertain weight modeling and stochastic routing
LSH [114] trajectory similarity 2004 locality-sensitive hashing
E2LSH [115] trajectory similarity 2020 Geohash of domain POI and locality-sensitive hashing
SST [116] trajectory similarity 2020 synchronously matching, grid indexing and query partitioning
Strain-Join [118] trajectory similarity 2017 bi-directional mapping and signature-based similarity
MTSAX [119] trajectory similarity 2018 GeoWard dynamic coding and trajectory partition
DISON [120] trajectory similarity 2019 disjoint partitions by load balance and prune irrelevant
Tb-TS-Join [117] trajectory similarity 2018 search space pruning and parallel processing
Par3PKM [121] trajectory clustering 2015 MapReduce-based parallel three-phase k-means
Cui et al. [122] trajectory clustering 2014 MapReduce-based k-means
Hu et al. [123] trajectory clustering 2015 MapReduce-based coarse-grained Dynamic Time Warping
Wang et al. [124] trajectory clustering 2017 distributed parallel clustering on trajectory density partition
DITA [125] trajectory clustering 2018 global and local index with partition and cost-based balance
the fog computing layer closer to the edge devices. Currently, research • Distributed Spatio-temporal Machine Learning. Spatio-temporal data
in these areas is focused on the need for crowdsourcing and machine are usually provided by multiple service providers, which do not
learning. allow raw data sharing between providers. Therefore, data shar-
ing and collaborative computing on the premise that data does not
• Spatio-temporal Crowdsourcing in Edge Computing. With the rapid de- leave the local area inspire spatio-temporal federated learning with
velopment of mobile Internet and the IoT, traditional crowdsourc- the concept of “computing moves rather than data move”. Spatio-
ing has developed into a new service called spatiotemporal crowd- temporal federated learning conducts model training on multiple
sourcing, also known as mobile crowdsourcing. Spatial crowdsourc- decentralized edge devices and builds a general machine-learning
ing utilizes mobile Internet, online crowdsourcing platforms, and model without sharing data. Space-time federated learning can be
location services to connect crowdsourcing workers with tempo- divided cross-device and cross-silo, with participants being edge de-
ral and spatial attributes in the real world, enabling crowdsourc- vices (e.g. vehicle-mounted IoT devices and traffic flow monitoring
ing workers to actively or passively complete crowdsourcing tasks sensors) and enterprises(e.g. service providers of shared bicycles and
with spatio-temporal attributes, such as online taxi platform Didi online car-hailing) respectively. Ye et al. [130] proposed a selective
and Uber. Using edge computing and fog computing technology to model aggregation method that individually trains local deep neural
improve spatio-temporal crowdsourcing systems has become a re- networks using local data at the edge devices. Tong et al. [131] dis-
search hotspot. Zhang et al. [126] proposed an edge computing- cussed the Federated Range Aggregation (FRA) problems and pro-
based Bidirectional K-Nearest Neighbor Crowdsourcing Allocation posed an efficient range-aggregation approximation model. And
Protocol. Wu et al. [127] proposed a Weighted and Multi-Objective Zhang et al. [132] proposed federated adversarial domain gener-
Particle Swarm Combination to optimize multi-objective task assign- alization (FedADG) to equip federated learning with domain gener-
ment. On this basis, Zhang et al. [128] proposed an Online Task alization capability. Furthermore, Hu-Fu [133] is a spatio-temporal
Assignment across Regions based on Prediction algorithm with a data query processing system based on spatio-temporal data federa-
two-stage graph-driven bilateral assignment strategy to solve the tion, which decomposes the processing of spatio-temporal query into
Cross-regional Online Task problem. Furthermore, edge cloud com- plain-text operation and security operation. For online car-hailing
puting [129] unified with cloud computing in architecture and in- services, Tong et al. [134] designed a federal learning-to-dispatch
terface capabilities forms a complementary relationship and shows (Fed-LTD) framework, which achieves effective cross-dispatch by
good performance. sharing the scheduling model and decision-making.
8
JID: FMRE
Fig. 4. Main research content of intelligent data mining.
3.3. Intelligent data mining: a toolbox for characterizing virtual space such as weather data, by capturing temporal dependencies and mod-
eling sequential patterns. And the recent development of diffusion
Based on the capability of spatial sensing, transmission, and compu- model have show great abilities in generate data that resembles the
tation, analyzing urban data to obtain valuable information and knowl- real world [148].
edge is possible. Spatio-temporal data can be classified into event data, • Data visualization [149] is essential for effectively presenting spatio-
trajectory data, reference point data, and raster data [135], each repre- temporal data in a visually intuitive manner, aiding in the under-
senting a kind of urban data with specific spatial and temporal charac- standing of their characteristics and dynamic trends. Heat maps are
teristics. Considering such complexity among attribute features, spatio- employed to depict the distribution of population density in cities, al-
temporal features, and correlation differences, leveraging the power of lowing for a quick assessment of population concentration. Contour
AI and machine learning, i.e. intelligent data mining, is of great impor- maps, conversely, offer insights into terrain elevation and tempera-
tance. The core research areas of intelligent data mining, as illustrated ture variations, enabling the visualization of topographical features
in Fig. 4, involve the models and algorithms for rule mining, abnormal and thermal distributions.
analysis, correlation analysis, and prediction and decision-making tasks.
3.3.2. Spatio-temporal abnormal analysis
3.3.1. Spatio-temporal rule mining Unlike the mining task in Section 3.3.1 for the rule under normal cir-
Spatio-temporal data exhibits notable characteristics such as spatio- cumstances, spatio-temporal data anomaly analysis aims to uncover the
temporal correlation, multidimensionality, large volume, dynamics, and causes and assess the impact of anomalies, through identifying and ana-
uncertainty. These properties pose challenges for data mining and call lyzing entities that deviate from the expected distribution. Based on the
for utilizing AI and machine learning techniques to uncover patterns combined space-time relationship, spatio-temporal data anomaly anal-
and extract valuable insights. The primary approaches for intelligent ysis can be categorized as follows:
spatio-temporal rule mining include:
• Spatial data anomaly detection: This method uncovers deviations from
• Traditional spatio-temporal statistical models, such as the historical av- the universal spatial patterns in a small portion of the data, revealing
erage and time series models [136] like moving average, ARMA, and the unique laws of geographic phenomena or processes. Represen-
ARIMA models, analyze and model temporal and spatial variables tative techniques include the distance-based method [150], cluster-
based on statistical principles and assumptions. based method [151], density-based method [152], and graph-based
• Domain understanding and physical modeling involves physical pro- method [152]. Spatial data anomaly detection aids in discovering
cess like levy flight [137], point process [138], or collective mo- abnormal points and patterns in spatial data during the construction
bility model [139] in the modeling or clustering process [140]. By of the CM.
capturing the inherent causalities within the data, this approach en- • Time series anomaly detection: This approach identifies abnormal pat-
hances the interpretability of the derived insights [141] through terns in time series data by considering periodicity, trend, and ran-
methods including dynamic knowledge-based methods or knowl- domness. Representative methods encompass statistical techniques
edge graph [142]. (e.g., hypothesis testing, ARMA model, ARIMA model), similarity-
• Machine learning models offer a range of techniques for various tasks based methods (e.g., KNN algorithm, LOF algorithm), and deep
in spatio-temporal data analysis. Classification methods, such as sup- learning-based methods [153,154].
port vector machines and random forests [143,144], are used for cat- • Spatio-temporal data anomaly detection: This method combines spa-
egorizing data into classes. Regression methods, including linear re- tial anomaly detection with time series anomaly detection to iden-
gression prediction and XGBoost [145], are employed for predicting tify spatio-temporal anomalies. Given the complex characteristics,
numerical values. Clustering methods [146] encompass hierarchy- diverse anomalies, and scarcity of samples, detecting anomalies in
based, partition-based, and density-based approaches, identifying spatio-temporal data is challenging and heavily relies on machine
meaningful groups within the data. learning and AI [155].
• Deep learning models [147], particularly Convolutional Neural Net-
works (CNNs) and Recurrent Neural Networks (RNNs), offer ad- Anomaly detection techniques are valuable for identifying and un-
vanced capabilities for spatio-temporal analysis. CNNs are well- derstanding irregularities in spatio-temporal data, enabling proactive
suited for processing spatio-temporal data represented in image for- measures in the CM construction process. These methods enhance data
mat, allowing for efficient feature extraction and pattern recogni- quality, anomaly detection accuracy, and the overall reliability of the
tion. On the other hand, RNNs are adept at handling time series data, CM’s infrastructure and services.
9
JID: FMRE
3.3.3. Spatio-temporal correlation analysis • Data-driven spatio-temporal decision-making: The visualization of

In addition to addressing normal and abnormal circumstances, spe- spatio-temporal knowledge graphs is pivotal for decision-making
cific associations between entities, such as proximity or causation, are capabilities [156,167]. Using graph mining algorithms, valuable
of great value. Spatio-temporal association analysis explores the depen- information is extracted and displayed through graph representa-
dencies and interactions among entities, delving deeper into the tempo- tion [168], such as the logistics decision optimization platform used
ral changes and spatial interactions between objects, providing valuable by Cainiao [169].
insights for decision-making [156]. • Interpretable spatio-temporal prediction and decision-making: Deep
Traditional association analysis primarily relies on association rules, learning techniques alone lack interpretability. Spatio-temporal
which first mine spatial association rules and then incorporate tempo- knowledge graphs integrate spatio-temporal data and knowledge to
ral association constraints, or vice versa [157]. However, these methods gain a better understanding of spatio-temporal rules [161,165,170].
are conducted independently, neglecting the spatio-temporal coupling Combining AI with knowledge graph technology greatly enhances
of data. Recently, the powerful feature learning capabilities of deep interpretability and provides more insights, and Liu et al. [171] pro-
learning have been leveraged to automatically extract spatio-temporal posed a versatile UrbanKG for prediction and decision-making.
correlation features from the data [147]. Consequently, recent studies • Deep reinforcement learning: Deep reinforcement learning provides a
have combined recurrent neural networks with convolutional neural path that has not been fully explored, that is, to directly model and
networks [158] and graph neural networks [159], enabling simultane- predict spatio-temporal data through trial-and-error learning with
ous convolution operations on the spatio-temporal dimensions of the agents. Using existing spatio-temporal data as expert knowledge,
data. deep reinforcement learning can learn knowledge from data very
By conducting spatio-temporal correlation analysis on real-world effectively [172]. Other decision-making tasks includes traffic sig-
data, we can gain a better understanding of user behavior and activities nal [173] or navigation [174].
in the city [160]. Incorporating spatio-temporal correlation analysis into
the CM construction can create a more realistic, intelligent, and inter- 3.4. Metaverse space construction: the entrance to the virtual space
active virtual reality experience. This integration enhances the overall
user experience and facilitates more applications. Completing the series of tasks to obtain information about the real
city, one can finally construct a digital space in the virtual realm, which
3.3.4. Spatio-temporal prediction and decision-making
is nowadays called Metaverse. The construction of metaverse spaces
Spatio-temporal prediction involves forecasting the future changes
serves as the gateway to the virtual realm, providing the foundation
in geographical events or phenomena in both time and space. Spatio-
of immersive experiences, interactions, simulations, and augmented re-
temporal decision-making, on the other hand, entails selecting the opti-
ality overlays. The metaverse space serves as the container of virtual
mal solution based on analysis results [161]. As the complexity of fore-
content and the platform of virtual applications, which is expected to
casting and decision-making continues to rise, traditional methods have
be close to real but easy to adjust. Here we discuss model construction,
struggled to meet the demands of decision-making processes. The ad-
model fusion, and model verification technologies, following a sequen-
vent of AI has significantly improved data processing efficiency, en-
tial process to realize the CM model (See Fig. 5).
hanced prediction accuracy, and enabled intelligent decision-making
optimization [162]. Consequently, AI-based prediction and decision-
making have become the prevailing approach [161,163]. 3.4.1. Metaverse model construction
The first step of a CM is to establish a digital model, which finds ap-
• Domain understanding and physical modeling: By amalgamating inher- plications in urban planning, simulation, and analysis. In recent years,
ent domain insights and empirical laws, this approach fortifies the the great abundance of data shifted research focus towards multi-source
reliability and interpretability of predictions and decisions. By inject- data. Souza L [175] provides a comprehensive overview of CM model
ing the inherent comprehension of the domain’s intricacies into the construction technology, categorizing the current urban model construc-
transformative capabilities of AI, methods like differentiable deci- tion approaches into the following categories according to their data
sion trees [164] or knowledge graph [165] can yield more informed source:
and contextually sound outcomes.
• Data-driven spatio-temporal prediction: Deep learning models like RNN • LiDAR model: This approach involves acquiring 3D point cloud data
and GNN possess powerful automatic feature learning capabilities of urban scenes using LiDAR technology. The data is then processed
in the spatio-temporal domain [147], which can be fine-grained to extract features such as position, size, and direction, which are
to achieve accurately driving styles recognition [166]. They enable used to construct a city model [176].
real-time dynamic analysis and prediction of massive heterogeneous • UAV high-altitude aerial photography: Utilizing UAVs equipped with
spatio-temporal data, facilitating accurate forecasts of medium and visual recognition technology, high-altitude aerial photographs are
long-term characteristics. taken to capture overlapping image areas. These images can be used
Fig. 5. Key technology involved in the review on Metaverse Space Construction.
10
JID: FMRE
for high-precision mapping and reconstruction [177]. For example, 3.4.3. Metaverse model verification
Google has employed this technology to build 3D city models in Model verification technology plays a crucial role in assessing the
Google Maps. accuracy and reliability of the established model from Sections 3.4.1 and
• Satellite remote sensing: This approach involves capturing high- 3.4.2.
resolution remote sensing images through satellites. Advanced deep One commonly employed method is simulation-based verification.
learning models [178] are now applied to extract ground object in- By constructing a virtual urban environment and simulating its devel-
formation, which is used to construct urban models. opment process under different scenarios, the accuracy and reliability
• Measuring instruments and sensors: Various physical information of the model can be assessed [49]. Furthermore, researchers can utilize
about urban scenes is obtained using measuring instruments and Virtual Reality (VR) technology to present the virtual city to users, al-
sensors. For instance, the city government of Munich uses sensors lowing them to experience and evaluate its practicality and feasibility.
to gather air quality data [179], which is utilized to simulate the This approach necessitates a comprehensive simulation platform, along
city’s meteorological environment. with various parameters and rules, serving as fundamental components
• Social media and crowdsourcing: This approach involves collecting ur- for establishing and optimizing the CM.
ban scene information from the public through social media, mobile During the simulation and verification process, various elements
Internet, and crowdsourcing platforms.. As an example, researchers within the city can be simulated to observe their movement, interaction,
from MIT have developed Treepedia [180], which constructs mod- and impact. By comparing and validating these simulations with real-
els of green coverage in major cities worldwide using photos and world observations, the quality and accuracy of the City Metaverse (CM)
location information uploaded by the public. can be further optimized. Simulation-based model verification methods
have gained popularity in recent years. For instance, a study simulat-
In summary, the selection of city model construction technology
ing traffic congestion in New York demonstrated that the CM could ac-
in the CM should be based on specific application scenarios and data
curately reflect the actual urban traffic situation [185]. Another study
sources. Each technology has its strengths and suitability for different
verified the feasibility of shared transportation in reducing environmen-
purposes. Choosing the appropriate technology for a given application
tal impact by simulating factors such as air quality and environmental
scenario is crucial in constructing the CM.
noise within the city [186].
Additionally, there are other model validation methods based on
3.4.2. Multi-source model fusion data analysis [187]. These methods rely on historical data to evaluate
Various data sources in Section 3.4.1 can all provide models at dif- the prediction accuracy and applicability of the model. Such approaches
ferent scales and resolutions, therefore how to integrate them should be are particularly valuable in traffic or climate models, where historical
carefully examined. Multi-source model fusion integrates models from data can be used to validate the model’s effectiveness.
diverse data sources to create a unified CM model, aiming to leverage However, when applying validation methods based on data analysis,
the strengths of different data sources for comprehensiveness and accu- certain considerations must be considered. Firstly, the data quality is
racy. The major approach includes the integration at the data level, the crucial, emphasizing the need for data cleaning and preprocessing. Sec-
model level, or afterward verification. ondly, selecting appropriate evaluation metrics should reflect the accu-
• Fusion technique based on data [181]: This approach utilizes data fu- racy and applicability of the model. Finally, the analysis and interpreta-
sion methods that consider factors such as weight, quality, or trust to tion of the results are essential to identify the strengths and weaknesses
integrate data from different sources. It aims to ensure the integrity of the model, providing valuable feedback for refining the CM.
and consistency of urban data by forming multi-level and multi-angle
representations. For example, Jia et al. [182] achieved precise urban 4. City metaverse empowers spatial data intelligence
area extraction and model establishment by fusing multi-source re-
mote sensing image data. 4.1. Smart city simulation: a city running in virtual space
• Fusion technique based on model integration: This technique com-
bines models from different data sources using similarity, associa- With the technical support of Section 3, we can extract and construct
tion, and fusion rules. By leveraging these integration methods, the a metaverse space from a real city, and apply the power of digital tech-
strengths of individual models can be combined to enhance the over- nology to benefit our real world. The first step is to provide a virtual
all model accuracy. For instance, [183] employed a random forest- representation of the city and simulate various components and behav-
based model integration algorithm to accurately monitor changes in iors through smart city simulation. Such virtual representation, often
urban objects. called a digital twin, allows urban planners, policymakers, and stake-
• Fusion technique based on model verification [184]: This approach in- holders to analyze and optimize the city in a simulated environment.
volves using different verification criteria, such as accuracy or ro- In the following sections, we will delve into the key aspects of smart
bustness, to evaluate the urban model. Various verification algo- city simulation, focusing on spatial-temporal data visualization, spatial-
rithms, including trust fusion, hierarchical fusion, and Bayesian net- temporal dynamic simulation, and the digital twin city in order accord-
work, can be utilized to assign weights and optimize multiple objec- ing to the depth of the simulation, which aims to create accurate and
tives. interactive virtual replicas of cities together (See Fig. 6).
Fig. 6. Key characteristics of the advanced technologies involved in the review on Smart City Simulation.
11
JID: FMRE
4.1.1. Spatio-temporal data visualization leverage large-scale datasets to learn patterns and rules of mobility, en-
Spatio-temporal data visualization, especially interactive explo- abling the simulation of realistic movements and interactions within a
ration and dynamic demonstration, is the simplest form to present com- city [196,197].
plex spatio-temporal data visually [188]. Due to its multi-source, multi- Moreover, the availability of high-resolution data has significantly
dimensional, interactive, and dynamic nature, it usually serves as the contributed to our understanding of crowd behaviors in densely popu-
output of smart city simulation. lated areas [198]. Analyzing this high-resolution data allows a compre-
The main advantages of spatio-temporal data visualization technol- hensive understanding of crowd dynamics [199]. Recent developments
ogy include the following aspects: focused on capturing the daily trajectories and behaviors of individuals,
which can uncover valuable insights into trajectories [172,200], activ-
• Intuitiveness: Spatio-temporal data visualization is designed to
ity patterns [201], and the correlation between spatial and temporal at-
present abstract spatio-temporal data in front of users through in-
tribute [148]. Integrating data-driven approaches, AI algorithms, and
tuitive visualization.
high-resolution datasets has revolutionized spatio-temporal dynamic
• Interactivity: Spatio-temporal data visualization supports indepen-
simulation, enabling more accurate and detailed representations of ur-
dent selection of data dimensions, attributes, and granularity, en-
ban dynamics. These advancements have far-reaching implications for
abling quick and accurate obtaining of the required information.
urban planning, transportation management, and emergency response.
• Real-time performance: Spatio-temporal data visualization support
The advantages of spatio-temporal dynamic simulation lie in:
displaying the spatio-temporal changes of data through dynamic
demonstration to better meet the needs of real-time data analysis. • High simulation accuracy: Based on actual data and physical laws,
• Globality: Spatio-temporal data visualization can integrate spatio- complex urban change processes like population migration, traffic
temporal data from different sources to obtain more comprehensive congestion, and climate change can be simulated, which is of value
and accurate data. to high-precision decision support for urban planning and manage-
ment.
Currently, spatio-temporal data visualization finds applications in
• Better visualization: The process of urban spatio-temporal dynamic
diverse fields, including map visualization [189], traffic visualiza-
change can be presented in simulation so that decision-makers and
tion [190], and epidemiology visualization [191]. Advancements in sen-
the public can have an intuitive understanding, which is of great
sor technology and wireless communication enable the acquisition and
significance in promoting urban democratization and transparency.
processing of large-scale, high-resolution spatial data. Looking ahead,
• Strong decision-making support: The urban change under different cir-
spatio-temporal data visualization technology is expected to progress in
cumstances can be simulated to provide a comparison and selection
the following directions:
of various schemes, thereby decision-makers can evaluate the pros
• Intelligence: Emerging AI-based visualization technologies are ex- and cons, and formulate more reasonable and effective strategies.
pected to automatically process and analyze large-scale spatio-
temporal data and be able to choose the best visualization form. 4.1.3. Digital twin city
• Interaction: Advancement of HCI will enhance visualization interac- With the help of visualization and simulation, the digital twin city
tion, and allow more free exploration through gesture recognition, creates digital replicas of real cities for intelligent management, focus-
speech recognition, AR/VR, etc. ing on real-time synchronization with the actual urban conditions. It
• Integration: As the data from the city metaverse would be multi- provides decision support for urban planning, design, operation, and
source and multi-modal, effective and efficient data integration will management by updating, simulating, and analyzing real-time situa-
become the key to spatio-temporal data visualization. tions [202]. This technology is a significant trend in smart city research
In conclusion, spurred by the demands of constructing and utilizing and finds applications in urban planning, traffic management [203], en-
the city metaverse, spatio-temporal data visualization is set to undergo vironmental protection [204], and disaster response.
rapid development and innovation. The development of digital twin city technology has several advan-
tages.
4.1.2. Spatio-temporal dynamic simulation • Digital twin city can provide high-precision, high-resolution urban data
Visualizing virtual cities alone does not provide enough insights that that comes directly from the real-time collection during the opera-
drive virtual cities. Spatio-temporal dynamic simulation [192] involves tion, which are beneficial for decision-making management. And the
simulating various events in urban space, such as dynamic traffic, hu- simulation in a digital twin city can avoid potential risks such as un-
man mobility, urban disasters, and urban energy, which is the subse- reasonable systems and models, thereby ensuring the authenticity of
quent step for digital twins. the simulation effect through interconnected urban systems.
The key components of spatio-temporal dynamic simulation include • Digital twin city can realize real-time monitoring and intelligent regu-
establishing urban models, formulating simulation strategies, determin- lation. With the help of AI models, real-time detection of road traffic
ing model parameters, conducting simulation experiments, and ana- flow can provide control of traffic signals. Similar methods can also
lyzing results. For detailed information on establishing urban models, be applied in environmental detection and crime prevention.
please refer to Section 3.4. When formulating simulation strategies, con- • Digital twin city can increase the participation and awareness of ur-
sider factors such as the target of the simulation, spatio-temporal scope, ban residents, which promotes urban democratization and commu-
and specific goals and requirements [193]. Determining model param- nity participancy. Digital twins in virtual spaces solve the problem
eters is crucial and should be done based on experience or experimen- of physical isolation of urban areas, enable a deeper understand-
tation. Following simulation experiments, it is important to visualize, ing across different kinds of borders, and provide powerful tools for
quantify, and analyze the results to draw meaningful conclusions and democratic urban governance.
provide decision-making recommendations.
Thanks to the availability of vast amounts of data and advancements It is important to acknowledge that the digital twin city technol-
in AI, spatio-temporal dynamic simulation is experiencing remarkable ogy still faces several challenges. Firstly, there is a need to enhance
improvements in accuracy and utility. Integrating AI algorithms with the quality and accuracy of urban data to improve credibility and real-
comprehensive datasets has unlocked new possibilities in simulating hu- ism. Secondly, the processing and computation of large-scale data pose
man mobilities within urban environments [194]. Deep learning mod- challenges for real-time updates and evolution. Thirdly, ensuring data
els, in particular, have demonstrated exceptional capabilities in captur- privacy and security, particularly with sensitive human data, must be
ing and reproducing complex human behaviors [195]. These models emphasized before advancing the digital twin city.
12
JID: FMRE
information hierarchically, which is crucial for generating content

that closely resembles real-life experiences [207].
4.2.2. Augmented reality

Unlike the full immersion of virtual environment in VR, Augmented
Reality (AR) overlaps virtual objects onto real-world environments, in-
cluding computer-generated images, sounds, 3D models, video, graph-
ics, animation sequences, games, and GPS. AR supplements reality
rather than providing an alternate reality, beginning with collecting
real-world data through cameras and sensors. The user’s spatial posi-
tion is continuously updated in real-time using cameras, gyroscopes,
and other accessories. This information is used to calculate relative po-
sitions and fuse virtual content, resulting in a synthesized video presen-
Fig. 7. Schematic diagram of Human-computer interaction for city meta- tation [208].
verse. As a crucial technology in the city metaverse, AR blends digital visual
effects with the real environment and is accessible through smartphones
and other digital devices.
4.2. Human-computer interaction: breaking the barriers between virtual
To support real-time AR operations, an efficient data transmission
and physical spaces
network like 5G (as discussed in Section 3.2.1) is essential. This network
enables both precise and approximate AR services. Precise AR detects
As a digital reality, CM can reconstruct and simulate the physical
all potential objects for an immersive AR experience, while approxi-
city. But without engagement with human interaction, the CM is nei-
mate AR focuses only on visually salient objects to reduce computation
ther self-completed nor adequate for practical use, highlighting the sig-
and communication overhead. Recent research proposes a self-adaptive
nificance of HCI technology (Fig. 7). The HCI process comprises three
AR services framework that adjusts to different network conditions and
main components: the real world, the virtual world, and the metaverse
computing capabilities. The AR service provider allocates computing re-
engine [205].
sources based on environmental information provided by AR users.
The real world component encompasses users who interact with the
Object detection [205] is another important aspect of AR, as it re-
metaverse through devices, while the virtual world component consists
quires accurate identification and localization of real-world objects for
of virtual characters and environments that respond to user input and
virtual object projection. Recent advancements in graph neural net-
exhibwit dynamic behavior. The metaverse engine facilitates interac-
works show promising potential in this area [209].
tions between the virtual and real worlds using technologies such as the
brain-computer interface, augmented reality (AR), virtual reality (VR),
4.2.3. Mixed reality
and mixed reality (MR). In the following sections, we will delve into the
Combining the advantages of VR and AR, Mixed Reality (MR) merges
concepts of VR, AR, MR, and the brain-computer interface.
real and virtual, aiming to create a unified space where real and virtual
objects coexist and interact in real-time. The goal is to seamlessly blend
4.2.1. Virtual reality the physical and digital worlds, allowing users to perceive and interact
Virtual reality (VR) is an immersive technology that combines com- with both simultaneously.
puter graphics systems and interface devices to create interactive 3D Virtual Reality, Augmented Reality, and Mixed Reality are related
environments. Using a VR device, the user’s visual range is expanded concepts with different display approaches (see Fig. 8). VR immerses
through a convex lens, while a gyroscope tracks the user’s head move- users in a fully computer-generated environment, disconnecting them
ments. The screen is continuously refreshed in real-time, allowing users from the real world. AR overlays virtual content onto the real world,
to experience a 360-degree, three-dimensional space, resulting in a enhancing the user’s perception of reality. MR combines elements of
highly immersive visual environment [206]. both VR and AR, merging the real and virtual worlds and allowing for
Virtual reality is characterized by immersion, interaction, and imag- seamless interaction between them [210].
ination, achieved through its three main components: hardware, soft- MR poses greater demands on perception and display compared
ware, and content. to VR and AR, resulting in additional challenges such as model cal-
ibration [211]. Achieving precise calibration is crucial for MR tech-
• Hardware encompasses input and output devices. Input devices can nology, including accurate initial calibration of hand-eye coordination
be hand-based or non-hand-based. Hand-based input includes VR and effective real-time calibration to ensure long-term accuracy. Re-
handles, VR gloves, and gesture input devices, with handles being cent research [212] has introduced a real-time latent active correction
simple and easy to use, while gloves offer more advanced motion
capture capabilities. Non-hand-based input includes eye tracking,
motion tracking, and voice input. The primary output device is a
head-mounted display that blocks the user’s sight to enhance the vi-
sual experience.
• Software creates virtual environments and objects to deliver an im-
mersive experience. It can reflect the real world or create imagi-
nary environments. Real-world reflection replicates existing physical
environments, such as digital twins (see Section 4.1.3). Imaginary-
environment creation involves painting, 3D modeling, or deep learn-
ing methods. Sound plays a vital role in setting the atmosphere and
enhancing the sense of presence, complementing the visual scenes
and objects.
• Content refers to the events, tasks, and experiences within the city
metaverse, created by humans or machines. Deep learning sys-
tems have demonstrated their ability to represent and combine Fig. 8. Differences of VR, AR and MR.
13
JID: FMRE
Fig. 9. Schematic diagram of IoT and Blockchain distributed technology.
algorithm for optical perspective and hand-eye coordination, which while blockchain provides data storage and security. Together, they sup-
helps mitigate error accumulation in hand-eye calibration. port the operation and control of the real city in the CM.
The conceptual diagram in Fig. 9 illustrates the integration of
4.2.4. Brain-computer interface blockchain and IoT, consisting of the physical layer (IoT-enabled phys-
In addition to transmitting information through visual means like ical objects), the connection layer (digital representation generated
AR or VR, the Brain-Computer Interface (BCI) enables direct commu- through IoT and corresponding NFT), and the blockchain layer (stor-
nication and information exchange between the brain and external de- age of relevant information using blockchain technology).
vices. The process can be divided into three steps: detection, analysis,
and control. Initially, specialized devices detect human brain activity,
such as brain waves or magnetic fields. Subsequently, these signals are 4.3.1. Internet of things
analyzed and processed to identify instructions from the brain. Finally, The Internet of Things (IoT) is a network that connects objects
the extracted instructions serve as input for controlling devices, facili- through information-sensing devices to enable their identification and
tating interaction between humans and computers. management. In the context of the city metaverse, IoT technology plays
BCI can be categorized into invasive, semi-invasive, and non-invasive a crucial role in collecting diverse data from urban sensors and deliver-
methods [213]. Invasive BCIs involve implanting electrodes into the ing operation instructions to devices. Compared to 5G which is mainly
cerebral cortex, while semi-invasive BCIs place electrodes in the cranial a communication technology in Section 3.2.1), IoT focuses more on the
cavity but outside the cerebral cortex. Non-invasive BCIs collect EEG heterogeneous, continuous, multi-dimensional, and multi-sourced con-
signals using wearable devices attached to the scalp. Invasive BCIs pro- nections between devices, which involves embedded devices, network
vide accurate data but pose surgical and tissue rejection risks, whereas architecture and application development. There is no doubt that 5G
non-invasive BCIs avoid safety risks but capture weaker signals. communication technology can greatly promote the progress of IoT, and
BCI holds promise as an output technique in the city metaverse. While such a close connection proves the synergy between SDI and CM.
conventional devices like speakers, headphones, screens, and VR/AR The architecture of IoT can be divided into four main components:
devices provide limited immersive experiences for hearing, vision, and
touch, BCI can directly transmit images and sounds to the brain, enhanc- • The perception layer, also known as the physical layer, includes sen-
ing auditory and visual interactions in real-time. Although currently sors, actuators and other devices to collect various information from
used primarily in the medical field, BCI has the potential to shine in the surrounding environment. These devices then send the collected
the metaverse. information to the network layer.
BCI also enhances input flexibility by enabling users to control ex- • The network layer, also known as the transport layer, forwards col-
ternal objects through encoding and decoding brain waves. In an ideal lected data from physical objects to information processing systems
scenario, EEG-based BCI allows users to control virtual characters in the through wired or wireless means such as WiFi, Bluetooth, or infrared.
city metaverse through thoughts and imagination. Existing applications • The platform layer, is linked to a database and responsible for service
such as the P300 speller [214] enable text input using EEG, and further management and data processing by technologies such as virtualiza-
research can enhance accuracy and efficiency. tion and cloud computing.
BCI research has made significant strides, but it still faces major • The application layer, is the interface between the IoT system and
challenges. The primary challenge lies in accurately deciphering human users that provides intelligent and corresponding management for
intentions from brain signals, which are often characterized by a low logistics, medical care, and urban construction.
signal-to-noise ratio. Brain signals are easily influenced by biological
and environmental factors, and their non-stationary nature makes in- IoT, as a distributed technology and a prerequisite for SDI, faces sig-
formation extraction difficult. Although various preprocessing and fea- nificant challenges in terms of security and privacy. At the perception
ture engineering techniques in the time and frequency domains have layer, the transmission of multi-sourced information carries the risk of
been developed, they tend to be time-consuming and prone to distortion. fake data and malicious tampering in public environments. To address
Moreover, feature engineering heavily relies on human expertise, lim- these challenges, targeted algorithms and authentication mechanisms
iting generalizability. Although AI-based approaches have shown some like Privacy-Preserving Data Publishing (PPDP) [216] are necessary. At
progress, the dynamic nature of human thinking poses difficulties in pre- the network layer, the efficient transmission speed exacerbates the prob-
cise classification. The city metaverse encompasses diverse and complex lem of fake data, compatibility issues, and cluster security concerns due
scenarios compared to medical applications, necessitating the develop- to the heterogeneity and complexity of the architecture. Secure routing
ment of new AI methods specifically tailored for BCI applications [215]. protocols and data protection schemes are essential to mitigate these
challenges and ensure the integrity and security of the IoT network.
4.3. Distributed technology: intermediary connection of physical spaces Blockchain-based IoT provenance mechanisms provide a solution for
ensuring data integrity and verifiability [217]. By recording all actions
After engaging human interaction in CM, the need for physical- in the blockchain using a consensus-driven mechanism and digital sig-
virtual synchronization emerges, aiming to reflect human behavior in natures, transparency, immutability, and auditability are achieved. This
the virtual world to every urban corner. Integrated distributed technolo- technology has made a significant impact in industries such as supply
gies, such as IoT and blockchain, provide such possibilities for CM. IoT chain management, where transparent and traceable records from pro-
enables the distributed collection and reflection of real-world devices, duction to sales are crucial.
14
JID: FMRE
4.3.2. Blockchain of addressing these concerns. Ongoing research focuses on technologies

The concept of blockchain was introduced by Satoshi Nakamoto in like stealth addresses, zero-knowledge proof schemes, and reversible to-
2008, through the article “Bitcoin: A Pemer-to-Peer Electronic Cash kens such as ERC-721R [214], aiming to enhance the security and pri-
System” [218]. Blockchain is a distributed digital ledger that enables vacy of NFTs.
the recording and sharing of information within peer-to-peer networks,
which provide authenticity and validity for IoT in Section 4.3.1 through 4.4. Smart city decision support: returning to the exit of physical space
a sequential chain structure and corresponding cryptographic algo-
rithms. The features of blockchain have led to its widespread adoption in After realizing smart city simulation, human-computer interac-
finance, information security, logistics, and manufacturing, which can tion and distributed control in CM, it will be a matter of course
be used as the extension of CM on real-world devices. to integrate these technologies for decision-making support of smart
Blockchain technology encompasses different types of chains based cities [221,222]. To achieve this goal, it is necessary to develop plat-
on their degree of decentralization: public chain, private chain, and forms or systems to carry the various modules from simulation, inter-
alliance chain. The public chain operates in a decentralized manner, action and control. Here, we will mainly introduce the three kinds of
where all network nodes participate and share information freely. How- platforms, namely the city information model, virtual geographic envi-
ever, it may have limitations in terms of scalability. On the other hand, ronment and urban middle platform. The three are closely related but
the private chain is centralized, providing higher transaction efficiency conceptually distinctive, providing different emphases for different fea-
and confidentiality. It operates under the control of a central entity. The tures and can be regarded as different solutions for CM.
alliance chain falls between the public and private chains, managed by
4.4.1. City information model
multiple organizations working together.
The City Information Model (CIM) is a digital representation used
In the context of Spatial Data Infrastructure (SDI), data plays a cru-
to describe and model urban spaces, infrastructure, environments, and
cial role, and blockchain technology can contribute to its storage and
social and economic systems [223]. It is derived from the Building In-
security [1]. Spatial data often contains location information that may
formation Model (BIM), which focuses on digitizing the physical and
involve personal privacy or state secrets. By utilizing cryptography al-
functional characteristics of buildings [224]. The CIM, similar to BIM,
gorithms, the private chain or alliance chain can enhance information
integrates various data sources to provide comprehensive, accurate, and
security and protect sensitive data. Furthermore, the blockchain’s abil-
real-time information support for enhancing urban sustainability, liv-
ity to store time-series data facilitates auditing and retrospective anal-
ability, and competitiveness [225].
ysis. It allows for the verification of results and decisions by tracing
The CIM and the city metaverse are mutually beneficial as they share
back through the recorded data, enabling the identification of issues in
data, integrate technologies, and facilitate convergence. By leveraging
spatio-temporal data applications [219].
the capabilities of the CIM, the city metaverse can use various models of
prediction and simulation from digital twins to address digital inequal-
4.3.3. Non-fungible token
ity in planning and data fragmentation [226]. In return, the CM can
Non-fungible Tokens (NFTs) are unique tokens recorded on a
enrich the CIM with immersive and interactive experiences, enabling
blockchain ledger, representing certificates of distinct digital assets. In
better decision-making and urban management. The symbiotic relation-
the CM, publishers convert digital items such as images, videos, and
ship between the CIM and the city metaverse contributes to developing
audio into NFTs, enabling users to freely trade these tokens through
smart and connected cities.
smart contracts on the blockchain (see Fig. 10). NFT provides advanced
• Data sharing: The CIM can provide abundant real-world data for the
blockchain applications, goes beyond the original intention of informa-
tion proof of device networks, and provides more possibilities for human city metaverse, while behavior information generated in CM can also
behavior in CM. be fed back to CIM to provide new insights for urban management
As a new application of blockchain, Non-Fungible Tokens (NFTs) and planning [227].
• Technology complementarity: Both CIM and City Metaverse rely on
inherit and expand upon the characteristics of blockchain, includ-
ing uniqueness, traceability, scarcity, and indivisibility. These features advanced digital technologies, such as VR, AR and AI, which can
make NFTs suitable for use in the city metaverse, where they can serve achieve mutual support and collaborative innovation in different
as intermediaries for interaction and proof of ownership in the virtual scenarios [228,229].
• Application integration: CIM and City Metaverse can realize interac-
world, as well as provide digital property protection. Given the diverse
and heterogeneous nature of information in the city metaverse, NFTs tion and integration in urban planning, design, management and op-
can provide a credential basis to validate the contributions of owners. eration, like virtual urban planning experiments, public safety drills,
However, despite the protection offered by blockchain, NFTs still emergency response training, etc.
face security and privacy challenges. Incidents such as those reported On the other hand, the increasing evolution from digitization to in-
by SlowMist Hacked [220], which documented 56 security incidents in telligence gives rise to the concept of the city brain. The city brain is
2022 resulting in a loss of 65.43 million USD, highlight the importance an AI-based application that leverages SDI and CM technologies to inte-
grate diverse data sources for real-time monitoring, analysis, and deci-
sion support [230,231], enabling intelligent, efficient, and sustainable
urban development.
As the practice of large-scale artificial intelligence in the real world,
the city brain can be viewed as an application of CM and CIM. On one
hand, the city brain has five major application scenarios: urban traffic
checkup, urban police monitoring, urban traffic micro-control, urban
special vehicles, and urban strategic planning [232], all of which rely on
advanced information and modeling technologies of CM. On the other
hand, The visualization infrastructure of CIM [233,234] provided a good
platform for these applications’ usages of authorities and planners.
4.4.2. Virtual geographical environment

Virtual Geographical Environment (VGE) is a new generation
Fig. 10. Schematic diagram of the Non-Fungible Token. of geographic analysis tools for modern smart city systems [9]. It
15
JID: FMRE
integrates several essential features, including geospatial analysis, geo-

visualization, and geography-related planning and decision-making, as
well as training, education, and entertainment. VGE reflects the early
effort of CM in the field of GIS, focusing on opportunities for data sup-
port and functional expansion, and now benefiting from the technology
advancement of CM [235]:
• Twin VGE: Twin VGE is developed based on the digital twin frame-
work of metaverse, aiming to quantify the real-time and fidelity VGE
with the constraint vectors and the attribute vector, enabling deep
human interaction with the geographic environment [236].
• Geographical perspective: Traditionally, VGE provides three appli-
cation levels, namely geo-object-based analysis, geo-process-based
simulation, and multi-participant-based collaborative experiments.
They can enrich the CM through knowledge collaboration, multi-
person collaboration, multiple visualizations, spatiotemporal expres-
sion [237], which can provide a scientific basis for decision-making Fig. 11. The national land space planning scene supported by virtual sim-
and management. ulation in city metaverse.
• More than one model: With the development of CM, it is now pos-
sible to catalog more than one model for any problem in VGE, with
a human-centrally loop but focusing on different articulations of the 5. Application
applicability. The virtual infinity of the CM model defined new forms
of VGE, which now forms the cutting edge of geospatial modeling SDI and CM can be extensively applied in various fields, and here
and analysis [235]. we provide a concise overview of the eight most relevant application
• Immersive interaction: VGE can better represent our sense of place scenarios to highlight the current technological applications
through the application of VR technologies [238], deepen our geo-
5.1. National land survey and management
experience from immersion to presence and further to embodi-
ment [239]. Such immersive interaction can also benefit from recent
Using 3D digitization and AI-driven knowledge bases in the CM,
machine learning methods like knowledge graph [240].
the potential impact of different planning schemes can be deduced
VGE has attracted researchers’ attention for more than two to address various spatial land-use planning and management scenar-
decades [241], firstly defined by Lin and Gong [242] as a sub-field of GIS ios [249] (Fig. 11).
and were designed for geographic understanding and problem-solving Establishing a real scene 3D surveying and mapping system serves as
tasks by virtually augmenting users senses [9]. With the introduction the foundational framework for Digital China and the development of
of contemporary technology packages, including HCIs, distributed tech- the CM. In February 2022, the General Office of the Ministry of Natural
nologies, and simulations, VGE can now be seen as an application be- Resources issued a notice titled “Notice on Comprehensively Promot-
yond CM [238]. With the continuous maturity of CM, VGE can enhance ing Real-Scene 3D China Construction,” which laid the policy ground-
strategic visioning, pre-planning, public consultation, and traditional work for the application of the City Metaverse in national land manage-
planning practices more [243], reflecting the vast possibilities in smart ment [250].
city decision support. An exemplary application of the CM in national land space planning
is demonstrated in the Xiong’an New Area. This region has pioneered
4.4.3. Urban middle platform an innovative approach known as the “integration of planning, construc-
The Urban Middle Platform (UMP) is an open infrastructure that tion, and management.” Leveraging technologies such as Building Infor-
aims to unify and coordinate urban management, services, and devel- mation Modeling (BIM), Macro Geospatial Data (GSD), and the IoT, the
opment. It provides efficient and convenient information services for Xiong’an New District’s CIM platform aggregates urban management-
the government, enterprises, and the public [244]. By establishing the related data and adopts a micro-service architecture.
UMP, data sharing, technology integration, service support, and intel- The platform encompasses the entire life cycle of a city, covering
ligent decision-making can be achieved, promoting the digitalization, six stages: planning, construction, management, development, opera-
intelligence, and sustainable development of cities. tion, and maintenance. It represents the first instance in China and the
Compared to the city brain mentioned in Section 4.4.1, the UMP pri- global context where digital city mapping and growth are seamlessly
marily focuses on data integration, technology integration, and an open integrated. The implementation of this platform is expected to signifi-
platform, while the city brain emphasizes intelligent decision-making cantly enhance the development and refinement of Xiong’an City.
and automated management [230–232]. The UMP focuses more on the
infrastructure level, providing unified data, technology, and service sup- 5.2. Low-carbon environmental protection
port across various urban fields, which can provide the city brain with
necessary data and platforms when carrying out higher-level decision- During the 14th Five-Year Plan, China aims to accelerate digital
making and applications [245]. transformation to achieve “emission peak” and “carbon neutrality”
The UMP and the CM can also be effectively combined. At the goals, focusing on building a clean, low-carbon, safe, and efficient smart
data level, the UMP collects and integrates various urban building energy system. The combination of SDI and CM technology, facilitated
data [246] to support the construction and operation of the CM, fos- by the digital twin platform, supports and drives the digital transforma-
tering innovation and collaboration. As a comprehensive platform, the tion of the energy sector [251].
UMP is open and scalable, enabling third-party developers and insti- Specifically, CM assists enterprises in low-carbon smart production
tutions to access it and provide rich applications and services for the in the following scenarios: First, SDI and CM enable efficient collection
CM [247]. The UMP can also facilitate the interconnection of the CM and analysis of production data and energy consumption data, facili-
with other fields like education, supporting remote education and train- tating timely adjustments for energy conservation and emission reduc-
ing in urban management, planning, and construction, and offering im- tion [252]. Second, the digital twin simulation system allows for the re-
mersive learning experiences for students and professionals [248]. hearsal and assessment of different production tasks, leading to shorter
16
JID: FMRE
decision-making cycles and cost reductions. Additionally, virtual pro- more scientific traffic planning solutions [256], and ensure better
duction scenarios simulate processes and assess risks, aiding in crisis solutions for sustainable traffic development [257].
management and risk response training. Real-time data analysis using • Public transportation operation management: SDI can use various
SDI provides early warning capabilities. sensors to collect and analyze real-time data of urban public
At a macro level, SDI and CM contribute to low-carbon smart city transportation, and reflect these data in CM to provide intelligent
management. Utilizing multiple spatio-temporal data, refined energy decision support for the design and scheduling of urban public
demand forecasting optimizes energy demand and adjusts energy sup- transportation, so as to improve efficiency and convenience of
ply systems. By integrating supply and demand data from the energy urban public transport.
industry with economic, social, environmental, and policy information, • Risk management and resilience assessment: Through digital twin in
the virtual model in CM assists managers in optimizing and manag- CM, elements such as urban traffic network, roads, vehicles, passen-
ing energy storage and transmission facilities for stable supply and effi- gers and traffic facilities can be accurately simulated for emergency
cient utilization. Virtual environmental simulation scenarios in CM en- response and traffic diversion strategies, providing a more compre-
hance public understanding and experience of zero-carbon energy ap- hensive and accurate guarantee for the safe operation.
plications, promoting low-carbon lifestyles and reducing high-carbon
emission activities like business meetings and travel. 5.4. Cultural and tourism activities
5.3. Traffic planning The combination of SDI and CM enables the expansion of audience
reach by offering virtual access to cultural facilities and tourism re-
The 14th Five-Year Plan prioritizes the acceleration of digital trans- sources. By integrating spatial data, CM creates captivating and immer-
formation in the transportation sector, advocating for the application of sive experiences for tourists, thereby significantly enhancing cultural
technologies such as big data, cloud computing, Internet of Things, and and tourism activities and driving the growth of the digital economy.
artificial intelligence in transportation. It aims to foster innovation and
development of digital transportation [253]. The integration of SDI and • Spatial data and CM technology enhance tourism by mapping attrac-
CM creates a platform with diverse scenes, as depicted in Fig. 12. tions and creating virtual travel experiences, offering insights into
tourist behavior for improved marketing and service strategies. In
• Real-time prediction and management: By collecting traffic spatio- cultural activities, spatial data generates digital representations of
temporal data such as urban road conditions [254], traffic flow, pub- heritage sites and landscapes, while CM enables participation in cul-
lic transportation, and parking lots, a virtual model of urban traffic tural events through VR and AR, making exhibitions more engaging
in CM can help to develop more effective traffic strategies. and pro- and accessible.
vide an accurate and scientific basis for planning [255]. • Urban spatial data includes information about the physical, environ-
• Autonomous driving: Using SDI to obtain vehicles and environmen- mental and cultural characteristics of the urban environment, which
tal information in real-time can use to realize adaptive cruise control can be used to identify urban areas and map the boundaries of dif-
of vehicles, including adaptive acceleration and braking and steer- ferent urban functional areas, such as areas of natural or cultural sig-
ing. nificance. Detailed and accurate city maps can highlight the unique
• Vehicle-road collaboration: SDI can effectively perceive the infor- characteristics of each region thereby providing a decision-making
mation and status among urban vehicles, roads, people, weather, basis for the tourism and cultural activities.
ground-air environment, traffic stations, traffic equipment and traf- • SDI-based Point of Interest (POI) recommendations utilize user pref-
fic events in real-time to realize the coordinated linkage of human- erences, locations, and scenic spots to suggest personalized travel
vehicle-road integration. routes. Multi-source spatial data, including traffic patterns and user
• Traffic planning and design: CM technology can simulate different check-ins, predict visitor flows for effective merchant service recom-
traffic and planning schemes in the digital twin model to formulate mendations. Real-time performance improvement while maintaining
accuracy is a research focus.
5.5. Urban health
Rapid urbanization has led to health challenges like air pollution,

unhealthy lifestyles, and an aging population. The “Healthy China
2030” Planning Outline and the New Urbanization Implementation
Plan [258] emphasize the need for healthy, livable, and safe cities. The
New Urbanization Implementation Plan of the 14th Five-Year Plan also
specifically pointed out to promote the healthy, livable and safe devel-
opment of cities.
Firstly, SDI aids in identifying urban health issues by analyzing spa-
tial data. The urban built environment, including overcrowded housing,
lack of green spaces, and tobacco and alcohol exposure, affects resi-
dents’ health [259,260]. Intelligent algorithms combined with spatial
data can calculate health indicators [261] and guide spatial improve-
ments or suitable policies [262,263].
Secondly, CM promotes medical equality between urban and rural
areas. Through CM, medical knowledge and technology can be shared
with remote regions, enabling online consultations and healthcare ac-
cess for rural patients. And patients in rural areas can conduct consul-
tations across spatial distances and receive online medical care.
Finally, metaverse technology facilitates health education and pro-
Fig. 12. Application of Spatial Data Intelligent Metaverse in Urban Trans- motion in an accessible manner. It provides a virtual platform for pub-
portation. lic engagement, showcasing the benefits of healthy lifestyles, proper
17
JID: FMRE
Fig. 13. Urban resilience improvement strategies based on city metaverse.
rest, nutritious diets, and clean environments, thereby fostering health 5.7. Urban supply chain
awareness and encouraging real-life actions.
The urban supply chain, which encompasses the distribution of var-
ious materials and goods to urban areas, undergoes reorganization to
5.6. Resilient cities accommodate mass customized consumption, digital retail, and instant
response. To enhance user experience, the integration of the city meta-
Urban resilience encompasses a city’s capacity to withstand and verse offers visibility into the entire supply chain process and enables
recover from internal and external pressures by resisting, recovering, precise decision-making at a “second-level” (shown in Fig. 14). This pro-
adapting, and transforming. SDI and CM have significant roles in en- motes improved efficiency and adaptability in time allocation and spa-
hancing urban resilience through the detection and assessment of ab- tial adjustments within the urban supply chain.
normal events, optimizing resource allocation, and enabling intelligent There are four major advantages involved:
emergency management (Fig. 13).
SDI plays a crucial role in detecting and evaluating urban abnor- • System simulation: By utilizing the digital twin model of infras-
mal events by monitoring and analyzing diverse data sources, enabling tructure, equipment, users, and operations within the CM, it be-
timely countermeasures and enhancing urban resilience. For instance, comes possible to predict and intervene in events such as conges-
Lu Cong et al. achieved over 70% prediction accuracy in detecting ur- tion, failures, and resource idleness in the supply chain. This en-
ban traffic abnormal events using SDI [264]. Motta M et al. utilized SDI ables proactive measures to address inaccurate planning, unreason-
to monitor and predict urban floods, providing accurate flood warn- able plans, and uncontrollable processes, enhancing trust and effec-
ing information [265]. Wang Q et al. employed an advanced temporal tiveness within the supply chain. CM applications like autonomous
graph convolutional neural network to accurately capture crime dynam- driving computing and intelligent warehouse systems have been suc-
ics, optimizing police resource allocation [266]. Vancouver City utilized cessfully implemented in port logistics.
SDI and satellite images to monitor and analyze green spaces, identify- • Capacity prediction: Distributed computing can be employed to pre-
ing and utilizing underutilized resources to enhance the urban ecolog- dict the total demand and transport capacity of urban logistics. Real-
ical environment [267]. Furthermore, the analysis of complex system time capture of decentralized performance and order information
empowered by deep learning can help to identified the dominate vari- allows for the integration of multi-modal data such as production
ables [268] and critical nodes [269] in urban systems. planning, transportation monitoring, weather sensing, and IoT data.
CM enables the visualization, analysis, and optimization of poten- This significantly improves the prediction accuracy of the entire lo-
tially vulnerable areas and risks within a city. For example, Z Al- gistics link and enhances the timeliness of goods delivery.
lam et al. developed a CM to dynamically simulate and predict ur- • Matching of supply and demand: By simulating the behavior of
ban transportation, energy, and the environment [2]. Y Han et al. pro- the supply chain and considering the varying demand types and
posed a hybrid evolutionary dynamics framework to provide consis- strengths of facilities in the city, it becomes possible to accelerate
tent services in CM [270]. China Southern Power Grid utilized CM for the adoption of multi-modal transport modes and establish suitable
multi-modal monitoring and early warning of the power grid, enhanc- cargo loading and transport route selection. Real-time perception
ing urban resilience through systematic transformation and dispatching of external changes and immediate response ensure the provision
schemes [271]. of more matching facilities and resources, enabling intelligent op-
As urban resilience continues to develop, the scientific guidance of timization of logistics contract fulfillment paths and enhancing the
adaptation and transformation of various risks becomes increasingly responsiveness and flexibility of the transportation network.
important [29]. SDI and CM undoubtedly play pivotal roles in lead- • Synchronization of virtuality and reality: The CM’s perception,
ing research, driving industrial development, and fostering practical decision-making, and interaction capabilities are crucial for op-
innovation. timizing supply chain operations. The integration of distributed
18
JID: FMRE
Fig. 14. The urban supply chain stack based on city metaverse.
technology and spatial intelligence forms a real-time feedback sys- immersive user experience in the city metaverse [285,286], rendering
tem that helps improve efficiency, reduce operating costs, and con- urban planning and management more intuitive and interactive. With
tinuously meet the demands of urban economic and social develop- the recent development in hand tracking [287], eye tracking [288] and
ment within the urban supply chain. other intelligent technologies [7], commercial applications of VR/AR
technology have already commenced addressing practical challenges
6. Future direction such as user engagement, interface interaction, and experience enhance-
ment.
6.1. Cutting-edge technology In terms of security and privacy, blockchain technology will enhance
data protection in the CM [289] and foster trust and reliability in digital
In the coming five to ten years, as related technical fields continue to cities [290]. As an accompanying security system for the city metaverse,
advance and mature, the prospects for the technological development of blockchain advancement like decentralized mixing services [291], ring
SDI and CM will be extensive and boundless. In this discussion, we will signature [292], non-interactive zero-knowledge proof [293], or homo-
explore the drivers of technological progress and prospects, focusing on morphic cryptosystem [294] stimulating new avenues for technological
the cutting-edge technologies that may play more important roles. development and innovative applications.
In terms of computational analysis methods, advanced deep learn- Besides those technologies advances, soft power such as data sharing
ing will expedite the processing and analysis of spatial data, enhanc- and open cooperation will serve as crucial driving forces for advanc-
ing the accuracy and efficiency of intelligent management in CM [147]. ing the CM [295,296]. Collaboration among government, enterprises,
We have witnessed the large generative model, such as GPT [272], and academia in building and utilizing digital city infrastructure and
LLaMA [273], Stable Diffusion [274], perform surprisingly well on tasks resources will be pivotal, which should be carefully examined on the
of text and images generation. These models leverage diverse urban data aspect of policy [297]. Technological progress in differential privacy
to generate images, voices, texts, or other outputs that resemble human is indispensable for promoting the openness [298], contributing to the
performance, bridging the gap between the city metaverse and real-life future intelligence of spatial data and the overall growth of the CM.
scenarios. Recent work [148,172,275] has proved that generative mod-
els can be quickly migrated to spatio-temporal data for generating hu- 6.2. Future of industrial chain
man trajectory, which is vital in SDI.
Meanwhile, the advancement in reinforcement learning which SDI and the associated industrial chain of the city metaverse act as
can learn from simulation environment [276] or expert knowl- catalysts for the coordinated development of various disciplines and in-
edge [277] can accelerate the simulation in CM and delve into the dustries across different sectors. The construction of SDI relies on sub-
simulation of human need [201]. Such a model can be utilized with stantial infrastructure support from upstream industries such as comput-
large-scale decision-making models that effectively employ urban infor- ers, network communications, sensors, the IoT, and cloud computing.
mation in dynamic environments, supporting various real-life decision- Furthermore, it extends to the manufacturing and operation of space
making processes like traffic signal [278], power grid [276] or base sta- data acquisition equipment like drones and satellites, as well as SAAS
tion [252]. providers offering data processing and analysis services. Similarly, the
In terms of hardware architecture and technical support, high- development of the CM involves upstream industries related to infor-
performance computing technologies like edge computing and fog com- mation technologies, including professional city modeling, virtual real-
puting are progressively maturing. These technologies offer enhanced ity, and data visualization. Additionally, it extends to emerging sectors
computational efficiency [279], greater storage capabilities [280], and like artificial intelligence and autonomous driving. The downstream ap-
improved support for the functioning of the city metaverse [281]. Those plications of SDI and the city metaverse are extensive and have been
cutting-edge advance in computation offloading [282], energy con- discussed in Section 5.
sumption [283] or metaverse application [284] drive the progress of Moreover, the application of SDI and CM collaborative development
constructing a CM and promoting virtual-real interaction. among different industries and fields. For instance, employing SDI for
From the perspective of user access and experience improvement, logistics trajectory monitoring and optimization enhances the efficiency
virtual reality and augmented reality technologies will deliver a more and service quality of urban logistics, thereby fostering the development
19
JID: FMRE
of related industries such as e-commerce and manufacturing. By extend- susceptible to natural disasters, human errors, and cyber threats.
ing the industrial chain of urban informatization based on CM [299], Failures of smart devices, system vulnerabilities, and attacks on soft-
the formation of industrial clusters interconnected becomes a signifi- ware systems can disrupt operations and compromise the integrity
cant transformation for future urban development. of the entire city metaverse.
Over the next five to ten years, the technological development of • Data privacy: The use of spatial data intelligence technologies in the
SDI will be closely intertwined with the construction of the CM. This city metaverse can lead to privacy breaches. Existing privacy protec-
synergy will facilitate the digital transformation of cities and the es- tion measures are limited in dealing with complex relationships in
tablishment of smart cities, encompassing infrastructure, software and urban spatio-temporal data. More efficient and comprehensive ap-
hardware equipment, services, and other aspects of technological devel- proaches are needed to address data privacy concerns.
opment. The ultimate goal is to create smarter, sustainable, and livable • Ethics and morals: The city metaverse reflects and extends the real
urban environments for the benefit of all. world, introducing complex ethical and moral challenges. Virtual
avatars and interactions in the metaverse can lead to behavior
6.3. Legal, regulatory and ethics changes and raise questions about morality and legality. Ensuring
ethical standards and preventing harmful behaviors within the meta-
The technological progress and development of SDI and the CM hold verse is crucial.
immense potential, but they are also accompanied by legal and regula-
tory challenges. It is crucial for the government to enhance supervision To effectively address these security and ethical issues, collabora-
over the city metaverse, establish clear legal frameworks, identify re- tion among governments, enterprises, and the public is necessary. The
sponsible entities, and define industry standards. This can be achieved government should provide oversight and regulation while remaining
through the implementation of laws, regulations, and guidelines. Fur- open and cautious. Enterprises should proactively assess and mitigate
thermore, regulatory authorities should improve their enforcement ca- risks through scene examination, algorithm design, privacy protection
pabilities and technical expertise to effectively oversee the city meta- mechanisms, and compliance reviews.
verse.
In addition to government supervision, CM service providers have a 7. Conclusion
responsibility to self-regulate their operations. They should take proac-
tive measures to ensure the safe and stable functioning of the city meta- In summary, we have organized the research and technologies re-
verse. This involves implementing robust security protocols, adhering to lated to the CM into a coherent framework. Firstly, we explore SDI tech-
privacy regulations, and adopting best practices in data management. nologies that enable the collection of real-world information for con-
By prioritizing self-regulation, CM service providers contribute to the structing a virtual CM. Next, we discuss the integration of these tech-
overall safety and reliability of the city metaverse ecosystem. nologies with SDI to enable interaction and management of real cities
Addressing legal and regulatory challenges is vital to foster a trust- from the virtual perspective. We also highlight the practical applications
worthy and responsible environment for the development and operation and potential of these methods in real-world scenarios. Finally, we ad-
of SDI and the city metaverse. Collaboration between the government, dress the new challenges in technological progress, the industrial chain,
regulatory authorities, and CM service providers is essential in estab- legal and regulatory aspects, as well as ethics and morality.
lishing a framework that promotes innovation while safeguarding the The CM relies on comprehensive and accurate data provided by SDI
interests of individuals and society as a whole. to create urban models and support urban planning, construction, and
management. Only with sufficient, diverse, and reliable data can the
• The legal status of the city metaverse remains ambiguous, posing CM achieve a more realistic, refined, and immersive experience. Addi-
challenges in terms of management and regulation. As the city meta- tionally, the development of the CM can drive advancements in SDI.
verse encompasses various economic activities, there is a need to es- Technologies such as virtual reality, augmented reality, and artificial
tablish clear legal frameworks that address virtual property rights, intelligence that are utilized in the CM present new opportunities and
privacy protection, and network security. It is essential to develop application scenarios for SDI, enhancing our understanding of spatial
comprehensive laws that define the legal nature, transactions, and data intelligence. For instance, AI applications within the CM can foster
taxation of virtual property, while also addressing concerns such as innovations in areas like smart transportation and environmental pro-
money laundering within the virtual space. tection. SDI requires an open, inclusive, and rapidly evolving platform
• Effective management and security of data within the city metaverse like the CM, which transforms our perception of urban spaces into ac-
present significant challenges. It is necessary to develop appropri- tionable intelligence that impacts urban operations. We believe that the
ate technical solutions and regulations to mitigate the risks of data integration of SDI and the CM represents the main direction for future
breaches. Regulatory authorities should focus on ensuring that ser- development in this field.
vice providers inform users about the scope of data usage, establish
robust consent mechanisms, and implement stringent self-regulatory Declaration of Competing Interest
procedures to safeguard data privacy.
• Governance of the city metaverse requires coordination among vari- The authors declare that they have no conflicts of interest in this
ous sectors and international cooperation. The governance of the city work.
metaverse mirrors that of a physical city, involving multiple depart-
ments and agencies. Lessons from governing real cities can inform References
the governance of the city metaverse, but adjustments tailored to its
[1] X. Song, Y. Gao, Y. Li, et al., Spatial data intelligence: concept, technology and
unique nature are necessary. Additionally, given the transnational
chanllenges, J. Comput. Res. Dev. 2 (2022) 255–263.
nature of the virtual space, it is essential to foster cooperation and [2] Z. Allam, A. Sharifi, S.E. Bibri, et al., The metaverse as a virtual form of smart cities:
coordination among different countries and regions to address reg- opportunities and challenges for environmental, economic, and social sustainability
ulatory challenges that extend beyond national boundaries. in urban futures, Smart Cities 5 (3) (2022) 771–801.
[3] Y. Wang, Z. Su, N. Zhang, et al., A survey on metaverse: fundamentals, security,
Besides, in the ideal future created by the symbiosis of the real world and privacy, IEEE Commun. Surv. Tutor. 25 (1) (2022) 319–352.
[4] R.E. Hall, B. Bowerman, J. Braverman, The Vision of a Smart City, Technical Report,
and the digital world, there are significant security risks and ethical Brookhaven National Lab.(BNL), Upton, NY (United States), 2000.
issues that need to be addressed in the city metaverse: [5] A. Kirimtat, O. Krejcar, A. Kertesz, et al., Future trends and current state of smart
city concepts: a survey, IEEE access 8 (2020) 86448–86467.
• Technical reliability: While advancements in information technolo- [6] Z. Tong, F. Ye, M. Yan, et al., A survey on algorithms for intelligent computing and
gies have improved security measures, the city metaverse is still smart city applications, Big Data Mining Anal. 4 (3) (2021) 155–172.
20
JID: FMRE
[7] M. Aloqaily, O. Bouachir, F. Karray, et al., Integrating digital twin and advanced [39] M. Cheng, G. Li, Y. Chen, et al., Dense point cloud completion based on generative
intelligent technologies to realize the metaverse, IEEE Consum. Electron. Mag. 12 adversarial network, IEEE Trans. Geosci. Remote Sens. 60 (2021) 1–10.
(6) (2022) 47–55. [40] A. Capponi, C. Fiandrino, B. Kantarci, et al., A survey on mobile crowdsensing
[8] Y. Wang, K.L. Siau, L. Wang, Metaverse and human-computer interaction: a tech- systems: challenges, solutions, and opportunities, IEEE Commun. Surv. Tutor. 21
nology framework for 3D virtual worlds, in: International Conference on Human– (3) (2019) 2419–2465.
Computer Interaction, Springer, 2022, pp. 213–221. [41] C. Chen, S. Yang, Y. Wang, et al., CrowdExpress: a probabilistic framework for on–
[9] H. Lin, M. Chen, G. Lu, et al., Virtual geographic environments (VGEs): a new time crowdsourced package deliveries, IEEE Trans. Big Data 8 (3) (2022) 827–842.
generation of geographic analysis tool, Earth-Sci. Rev. 126 (2013) 74–84. [42] D. Zhang, B. Guo, Z. Yu, The emergence of social and community intelligence,
[10] M. Chen, H. Lin, Virtual geographic environments (VGEs): originating from or be- Computer 44 (7) (2011) 21–28.
yond virtual reality (VR)? Int. J. Digit. Earth 11 (4) (2018) 329–333. [43] B. Guo, Y. Liu, S. Liu, et al., CrowdHMT: crowd intelligence with the deep fu-
[11] H. Ning, H. Wang, Y. Lin, et al., A survey on the metaverse: the state-of-the-art, sion of human, machine, and IoT, IEEE Internet Things J. 9 (24) (2022) 24822–
technologies, applications, and challenges, IEEE Internet Things J. 10 (16) (2023) 24842.
14671–14688. [44] J. Li, X. Huang, J. Gong, Deep neural network for remote-sensing image interpre-
[12] L. Chettri, R. Bera, A comprehensive survey on internet of things (IoT) toward 5G tation: status and perspectives, Nat. Sci. Rev. 6 (6) (2019) 1082–1086.
wireless systems, IEEE Internet Things J. 7 (1) (2019) 16–32. [45] C. Yang, Y. Shen, B. Zhou, Semantic hierarchy emerges in deep generative repre-
[13] D.C. Nguyen, P.N. Pathirana, M. Ding, et al., Blockchain for 5G and beyond net- sentations for scene synthesis, Int. J. Comput. Vis. 129 (2021) 1451–1466.
works: a state of the art survey, J. Netw. Comput. Appl. 166 (2020) 102693. [46] K. Chen, C. Liu, H. Chen, et al. RSPrompter: learning to prompt for remote
[14] G. D’Amico, P. L’Abbate, W. Liao, et al., Understanding sensor cities: insights from sensing instance segmentation based on visual foundation model, arXiv preprint
technology giant company driven smart urbanism practices, Sensors 20 (16) (2020) arXiv:2306.16269 (2023).
4391. [47] P. Helber, B. Bischke, A. Dengel, et al., EuroSAT: a novel dataset and deep learning
[15] J. Xie, H. Tang, T. Huang, et al., A survey of blockchain technology applied to benchmark for land use and land cover classification, IEEE J. Sel. Top. Appl. Earth
smart cities: research issues and challenges, IEEE Commun. Surv. Tutor. 21 (3) Obs. Remote Sens. 12 (7) (2019) 2217–2226.
(2019) 2794–2830. [48] N. Tyagi, J. Singh, S. Singh, A review of routing algorithms for intelligent route
[16] M. Zhang, F. Tao, B. Huang, et al., Digital twin data: methods and key technologies, planning and path optimization in road navigation, in: Recent Trends in Product
Digit. Twin 1 (2022) 2. Design and Intelligent Manufacturing Systems: Select Proceedings of IPDIMS 2021,
[17] S. Boschert, R. Rosen, Digital twin–the simulation aspect, in: Mechatronic Futures: 2022, pp. 851–860.
Challenges and Solutions for Mechatronic Systems and Their Designers, 2016, [49] C. Gu, X. Duan, T. Yu, et al., A study on the key techniques of the digital city and
pp. 59–74. its 3D re-appearing, Geographocal Res. 21 (1) (2002) 14–24.
[18] H. Wu, Z. He, J. Gong, A virtual globe-based 3D visualization and interactive frame- [50] H. Ham, J. Wesley, H. Hendra, Computer vision based 3D reconstruction: a review,
work for public participation in urban planning processes, Comput. Environ. Urban Int. J. Electr. Comput. Eng. 9 (4) (2019) 2394.
Syst. 34 (4) (2010) 291–298. [51] H. Karim, A. Abdul Rahman, P. Boguslawski, et al., The potential of the 3D dual
[19] H. Duan, J. Li, S. Fan, et al., Metaverse for social good: a university campus pro- half-edge (DHE) data structure for integrated 2D-space and scale modelling: a re-
totype, in: Proceedings of the 29th ACM International Conference on Multimedia, view, in: Advances in 3D Geoinformation, 2017, pp. 477–493.
2021, pp. 153–161. [52] P. Cheng, Y. Lu, Y. Du, et al., Optimizing data placement on hierarchical storage
[20] K. Koperski, J. Adhikary, J. Han, Spatial data mining: progress and challenges sur- architecture via machine learning, in: Network and Parallel Computing: 16th IFIP
vey paper, in: Proc. ACM SIGMOD Workshop on Research Issues on Data Mining WG 10.3 International Conference, NPC 2019, Hohhot, China, August 23–24, 2019,
and Knowledge Discovery, Montreal, Canada, Citeseer, 1996, pp. 1–10. Proceedings 16, Springer, 2019, pp. 289–302.
[21] H. Wang, J. Wang, Research on progress of spatial data mining, Geogr. Geo-Inform. [53] W. Yuhui, L. Tian, L. Xinxiong, Reliability of perceived usability assessment via
Sci. 4 (6-10) (2005). crowdsourcing platform: retrospective analysis and novel feedback quality inspec-
[22] Y.-W. Zhou, Z.-Z. Hu, J.-R. Lin, et al., A review on 3D spatial data analytics tion method, Int. J. Hum.–Comput. Interact. 36 (11) (2020) 1070–1084.
for building information models, Arch. Comput. Methods Eng. 27 (2020) 1449– [54] G. Lee, B.-C. Moon, M. Park, Crowdsourcing-based learning data collection for
1463. real-time sensor error correction in indoor environments, IEEE Access 8 (2020)
[23] N. Alsaedi, A.S.A. Jalal, Big spatial data systems-a review, in: 2022 5th Interna- 127353–127367.
tional Conference on Engineering Technology and its Applications (IICETA), IEEE, [55] S. Yu, X. Chen, S. Wang, et al., An edge computing-based photo crowdsourc-
2022, pp. 147–152. ing framework for real-time 3D reconstruction, IEEE Trans. Mob. Comput. 21 (2)
[24] Z. Lv, W.-L. Shang, M. Guizani, Impact of digital twins and metaverse on cities: (2020) 421–432.
history, current situation, and application perspectives, Appl. Sci. 12 (24) (2022) [56] Y. Pan, S. Zeng, W. Chen, et al., Service quality evaluation of crowdsourcing logis-
12820. tics platform based on Fermatean fuzzy TODIM and regret theory, Eng. Appl. Artif.
[25] A.T. Kusuma, S.H. Supangkat, Metaverse fundamental technologies for smart city: Intell. 123 (2023) 106385.
a literature review, in: 2022 International Conference on ICT for Smart Society [57] L. Li, W. Jing, N. Wang, An improved distributed storage model of remote sensing
(ICISS), IEEE, 2022, pp. 1–7. images based on the HDFS and pyramid structure, Int. J. Comput. Appl. Technol.
[26] I. Yaqoob, K. Salah, R. Jayaraman, et al., Metaverse applications in smart cities: 59 (2) (2019) 142–151.
enabling technologies, opportunities, challenges, and future directions, Internet [58] W. Sun, Q. Du, Hyperspectral band selection: a review, IEEE Geosci. Remote Sens.
Things (2023) 100884. Mag. 7 (2) (2019) 118–139.
[27] S.E. Bibri, Z. Allam, The metaverse as a virtual form of data-driven smart cities: the [59] J.R. Beattie, F.W.L. Esmonde-White, Exploration of principal component analysis:
ethics of the hyper-connectivity, datafication, algorithmization, and platformiza- deriving principal component analysis visually using spectra, Appl. Spectrosc. 75
tion of urban society, Comput. Urban Sci. 2 (1) (2022) 22. (4) (2021) 361–375.
[28] S.E. Bibri, Z. Allam, The metaverse as a virtual form of data-driven smart urban- [60] L. Mou, S. Saha, Y. Hua, et al., Deep reinforcement learning for band selection
ism: on post-pandemic governance through the prism of the logic of surveillance in hyperspectral image classification, IEEE Trans. Geosci. Remote Sens. 60 (2021)
capitalism, Smart Cities 5 (2) (2022). 1–14.
[29] S.E. Bibri, Z. Allam, J. Krogstie, The metaverse as a virtual form of data-driven [61] L. Jingnan, Z. Jiao, G. Chi, et al., Data logic structure and key technologies on
smart urbanism: platformization and its underlying processes, institutional dimen- intelligent high-precision map, J. Geodesy Geoinf. Sci. 3 (3) (2020) 1–17.
sions, and disruptive impacts, Comput. Urban Sci. 2 (1) (2022) 24. [62] Y. Zhenkai, H. Yixin, Z.I. Lu, et al., Analysis of the development status and key
[30] X.X. Yang, Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Ur- technologies of high-precision map, Bull. Surv. Mapp. 0 (6) (2021) 54.
ban Environment, John Wiley & Sons, 2021. [63] I. Yildirim, A. Uyrus, E. Basar, Modeling and analysis of reconfigurable intelli-
[31] N. Mohd Noor, A. Abdullah, M. Hashim, Remote sensing UAV/drones and its ap- gent surfaces for indoor and outdoor applications in future wireless networks, IEEE
plications for urban areas: a review, in: IOP Conference Series: Earth and Environ- Trans. Commun. 69 (2) (2020) 1290–1301.
mental Science, vol. 169, IOP Publishing, 2018, p. 012003. [64] H. Taheri, Z.C. Xia, SLAM; definition and evolution, Eng. Appl. Artif. Intell. 97
[32] F. Calabrese, L. Ferrari, V.D. Blondel, Urban sensing using mobile phone network (2021) 104032.
data: a survey of research, ACM Comput. Surv. (CSUR) 47 (2) (2014) 1–20. [65] S. Lan, R. Yu, G. Yu, et al., Modeling local geometric structure of 3D point clouds
[33] F. Salamone, M. Masullo, S. Sibilio, Wearable devices for environmental monitoring using Geo-CNN, in: Proceedings of the IEEE/CVF Conference on Computer Vision
in the built environment: a systematic review, Sensors 21 (14) (2021) 4727. and Pattern Recognition, 2019, pp. 998–1008.
[34] H. Kim, H. Choi, H. Kang, et al., A systematic review of the smart energy conserva- [66] Z. Liu, Z. Zhang, Y. Cao, et al., Group-free 3D object detection via transformers, in:
tion system: from smart homes to sustainable smart cities, Renew. Sustain. Energy Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021,
Rev. 140 (2021) 110755. pp. 2949–2958.
[35] B. Rashid, M.H. Rehmani, Applications of wireless sensor networks for urban areas: [67] W. Zhang, C. Xiao, PCAN: 3D attention map learning using contextual information
a survey, J. Netw. Comput. Appl. 60 (2016) 192–219. for point cloud based retrieval, in: Proceedings of the IEEE/CVF Conference on
[36] Y. Zhang, Q. Hu, G. Xu, et al., Not all points are equal: learning highly efficient Computer Vision and Pattern Recognition, 2019, pp. 12436–12445.
point-based detectors for 3D LiDAR point clouds, in: Proceedings of the IEEE/CVF [68] Z. Liangpei, H.E. Jiang, Y. Qianqian, et al., Data-driven multi-source remote sensing
Conference on Computer Vision and Pattern Recognition, 2022, pp. 18953–18962. data fusion: progress and challenges, Acta Geodaetica et Cartographica Sinica 51
[37] T. Laidlow, J. Czarnowski, S. Leutenegger, DeepFusion: real-time dense 3D recon- (7) (2022) 1317.
struction for monocular SLAM using single-view depth and gradient predictions, in: [69] F. Bastani, S. He, S. Jagwani, et al., Updating street maps using changes detected in
2019 International Conference on Robotics and Automation (ICRA), IEEE, 2019, satellite imagery, in: Proceedings of the 29th International Conference on Advances
pp. 4068–4074. in Geographic Information Systems, 2021, pp. 53–56.
[38] S. Yu, C. Wang, C. Wen, et al., LiDAR-based localization using universal encoding [70] K. Kim, S. Cho, W. Chung, HD map update for autonomous driving with crowd-
and memory-aware regression, Pattern Recognit. 128 (2022) 108685. sourced data, IEEE Rob. Autom. Lett. 6 (2) (2021) 1895–1901.
21
JID: FMRE
[71] Y. Zhao, K. Zheng, J. Guo, et al., Fairness-aware task assignment in spatial crowd- [103] E. Cohen, E. Halperin, H. Kaplan, et al., Reachability and distance queries via 2-hop
sourcing: game-theoretic approaches, in: 2021 IEEE 37th International Conference labels, SIAM J. Comput. 32 (5) (2003) 1338–1355.
on Data Engineering (ICDE), IEEE, 2021, pp. 265–276. [104] T. Akiba, Y. Iwata, Y. Yoshida, Fast exact shortest-path distance queries on large
[72] J. Wei, Y. Lin, X. Yao, et al., Differential privacy-based location protection in spatial networks by pruned landmark labeling, in: Proceedings of the 2013 ACM SIGMOD
crowdsourcing, IEEE Trans. Serv. Comput. 15 (1) (2019) 45–58. International Conference on Management of Data, 2013, pp. 349–360.
[73] S. Han, J. Lin, S. Zhao, et al., Location privacy-preserving distance computation for [105] T. Akiba, Y. Iwata, K.-i. Kawarabayashi, et al., Fast shortest-path distance queries
spatial crowdsourcing, IEEE Internet Things J. 7 (8) (2020) 7550–7563. on road networks by pruned highway labeling, in: 2014 Proceedings of the Six-
[74] X. Deng, F. Zhang, H. Jin, Data anonymization for big crowdsourcing data, in: teenth Workshop on Algorithm Engineering and Experiments (ALENEX), SIAM,
IEEE INFOCOM 2019 - IEEE Conference on Computer Communications Workshops 2014, pp. 147–154.
(INFOCOM WKSHPS), 2019, pp. 1–6. [106] D. Delling, A.V. Goldberg, T. Pajor, et al., Robust Exact Distance Queries on Massive
[75] A. Gupta, R.K. Jha, A survey of 5G network: architecture and emerging technolo- Networks, Tech. Rep 2, Microsoft Research, USA, 2014.
gies, IEEE Access 3 (2015) 1206–1232. [107] Y. Li, L.H. U, M.L. Yiu, et al., An experimental study on hub labeling based shortest
[76] J. Zhao, J. Liu, L. Yang, et al., Future 5G-oriented system for urban rail transit: path algorithms, Proc. VLDB Endow. 11 (4) (2017) 445–457.
opportunities and challenges, China Commun. 18 (2) (2021) 1–12. [108] G. Malewicz, M.H. Austern, A.J.C. Bik, et al., Pregel: a system for large-scale graph
[77] W. Duan, J. Gu, M. Wen, et al., Emerging technologies for 5G-IoV networks: appli- processing, in: Proceedings of the 2010 ACM SIGMOD International Conference on
cations, trends and opportunities, IEEE Netw. 34 (5) (2020) 283–289. Management of Data, 2010, pp. 135–146.
[78] Q.C. Li, H. Niu, A.T. Papathanassiou, et al., 5G network capacity: key elements and [109] Y. Li, Y. Yuan, Y. Wang, et al., Distributed multimodal path queries, IEEE Trans.
technologies, IEEE Veh. Technol. Mag. 9 (1) (2014) 71–78. Knowl. Data Eng. 34 (7) (2020) 3196–3210.
[79] I. Ahmad, T. Kumar, M. Liyanage, et al., Overview of 5G security challenges and [110] S. Aridhi, P. Lacomme, L. Ren, et al., A MapReduce-based approach for short-
solutions, IEEE Commun. Stand. Mag. 2 (1) (2018) 36–43. est path problem in large-scale networks, Eng. Appl. Artif. Intell. 41 (2015) 151–
[80] Y.C. Hu, M. Patel, D. Sabella, et al., Mobile edge computing–a key technology to- 165.
wards 5G, ETSI White Paper 11 (11) (2015) 1–16. [111] D. Yang, D. Zhang, K.-L. Tan, et al., CANDS: continuous optimal navigation via
[81] A. Kumari, S. Tanwar, S. Tyagi, et al., Fog computing for smart grid systems in distributed stream processing, Proc. VLDB Endow. (PVLDB) 8 (2) (2014) 137–148.
the 5G environment: challenges and solutions, IEEE Wirel. Commun. 26 (3) (2019) [112] Z. Yu, X. Yu, N. Koudas, et al., Distributed processing of k shortest path queries over
47–53. dynamic road networks, in: Proceedings of the 2020 ACM SIGMOD International
[82] N. Pant, M. Fouladgar, R. Elmasri, et al., A survey of spatio-temporal database Conference on Management of Data, 2020, pp. 665–679.
research, in: Intelligent Information and Database Systems: 10th Asian Conference, [113] A.P. Simon, Y. Bin, S.J. Christian, Fast stochastic routing under time-varying un-
ACIIDS 2018, Dong Hoi City, Vietnam, March 19-21, 2018, Proceedings, Part II 10, certainty, VLDB J. 29 (4) (2020) 819–839.
Springer, 2018, pp. 115–126. [114] M. Datar, N. Immorlica, P. Indyk, et al., Locality-sensitive hashing scheme based
[83] C. Sommer, Shortest-path queries in static networks, ACM Comput. Surv. (CSUR) on p-stable distributions, in: Proceedings of the Twentieth Annual Symposium on
46 (4) (2014) 1–31. Computational Geometry, 2004, pp. 253–262.
[84] G. Yuan, P. Sun, J. Zhao, et al., A review of moving object trajectory clustering [115] L. Qiu, Z. Wu, E2LSH-based algorithm for trajectory KNN query, Comput. Technol.
algorithms, Artif. Intell. Rev. 47 (2017) 123–144. Dev. 30 (3) (2020) 13–18.
[85] A. Guttman, R-trees: a dynamic index structure for spatial searching, in: Proceed- [116] P. Zhao, W. Rao, C. Zhang, et al., SST: synchronized spatial-temporal trajectory
ings of the 1984 ACM SIGMOD International Conference on Management of Data, similarity search, GeoInformatica 24 (2020) 777–800.
1984, pp. 47–57. [117] S. Shang, L. Chen, Z. Wei, et al., Parallel trajectory similarity joins in spatial net-
[86] H. Samet, The quadtree and related hierarchical data structures, ACM Comput. works, VLDB J. 27 (3) (2018) 395–420.
Surv. (CSUR) 16 (2) (1984) 187–260. [118] N. Ta, G. Li, Y. Xie, et al., Signature-based trajectory similarity join, IEEE Trans.
[87] J.L. Bentley, Multidimensional binary search trees used for associative searching, Knowl. Data Eng. 29 (4) (2017) 870–883.
Commun. ACM 18 (9) (1975) 509–517. [119] F. Wang, Y. Pang, X. Zhou, et al., A method of track index for similarity search,
[88] J. Sankaranarayanan, H. Alborzi, H. Samet, Efficient query processing on spatial Comput. Appl. Softw. 34 (11) (2017) 1–5.
networks, in: Proceedings of the 13th Annual ACM International Workshop on Ge- [120] H. Yuan, G. Li, Distributed in-memory trajectory similarity search and join on road
ographic Information Systems, 2005, pp. 200–209. network, in: 2019 IEEE 35th International Conference on Data Engineering (ICDE),
[89] H. Hu, D.L. Lee, J. Xu, Fast nearest neighbor search on road networks, in: Advances IEEE, 2019, pp. 1262–1273.
in Database Technology-EDBT 2006: 10th International Conference on Extending [121] D. Xia, B. Wang, Y. Li, et al., An efficient mapreduce-based parallel clustering algo-
Database Technology, Munich, Germany, March 26-31, 2006 10, Springer, 2006, rithm for distributed traffic subarea division, Discrete Dyn. Nat. Soc. 2015 (2015)
pp. 186–203. 793010.
[90] X. Huang, C.S. Jensen, H. Lu, et al., S-GRID: a versatile approach to efficient query [122] X. Cui, P. Zhu, X. Yang, et al., Optimized big data k-means clustering using mapre-
processing in spatial networks, in: Advances in Spatial and Temporal Databases: duce, J. Supercomput. 70 (2014) 1249–1259.
10th International Symposium, SSTD 2007, Boston, MA, USA, July 16-18, 2007. [123] C. Hu, X. Kang, N. Luo, et al., Parallel clustering of big data of spatio-temporal
Proceedings 10, Springer, 2007, pp. 93–111. trajectory, in: 2015 11th International Conference on Natural Computation (ICNC),
[91] B. Shen, Y. Zhao, G. Li, et al., V-Tree: efficient kNN search on moving objects with IEEE, 2015, pp. 769–774.
road-network constraints, in: 2017 IEEE 33rd International Conference on Data [124] J. Wang, Z. Zhang, Z. Chu, et al., A trajectory data density partition based dis-
Engineering (ICDE), IEEE, 2017, pp. 609–620. tributed parallel clustering method, J. Univ. Sci. Technol. China 48 (1) (2018)
[92] R. Zhong, G. Li, K.-L. Tan, et al., G-tree: an efficient and scalable index for spa- 47–56.
tial search on road networks, IEEE Trans. Knowl. Data Eng. 27 (8) (2015) 2175– [125] Z. Shang, G. Li, Z. Bao, DITA: distributed in-memory trajectory analytics, in: Pro-
2189. ceedings of the 2018 International Conference on Management of Data, 2018,
[93] S. Luo, B. Kao, G. Li, et al., TOAIN: a throughput optimizing adaptive index for an- pp. 725–740.
swering dynamic kNN queries on road networks, Proc. VLDB Endow. 11 (5) (2018) [126] J. Zhang, Q. Ding, B. Li, et al., Bidirectional k-nearest neighbor spatial crowdsourc-
594–606. ing allocation protocol based on edge computing, PeerJ Comput. Sci. 9 (2023)
[94] D. He, S. Wang, X. Zhou, et al., GLAD: a grid and labeling framework with schedul- e1244.
ing for conflict-aware 𝑘 kNN queries, IEEE Trans. Knowl. Data Eng. 33 (4) (2019) [127] S. Wu, Y. Wang, X. Tong, Multi-objective task assignment for maximizing social
1554–1566. welfare in spatio-temporal crowdsourcing, China Commun. 18 (11) (2021) 11–25.
[95] J. Li, C. Ni, D. He, et al., Efficient kNN query for moving objects on time-dependent [128] Q. Zhang, Y. Wang, Z. Cai, et al., Multi-stage online task assignment driven by
road networks, VLDB J. 32 (2022) 1–20. offline data under spatio-temporal crowdsourcing, Digit. Commun. Netw. 8 (4)
[96] Z. Li, L. Chen, Y. Wang, G-tree: an efficient spatial index on road networks, in: (2022) 516–530.
2019 IEEE 35th International Conference on Data Engineering (ICDE), IEEE, 2019, [129] H. Tianfield, Towards edge-cloud computing, in: 2018 IEEE International Confer-
pp. 268–279. ence on Big Data (Big Data), IEEE, 2018, pp. 4883–4885.
[97] Z. Yu, Y. Liu, X. Yu, et al., Scalable distributed processing of k nearest neigh- [130] D. Ye, R. Yu, M. Pan, et al., Federated learning in vehicular edge computing: a
bor queries over moving objects, IEEE Trans. Knowl. Data Eng. 27 (5) (2014) selective model aggregation approach, IEEE Access 8 (2020) 23920–23935.
1383–1396. [131] Y. Shi, Y. Tong, Y. Zeng, et al., Efficient approximate range aggregation over large-s-
[98] S. Luo, B. Kao, X. Wu, et al., MPR–A partitioning-replication framework for multi- cale spatial data federation, IEEE Trans. Knowl. Data Eng. 35 (1) (2021) 418–430.
-processing KNN search on road networks, in: 2019 IEEE 35th International Con- [132] L. Zhang, X. Lei, Y. Shi, et al., Federated learning for IoT devices with domain
ference on Data Engineering (ICDE), IEEE, 2019, pp. 1310–1321. generalization, IEEE Internet Things J. 10 (11) (2023) 9622–9633.
[99] C. Li, Y. Gu, J. Qi, et al., A GPU accelerated update efficient index for KNN queries [133] Y. Tong, X. Pan, Y. Zeng, et al., Hu-Fu: efficient and secure spatial queries over
in road networks, in: 2018 IEEE 34th International Conference on Data Engineering data federation, Proc. VLDB Endow. 15 (6) (2022) 1159.
(ICDE), IEEE, 2018, pp. 881–892. [134] Y. Wang, Y. Tong, Z. Zhou, et al., Fed-LTD: towards cross-platform ride hailing via
[100] E.W. Dijkstra, A note on two problems in connexion with graphs, in: Edsger Wybe federated learning to dispatch, in: Proceedings of the 28th ACM SIGKDD Confer-
Dijkstra: His Life, Work, and Legacy, Association for Computing Machinery, 2022, ence on Knowledge Discovery and Data Mining, 2022, pp. 4079–4089.
pp. 287–290. [135] S. Wang, J. Cao, P. Yu, Deep learning for spatio-temporal data mining: a survey,
[101] P.E. Hart, N.J. Nilsson, B. Raphael, A formal basis for the heuristic determination IEEE Trans. Knowl. Data Eng. 34 (8) (2020) 3681–3700.
of minimum cost paths, IEEE Trans. Syst. Sci. Cybern. 4 (2) (1968) 100–107. [136] Z. Yunjun, H. Fattahi, F. Amelung, Small baseline inSAR time series analysis:
[102] R. Geisberger, P. Sanders, D. Schultes, et al., Contraction hierarchies: faster and unwrapping error correction and noise reduction, Comput. Geosci. 133 (2019)
simpler hierarchical routing in road networks, in: Experimental Algorithms: 7th 104331.
International Workshop, WEA 2008 Provincetown, MA, USA, May 30-June 1, 2008 [137] D. Brockmann, L. Hufnagel, T. Geisel, The scaling laws of human travel, Nature
Proceedings 7, Springer, 2008, pp. 319–333. 439 (7075) (2006) 462–465.
22
JID: FMRE
[138] C. Geyer, Likelihood inference for spatial point processes, in: Stochastic Geometry, [170] H. Wang, Q. Yu, Y. Liu, et al., Spatio-temporal urban knowledge graph enabled
Routledge, 2019, pp. 79–140. mobility prediction, Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 5 (4)
[139] F. Xu, Y. Li, D. Jin, et al., Emergence of urban growth patterns from human mobility (2021) 1–24.
behavior, Nat. Comput. Sci. 1 (12) (2021) 791–800. [171] Y. Liu, J. Ding, Y. Fu, et al., UrbanKG: an urban knowledge graph system, ACM
[140] O.R. Zaiane, C.-H. Lee, Clustering spatial data when facing physical constraints, Trans. Intell. Syst. Technol. 14 (4) (2023).
in: 2002 IEEE International Conference on Data Mining, 2002. Proceedings., IEEE, [172] Y. Yuan, J. Ding, H. Wang, et al., Activity trajectory generation via modeling spa-
2002, pp. 737–740. tiotemporal dynamics, in: Proceedings of the 28th ACM SIGKDD Conference on
[141] P. Waibel, L. Pfahlsberger, K. Revoredo, et al. Causal process mining from relational Knowledge Discovery and Data Mining, 2022, pp. 4752–4762.
databases with domain knowledge, arXiv preprint arXiv:2202.08314 (2022). [173] P. Mannion, J. Duggan, E. Howley, An experimental review of reinforcement learn-
[142] X. Chen, S. Jia, Y. Xiang, A review: knowledge reasoning over knowledge graph, ing algorithms for adaptive traffic signal control, in: Autonomic Road Transport
Expert Syst. Appl. 141 (2020) 112948. Support Systems, 2016, pp. 47–66.
[143] S. Ali, S.S. Tirumala, A. Sarrafzadeh, SVM aggregation modelling for spatio-tempo- [174] L. Liu, D. Dugas, G. Cesari, et al., Robot navigation in crowded environments us-
ral air pollution analysis, in: 17th IEEE International Multi Topic Conference 2014, ing deep reinforcement learning, in: 2020 IEEE/RSJ International Conference on
IEEE, 2014, pp. 249–254. Intelligent Robots and Systems (IROS), IEEE, 2020, pp. 5671–5677.
[144] T. Hengl, M. Nussbaum, M.N. Wright, et al., Random forest as a generic framework [175] L. Souza, C. Bueno, City information modelling as a support decision tool for plan-
for predictive modeling of spatial and spatio-temporal variables, PeerJ 6 (2018) ning and management of cities: a systematic literature review and bibliometric
e5518. analysis, Build. Environ. 207 (2022) 108403.
[145] J. Jiading, W. Feng, T. Rui, et al., TS_XGB: ultra-short-term wind power forecast- [176] S. Pu, G. Vosselman, Knowledge based reconstruction of building models from ter-
ing method based on fusion of time-spatial data and XGBoost algorithm, Procedia restrial laser scanning data, ISPRS J. Photogramm. Remote Sens. 64 (6) (2009)
Comput. Sci. 199 (2022) 1103–1111. 575–584.
[146] M.Y. Ansari, A. Ahmad, S.S. Khan, et al., Spatiotemporal clustering: a review, Artif. [177] M. Guo, Z. Zhang, H. Liu, et al., NDSRGAN: a novel dense generative adversarial
Intell. Rev. 53 (2020) 2381–2423. network for real aerial imagery super-resolution reconstruction, Remote Sens. 14
[147] S. Wang, J. Cao, S.Y. Philip, Deep learning for spatio-temporal data mining: a sur- (7) (2022) 1574.
vey, IEEE Trans. Knowl. Data Eng. 34 (8) (2020) 3681–3700. [178] Z. Chen, D. Li, W. Fan, et al., Self-attention in reconstruction bias u-net for semantic
[148] Y. Yuan, J. Ding, C. Shao, et al., Spatio-temporal diffusion point processes, in: KDD segmentation of building rooftops in optical remote sensing images, Remote Sens.
’23: Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and 13 (13) (2021) 2524.
Data Mining, Association for Computing Machinery, New York, NY, USA, 2023, [179] A. Wenzel, J. Chen, F. Dietrich, et al., Stand-alone low-cost sensor network in the
pp. 3173–3184. inner city of munich for modeling urban air pollutants, in: EGU General Assembly
[149] H. Yao, Y. Liu, Y. Wei, et al., Learning from multiple cities: a meta-learning ap- Conference Abstracts, 2021, pp. EGU21–15182.
proach for spatial-temporal prediction, in: The World Wide Web Conference, 2019, [180] B. Cai, X. Li, C. Ratti, Quantifying urban canopy cover with deep convolutional
pp. 2181–2191. neural networks, arXiv preprint arXiv:1912.02109 (2019).
[150] E.M. Knox, R.T. Ng, Algorithms for mining distancebased outliers in large datasets, [181] Y. He, Y. Peng, D. Lu, Survey of multisensor data fusion models, J. Tsinghua Unive.
in: Proceedings of the International Conference on Very Large Data Bases, Citeseer, (Sci & Tech) 36 (9) (1996) 14–20.
1998, pp. 392–403. [182] Y. Jia, D. Li, J. Sun, Data fusion techniques for multisources remotely sensed im-
[151] M.M. Breunig, H.-P. Kriegel, R.T. Ng, et al., LOF: identifying density-based local agery, Remote Sens. Technol. Appl. 15 (1) (2000) 41–44.
outliers, in: Proceedings of the 2000 ACM SIGMOD International Conference on [183] A. Puissant, S. Rougier, A. Stumpf, Object-oriented mapping of urban trees using
Management of Data, 2000, pp. 93–104. random forest classifiers, Int. J. Appl. Earth Obs. Geoinform. 26 (2014) 235–245.
[152] Q. Liu, M. Deng, Y. Shi, et al., A novel spatial clustering method based on multi– [184] M.S. Tehrany, B. Pradhan, M.N. Jebur, Flood susceptibility mapping using a novel
constraints, Acta Geodaetica et Cartographica Sinica 40 (4) (2011) 0. ensemble weights-of-evidence and support vector machine models in GIS, J. Hy-
[153] C. David, K. Tung, G. Chenjuan, et al., Unsupervised time series outlier detection drol. 512 (2014) 332–343.
with diversity-driven convolutional ensembles, Proc. VLDB Endow. 15 (3) (2022) [185] B.Y. He, J. Zhou, Z. Ma, et al., A validated multi-agent simulation test bed to evalu-
611–623. ate congestion pricing policies on population segments by time-of-day in new york
[154] K. Tung, Y. Bin, G. Chenjuan, et al., Robust and explainable autoencoders for un- city, Transp. Policy 101 (2021) 145–161.
supervised time series outlier detection, in: ICDE, 2022, pp. 3038–3050. [186] D.J. Fagnant, K.M. Kockelman, The travel and environmental implications of
[155] L. Deng, D. Lian, Z. Huang, et al., Graph convolutional adversarial networks for shared autonomous vehicles, using agent-based model scenarios, Transp. Res. Part
spatiotemporal anomaly detection, IEEE Trans. Neural Netw. Learn. Syst. 33 (6) C Emerg. Technol. 40 (2014) 1–13.
(2022) 2416–2428. [187] T. Xu, Study on the Data Cheeking for Urban Pipe-Line GIS Based on Knowledge
[156] Shanghai Wisdom Zhuoxin Information Technology Co.Ltd, Spatio-Temporal Arti- and Rules, Nanjing Normal University, 2005 Master’s thesis.
ficial Intelligence Powering The Digital Twin City White Paper, Technical Report, [188] J. Wang, Y. Ge, L. Li, et al., Spatiotemporal data analysis in geography, Acta Geo-
Shanghai Wisdom Zhuoxin Information Technology Co., Ltd., 2021. daetica Sin. 69 (9) (2014) 1326–1345.
[157] P. Reddy, R.U. Kiran, K. Zettsu, et al., Discovering spatial high utility frequent [189] W. Lu, C. Li, W. Sun, Based on the newmap API web front-end geographic data
itemsets in spatiotemporal databases, in: Big Data Analytics: 7th International Con- visualization, Sci. Surv. Mapp. 44 (9) (2019) 171–175.
ference, BDA 2019, Ahmedabad, India, December 17–20, 2019, Proceedings 7, [190] G. Sun, Y. Yang, X. Pan, et al., A survey on visual analysis of epidemic data, Chin.
Springer, 2019, pp. 287–306. J. Comput. 45 (3) (2022) 601–623.
[158] Y. Ding, Y. Zhu, J. Feng, et al., Interpretable spatio-temporal attention LSTM model [191] J. Zhang, Design and Implementation of A Platform for Traffic Spatial Temporal
for flood forecasting, Neurocomputing 403 (2020) 348–359. Data Processing and Interactive Visualization, Lanzhou Jiaotong University, 2017.
[159] K.-H.N. Bui, J. Cho, H. Yi, Spatial-temporal graph neural network for traffic [192] Y. Zhao, Y. Dong, Tourisim land use change and its eco-environmental impacts,
forecasting: an overview and open research issues, Appl. Intell. 52 (3) (2022) Sci. Geogr. Sin. 29 (2) (2009) 294–299.
2763–2774. [193] N. Andrienko, G. Andrienko, P. Gatalsky, Exploratory spatio-temporal visualiza-
[160] G. Jin, Y. Liang, Y. Fang, et al. Spatio-temporal graph neural networks for pre- tion: an analytical review, J. Visual Lang. Comput. 14 (6) (2003) 503–541.
dictive learning in urban computing: a survey, arXiv preprint arXiv:2303.14483 [194] J. Piao, J. Liu, F. Zhang, et al., Human–AI adaptive dynamics drives the emergence
(2023). of information cocoons, Nat. Mach. Intell. (2023) 1–11.
[161] Z.-H. Zhou, Rehearsal: learning from prediction to decision, Front. Comput. Sci. 16 [195] J. Feng, Y. Li, C. Zhang, et al., DeepMove: predicting human mobility with atten-
(4) (2022) 164352. tional recurrent networks, in: Proceedings of the 2018 World Wide web conference,
[162] A. Ghaderi, B.M. Sanandaji, F. Ghaderi, Deep forecast: Deep learning-based spatio- 2018, pp. 1459–1468.
temporal forecasting, arXiv preprint arXiv:1707.08110 (2017). [196] J. Feng, Z. Yang, F. Xu, et al., Learning to simulate human mobility, in: Proceedings
[163] L. Yu, Forecasting and decision optimization theory and methods based on artificial of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data
intelligence, J. Manage. Sci. 35 (1) (2022) 60–66. Mining, 2020, pp. 3426–3433.
[164] A. Silva, M. Gombolay, T. Killian, et al., Optimization methods for interpretable [197] Q. Hao, L. Chen, F. Xu, et al., Understanding the urban pandemic spreading
differentiable decision trees applied to reinforcement learning, in: International of COVID-19 with real world mobility data, in: Proceedings of the 26th ACM
Conference on Artificial Intelligence and Statistics, PMLR, 2020, pp. 1855–1865. SIGKDD International Conference on Knowledge Discovery & Data Mining, 2020,
[165] W. Bi, X. Cheng, B. Xu, et al. Bridged-GNN: knowledge bridge learning for effective pp. 3485–3492.
knowledge transfer, arXiv preprint arXiv:2308.09499 (2023). [198] Z. Zong, J. Feng, K. Liu, et al., DeepDPM: dynamic population mapping via deep
[166] C. Chen, Q. Liu, X. Wang, et al., semi-Traj2Graph identifying fine-grained driving neural network, in: Proceedings of the AAAI Conference on Artificial Intelligence,
style with GPS trajectory data via multi-task learning, IEEE Trans. Big Data 8 (6) volume 33, 2019, pp. 1294–1301.
(2022) 1550–1565. [199] G. Zhang, Z. Yu, D. Jin, et al., Physics-infused machine learning for crowd simula-
[167] Y. Liu, X. Zhang, J. Ding, et al., Knowledge-infused contrastive learning for urban tion, in: Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery
imagery-based socioeconomic prediction, in: WWW ’23: Proceedings of the ACM and Data Mining, 2022, pp. 2439–2449.
Web Conference 2023, Association for Computing Machinery, New York, NY, USA, [200] H. Shi, Q. Yao, Y. Li, Learning to simulate crowd trajectories with graph networks,
2023, pp. 4150–4160. in: Proceedings of the ACM Web Conference 2023, 2023, pp. 4200–4209.
[168] Z. Zhou, Y. Liu, J. Ding, et al., Hierarchical knowledge graph learning enabled so- [201] Y. Yuan, H. Wang, J. Ding, et al., Learning to simulate daily activities via mod-
cioeconomic indicator prediction in location-based social network, in: WWW ’23: eling dynamic human needs, in: WWW ’23: Proceedings of the ACM Web Con-
Proceedings of the ACM Web Conference 2023, Association for Computing Machin- ference 2023, Association for Computing Machinery, New York, NY, USA, 2023,
ery, New York, NY, USA, 2023, pp. 122–132. pp. 906–916.
[169] B. Yuan, Z. Deng, N. Geng, et al., Practice summary: cainiao optimizes the fulfill- [202] Y. Zhou, C. Liu, The logic and innovation of building digital twin city in Xiong’an
ment routes of parcels, INFORMS J. Appl. Anal. 53 (6) (2023) 446–450. new area, Urban Dev. Stud. 25 (10) (2018) 60–67.
23
JID: FMRE
[203] Y. Gao, S. Qian, Z. Li, et al., Digital twin and its application in transportation infras- [236] Q. Jing, Y. Chen, H. Lin, Framework of twin virtual geographic environment, in:
tructure, in: 2021 IEEE 1st International Conference on Digital Twins and Parallel IGARSS 2022–2022 IEEE International Geoscience and Remote Sensing Sympo-
Intelligence (DTPI), IEEE, 2021, pp. 298–301. sium, IEEE, 2022, pp. 287–290.
[204] E. Shahat, C.T. Hyun, C. Yeom, City digital twin potentials: a review and research [237] H. Lin, B. Xu, Y. Chen, et al., The virtual geographic environments: more than
agenda, Sustainability 13 (6) (2021) 3386. the digital twin of the physical geographical environments, in: New thinking in
[205] M. Xu, W.C. Ng, W.Y.B. Lim, et al., A full dive into realizing the edge-enabled GIScience, Springer, 2022, pp. 17–28.
metaverse: visions, enabling technologies, and challenges, IEEE Commun. Surv. [238] M. Rzeszewski, J. Naji, Literary placemaking and narrative immersion in extended
Tutor. 25 (1) (2022) 656–700. reality virtual geographic environments, Int. J. Digit. Earth 15 (1) (2022) 853–867.
[206] D. Martin, S. Malpica, D. Gutierrez, et al., Multimodality in VR: a survey, ACM [239] F. Hruby, R. Ressl, G. de la Borbolla Del Valle, Geovisualization with immersive vir-
Comput. Surv. (CSUR) 54 (10s) (2022) 1–36. tual environments in theory and practice, Int. J. Digit. Earth 12 (2) (2019) 123–136.
[207] M. Wang, X.-Q. Lyu, Y.-J. Li, et al., VR content creation and exploration with deep [240] B. Jiang, L. Tan, Y. Ren, et al., Intelligent interaction with virtual geographical
learning: a survey, Comput. Vis. Media 6 (2020) 3–28. environments based on geographic knowledge graph, ISPRS Int. J. Geo-Inform. 8
[208] A.D. Kaplan, J. Cruit, M. Endsley, et al., The effects of virtual reality, augmented (10) (2019) 428.
reality, and mixed reality as training enhancement methods: a meta-analysis, Hum. [241] Z. Chen, Y. Wang, T. Sun, et al., Exploring the design space of immersive urban
Factors 63 (4) (2021) 706–726. analytics, Visual Inform. 1 (2) (2017) 132–142.
[209] J. Yang, J. Lu, S. Lee, et al., Graph R-CNN for scene graph generation, in: Pro- [242] H. Lin, J. Gong, Exploring virtual geographic environments, Geogr. Inf. Sci. 7 (1)
ceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 670– (2001) 1–7.
685. [243] R. Kitchin, G.W. Young, O. Dawkins, Planning and 3D spatial media: progress,
[210] D. Buhalis, N. Karatay, Mixed reality (MR) for generation Z in cultural heritage prospects, and the knowledge and experiences of local government planners in
tourism towards metaverse, in: Information and Communication Technologies in ireland, Plann. Theory Pract. 22 (3) (2021) 349–367.
Tourism 2022: Proceedings of the ENTER 2022 eTourism Conference, January [244] I. iResearch, 2021 white paper on China’s AI mid-taiwan empowering ur-
11–14, 2022, Springer, 2022, pp. 16–27. ban space management, 2022. https://report.iresearch.cn/report/202201/3915.
[211] S. Rokhsaritalemi, A. Sadeghi-Niaraki, S.-M. Choi, A review on mixed reality: Cur- shtml?s=enable.
rent trends, challenges and prospects, Appl. Sci. 10 (2) (2020) 636. [245] M. Su, X. Jia, X. Du, et al., Research on the recent development and future trends
[212] Z. Zhang, Y. Li, J. Guo, et al., Task-driven latent active correction for physics-in- of data mid-end technology, Front. Data Comput. 1 (1) (2019) 116–126.
spired input method in near-field mixed reality applications, J. Soc. Inform. Disp. [246] B. Kalluri, C. Miller, B. Seshadri, et al., A cyber-physical middleware platform for
26 (8) (2018) 496–509. buildings in smart cities, in: Advances in Informatics and Computing in Civil and
[213] X. Zhang, L. Yao, X. Wang, et al. A survey on deep learning based brain computer Construction Engineering: Proceedings of the 35th CIB W78 2018 Conference: IT
interface: recent advances and new frontiers, arXiv preprint arXiv:1905.04149 66 in Design, Construction, and Management, Springer, 2019, pp. 645–652.
(2019). [247] R.-M. Soe, T. Ruohomäki, H. Patzig, Urban open platform for borderless smart
[214] X. Xu, F. Huijuan, A p300-based BCI system for online chinese input, J. Huaqiao cities, Appl. Sci. 12 (2) (2022) 700.
Univ. (Nat. Sci.) 36 (3) (2015) 269–274. [248] H. Lin, S. Wan, W. Gan, et al., Metaverse in education: vision, opportunities, and
[215] F. Fahimi, S. Dosen, K.K. Ang, et al., Generative adversarial networks-based data challenges, in: 2022 IEEE International Conference on Big Data (Big Data), IEEE,
augmentation for brain–computer interface, IEEE Trans. Neural Netw. Learn. Syst. 2022, pp. 2857–2866.
32 (9) (2020) 4039–4051. [249] V. Chaturvedi, W.T. de Vries, Machine learning algorithms for urban land use plan-
[216] A. Anjum, K.-K.R. Choo, A. Khan, et al., An efficient privacy mechanism for elec- ning: a review, Urban Sci. 5 (3) (2021) 68.
tronic health records, Comput. Secur. 72 (2018) 196–211. [250] G. O. of the Ministry of Natural Resources, Notice of the general office of the min-
[217] C. Guo, Y. Zhao, Research on application of blockchain technology in field of spatial istry of natural resources on comprehensively promoting real-scene 3d construction
information intelligent perception, Comput. Sci. 47 (S2) (2020) 354–358. of china, 2022. http://gi.mnr.gov.cn/202202/t20220225_2729401.html.
[218] S. Nakamoto, Bitcoin: a peer-to-peer electronic cash system, in: Decentralized busi- [251] S.M.C.o. Economy, Informatization, Relevant situation of Shanghai’s three action
ness review, 2008, p. 21260. plans for promoting the development of green and low-carbon industries, cultivat-
[219] O. Bouachir, M. Aloqaily, F. Karray, et al., Ai-based blockchain for the meta- ing a new “metaverse” track, and promoting the high-quality development of the
verse: approaches and challenges, in: 2022 Fourth International Conference on smart terminal industry, 2022. https://app.sheitc.sh.gov.cn/ttxw/692936.htm.
Blockchain Computing and Applications (BCCA), IEEE, 2022, pp. 231–236. [252] T. Li, L. Yu, Y. Ma, et al., Carbon emissions of 5G mobile networks in china, Nature
[220] Slowmist hacked, 2023. https://hacked.slowmist.io/?c=NFT. Sustain. (2023).
[221] D. Jung, V. Tran Tuan, D. Quoc Tran, et al., Conceptual framework of an intelli- [253] S.C.o. China, Notice of the state council on printing and distributing the “14th five-
gent decision support system for smart city disaster management, Appl. Sci. 10 (2) year plan” modern comprehensive transportation system development plan, 2022.
(2020) 666. http://www.gov.cn/zhengce/content/2022-01/18/content_5669049.htm.
[222] B. Akhmetov, L. Balgabayeva, V. Lakhno, et al., Mobile platform for decision sup- [254] F. Yu, W. Ao, H. Yan, et al., Spatio-temporal vehicle trajectory recovery on road
port system during mutual continuous investment in technology for smart city, network based on traffic camera video data, in: KDD ’22: Proceedings of the 28th
in: Recent Research in Control Engineering and Decision Making, Springer, 2019, ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Association
pp. 731–742. for Computing Machinery, New York, NY, USA, 2022, pp. 4413–4421.
[223] M. of Housing, U.-R. D. of the People Republic of China, Technical guidelines for [255] M. Wang, H. Yan, H. Sui, et al., Learning to discover causes of traffic congestion
basic platform of city information modeling (CIM), 2020. https://www.mct.gov. with limited labeled data, in: KDD ’22: Proceedings of the 28th ACM SIGKDD Con-
cn/preview/whhlyqyzcxxfw/zhgl/202012/P020201226505222859472.pdf. ference on Knowledge Discovery and Data Mining, Association for Computing Ma-
[224] X. Xu, L. Ding, H. Luo, et al., From building information modeling to city informa- chinery, New York, NY, USA, 2022, pp. 4041–4049.
tion modeling, J. Inform. Technol. Constr. (ITcon) 19 (2014) 292–307. [256] Y. Zheng, Y. Lin, L. Zhao, et al., Spatial planning of urban communities via deep
[225] C. A. of China, “the 14th five-year plan” national informatization plan expert talk: reinforcement learning, Nat. Comput. Sci. 3 (2023) 748–762.
accelerate the construction of urban information model platform and operation [257] Y. Zheng, H. Su, J. Ding, et al., Road planning for slums via deep reinforcement
management service platform to help build a new pattern of urban development, learning, in: KDD ’23: Proceedings of the 29th ACM SIGKDD Conference on Knowl-
2022. http://www.cac.gov.cn/2022-03/11/c_1648520393726461.htm. edge Discovery and Data Mining, Association for Computing Machinery, New York,
[226] P. Cureton, E. Hartley, City information models (CIMs) as precursors for urban NY, USA, 2023, pp. 5695–5706.
digital twins (UDTs): a case study of lancaster, Front. Built Environ. 9 (2023) [258] X. Tan, X. Liu, H. Shao, Healthy china 2030: a vision for health care, Value Health
1048510. Reg. Issues 12 (2017) 112–114.
[227] C. Carrión, Research streams and open challenges in the metaverse, J. Supercom- [259] Y. Zhang, F. Xu, T. Li, et al., Passive health monitoring using large scale mobility
put. (2023) 1–42. data, Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 5 (1) (2021).
[228] Z. Xu, M. Qi, Y. Wu, et al., City information modeling: State of the art, Appl. Sci. [260] Z. Han, H. Fu, F. Xu, et al., Who will survive and revive undergoing the epidemic:
11 (19) (2021) 9333. analyses about POI visit behavior in Wuhan via check-in records, Proc. ACM Inter-
[229] J. Hu, Z. Lv, D. Yuan, et al., A spatiotemporal intelligent framework and experimen- act. Mob. Wearable Ubiquitous Technol. 5 (2) (2021).
tal platform for urban digital twins, Virtual Reality Intell. Hardware 5 (3) (2023) [261] Y. Zhang, F. Xu, T. Xia, et al., Quantifying the causal effect of individual mobility
213–231. on health status in urban space, Proc. ACM Interact. Mob. Wearable Ubiquitous
[230] S. Guo, Y. Lin, N. Feng, et al., Attention based spatial-temporal graph convolutional Technol. 5 (4) (2022).
networks for traffic flow forecasting, in: Proceedings of the AAAI Conference on [262] L. Chen, F. Xu, Z. Han, et al., Strategic COVID-19 vaccine distribution can si-
Artificial Intelligence, 33, 2019, pp. 922–929. multaneously elevate social utility and equity, Nat. Hum. Behav. 6 (11) (2022)
[231] C.E.S. Institute, White paper on urban brain development, 2022. http://www.cesi. 1503–1514.
cn/202201/8200.html. [263] Q. Hao, F. Xu, L. Chen, et al., Hierarchical reinforcement learning for scarce medical
[232] J. Zhang, X.-S. Hua, J. Huang, et al., City brain: practice of large-scale artificial resource allocation with imperfect information, in: KDD ’21: Proceedings of the
intelligence in the real world, IET Smart Cities 1 (1) (2019) 28–37. 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, Association
[233] W. Gao, S. Ma, L. Duan, et al., Digital retina: a way to make the city brain more for Computing Machinery, New York, NY, USA, 2021, pp. 2955–2963.
efficient by visual coding, IEEE Trans. Circuits Syst. Video Technology 31 (11) [264] C. Lu, Research and Implementation of Detection Methods of Abnormal Events in
(2021) 4147–4161. Urban Traffic Data, Xidian University, 2021 Master’s thesis.
[234] Y. Lou, L.-Y. Duan, S. Wang, et al., Front-end smart visual sensing and back-end [265] M. Motta, M. de Castro Neto, P. Sarmento, A mixed approach for urban flood pre-
intelligent analysis: a unified infrastructure for economizing the visual system of diction using machine learning and GIS, Int. J. Disaster Risk Reduct. 56 (2021)
city brain, IEEE J. Sel. Areas Commun. 37 (7) (2019) 1489–1503. 102154.
[235] A. Hudson-Smith, M. Batty, Ubiquitous geographic information in the emergent [266] Q. Wang, G. Jin, X. Zhao, et al., CSAN: a neural network benchmark model for
metaverse, Trans. GIS 26 (3) (2022) 1147–1157. crime forecasting in spatio-temporal scale, Knowl.-Based Syst. 189 (2020) 105120.
24
JID: FMRE
[267] C. Haaland, C.K. van Den Bosch, Challenges and strategies for urban green-space [290] C. Esposito, M. Ficco, B.B. Gupta, Blockchain-based authentication and authoriza-
planning in cities undergoing densification: a review, Urban For. Urban Greening tion for smart city applications, Inf. Process. Manage. 58 (2) (2021) 102468.
14 (4) (2015) 760–771. [291] R. Xiao, W. Ren, T. Zhu, et al., A mixing scheme using a decentralized signature
[268] R. Li, H. Wang, Y. Li, Learning slow and fast system dynamics via automatic separa- protocol for privacy protection in bitcoin blockchain, IEEE Trans. Dependable Se-
tion of time scales, in: KDD ’23: Proceedings of the 29th ACM SIGKDD Conference cure Comput. 18 (4) (2019) 1793–1803.
on Knowledge Discovery and Data Mining, Association for Computing Machinery, [292] X. Li, Y. Mei, J. Gong, et al., A blockchain privacy protection scheme based on ring
New York, NY, USA, 2023, pp. 4380–4390. signature, IEEE Access 8 (2020) 76765–76772.
[269] J. Mao, L. Cao, C. Gao, et al., Detecting vulnerable nodes in urban infrastructure [293] A.D. Dwivedi, R. Singh, U. Ghosh, et al., Privacy preserving authentication system
interdependent network, in: KDD ’23: Proceedings of the 29th ACM SIGKDD Con- based on non-interactive zero knowledge proof suitable for internet of things, J.
ference on Knowledge Discovery and Data Mining, Association for Computing Ma- Ambient Intell. Hum. Comput. 13 (2021) 4639–4649.
chinery, New York, NY, USA, 2023, pp. 4617–4627. [294] B. Alaya, L. Laouamer, N. Msilini, Homomorphic encryption systems statement:
[270] Y. Han, D. Niyato, C. Leung, et al., A dynamic resource allocation framework for trends and challenges, Comput. Sci. Rev. 36 (2020) 100235.
synchronizing metaverse with IoT service and data, in: ICC 2022-IEEE International [295] Y. Wang, Z. Su, M. Yan, Social metaverse: challenges and solutions, arXiv preprint
Conference on Communications, IEEE, 2022, pp. 1196–1201. arXiv:2301.10221 (2023).
[271] H. Bai, C. Zhou, Z. Yuan, et al., Prospect and thinking of digital power grid based [296] I. Yaqoob, K. Salah, R. Jayaraman, et al., Metaverse applications in smart cities:
on digital twin, Southern Power Syst. Technol. 14 (8) (2020) 18–24. enabling technologies, opportunities, challenges, and future directions, Internet
[272] T. Brown, B. Mann, N. Ryder, et al., Language models are few-shot learners, Adv. Things 23 (2023) 100884.
Neural Inform. Process. Syst. 33 (2020) 1877–1901. [297] Y.K. Dwivedi, L. Hughes, A.M. Baabdullah, et al., Metaverse beyond the hype: mul-
[273] H. Touvron, L. Martin, K. Stone, et al. Llama 2: open foundation and fine-tuned tidisciplinary perspectives on emerging challenges, opportunities, and agenda for
chat models, arXiv preprint arXiv:2307.09288 (2023). research, practice and policy, Int. J. Inform. Manage. 66 (2022) 102542.
[274] R. Rombach, A. Blattmann, D. Lorenz, et al., High-resolution image synthesis with [298] H. Liu, D. Xu, Y. Tian, et al., Wasserstein generative adversarial networks based
latent diffusion models, in: Proceedings of the IEEE/CVF Conference on Computer differential privacy metaverse data sharing, IEEE J. Biomed. Health Inform. (2023)
Vision and Pattern Recognition, 2022, pp. 10684–10695. 1–12.
[275] Q. Long, H. Wang, T. Li, et al., Practical synthetic human trajectories gener- [299] M.M. Queiroz, S. Fosso Wamba, S.C.F. Pereira, et al., The metaverse as a break-
ation based on variational point processes, in: Proceedings of the 29th ACM through for operations and supply chain management: Implications and call for
SIGKDD Conference on Knowledge Discovery and Data Mining, 2023, pp. 4561– action, Int. J. Oper. Prod. Manage. 43 (10) (2023) 1539–1553.
4571.
[276] Z. Zhu, K. Lin, A.K. Jain, et al., Transfer learning in deep reinforcement learn- Xiaofeng Meng (IEEE Member) is currently a full professor
ing: a survey, IEEE Trans. Pattern Anal. Mach. Intell. 45 (11) (2023) 13344– at the School of Information, Renmin University of China. He
13362. is also a CCF Fellow, the General Chair of ACM SIGSPATIAL
[277] S. Arora, P. Doshi, A survey of inverse reinforcement learning: Challenges, methods 2021, and the General Vice Chair of IEEE ICDE 2009. He has
and progress, Artif. Intell. 297 (2021) 103500. served on the Program Committee of SIGMOD, ICDE, CIKM,
[278] T. Chu, J. Wang, L. Codecà, et al., Multi-agent deep reinforcement learning for and DASFAA. His research interests include mobile data man-
large-scale traffic signal control, IEEE Trans. Intell. Transp. Syst. 21 (3) (2019) agement, privacy protection in mobile web, social computing,
1086–1095. and spatial databases. He has published over 200 papers in
[279] Y. Hu, T.-M. Li, L. Anderson, et al., Taichi: a language for high-performance compu- prestigious journals and conferences, including IEEE Transac-
tation on spatially sparse data structures, ACM Trans. Graph. (TOG) 38 (6) (2019) tions on Knowledge and Data Engineering, VLDBJ, and SIGMOD.
1–16.
[280] F. Cappello, S. Di, S. Li, et al., Use cases of lossy compression for floating-point
data in scientific data sets, Int. J. High Perform. Comput. Appl. 33 (6) (2019)
1201–1220. Yong Li is currently a tenured associate professor at the De-
[281] S. Dhelim, T. Kechadi, L. Chen, et al. Edge-enabled metaverse: the conver- partment of Electronic Engineering, Tsinghua University. He
gence of metaverse and mobile edge computing, arXiv preprint arXiv:2205.02764 received a Ph. D. degree in electronic engineering from Ts-
(2022). inghua University in 2012. His research interests include city
[282] Y. Wang, P. Lang, D. Tian, et al., A game-based computation offloading method science and urban computing. Dr. Li has served as General
in vehicular multiaccess edge computing networks, IEEE Internet Things J. 7 (6) Chair, TPC Chair, and SPC/TPC Member for several inter-
(2020) 4987–4996. national workshops and conferences, and is on the editorial
[283] M. Guo, W. Wang, X. Huang, et al., Lyapunov-based partial computation offloading board of two IEEE journals. He has published over 100 papers
for multiple mobile devices enabled by harvested energy in MEC, IEEE Internet on firsttier international conferences and journals, including
Things J. 9 (11) (2021) 9025–9035. Nature Human Behavior, Nature Computational Science, KDD,
[284] Y. Wang, J. Zhao, Mobile edge computing, metaverse, 6G wireless communications, NeurIPS, WWW, UbiComp, and his papers have total citations
artificial intelligence, and blockchain: survey and their convergence, in: 2022 IEEE of more than 25000.
8th World Forum on Internet of Things (WF-IoT), IEEE, 2022, pp. 1–8. Yuming Lin is a postdoc fellow of the Department of Elec-
[285] W. Xu, M.J. Dainoff, L. Ge, et al., Transitioning to human interaction with AI sys- tronic Engineering, Tsinghua University. He received a doc-
tems: new challenges and opportunities for HCI professionals to enable human– toral degree in environmental design from Kanazawa Univer-
centered AI, Int. J. Hum.–Comput. Interact. 39 (3) (2023) 494–518. sity and a master’s degree in architecture from Tsinghua Uni-
[286] J. Garbett, T. Hartley, D. Heesom, A multi-user collaborative BIM-AR system to versity. His research focuses on the resilience of urban sys-
support design and construction, Autom. Constr. 122 (2021) 103487. tems concerning the impact of the COVID-19 pandemic. His
[287] J. Romero, D. Tzionas, M.J. Black, Embodied hands: Modeling and capturing hands research interest also covers the broad application of human
and bodies together, arXiv preprint arXiv:2201.02610 (2022). mobility data and machine learning technology in the sustain-
[288] A.F. Klaib, N.O. Alsrehin, W.Y. Melhem, et al., Eye tracking algorithms, techniques, able development of architecture and urban planning.
tools, and applications with an emphasis on machine learning and internet of things
technologies, Expert Syst. Appl. 166 (2021) 114037.
[289] T.R. Gadekallu, W. Wang, G. Yenduri, et al., Blockchain for the metaverse: a review,
Future Gener. Comput. Syst. 143 (2023) 401–419.
25

Spatial Data Intelligence and City Metaverse - A Review

Uploaded by

Copyright:

Available Formats

Spatial Data Intelligence and City Metaverse - A Review

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Spatial Data Intelligence and City Metaverse - A Review

Uploaded by

Copyright:

Available Formats

JID: FMRE

ARTICLE IN PRESS [m5GeSdc;February 2, 2024;14:31]

Fundamental Research xxx (xxxx) xxx

Contents lists available at ScienceDirect

Spatial data intelligence and city metaverse: A review

1. Introduction to meteorological, population, and socioeconomic data, have ampliﬁed

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

Fig. 1. The conceptual connection between SDI and SM.

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

Fig. 2. Key technology involved in the review on Spatial intelligent sensing.

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

Fig. 3. Schematic diagram of high-performance computing framework for spatial data.

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

Algorithm Task Year Key features

R-trees [85] general index 1984 fast nearest neighbours queries

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

Fig. 4. Main research content of intelligent data mining.

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

3.3.3. Spatio-temporal correlation analysis • Data-driven spatio-temporal decision-making: The visualization of

Fig. 5. Key technology involved in the review on Metaverse Space Construction.

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

information hierarchically, which is crucial for generating content

4.2.2. Augmented reality

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

Fig. 9. Schematic diagram of IoT and Blockchain distributed technology.

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

4.3.2. Blockchain of addressing these concerns. Ongoing research focuses on technologies

4.4.2. Virtual geographical environment

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

integrates several essential features, including geospatial analysis, geo-

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

5.5. Urban health

Rapid urbanization has led to health challenges like air pollution,

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

Fig. 13. Urban resilience improvement strategies based on city metaverse.

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

X. Meng, Y. Li, K. Liu et al. Fundamental Research xxx (xxxx) xxx

You might also like