Energy consumption in 5g systems

– challenges and solutions

Florin Paulescu Gheorghița Daniela Năchescu
Department of Communications Department of Communications
Politehnica University Timisoara Politehnica University Timisoara
Timisoara, Romania Timisoara, Romania
florin.paulescu@student.upt.ro gheorghita.nachescu@student.upt.ro

Abstract— With the introduction of 5G networks in our

lives, we are faced with the challenge to lower energy
consumption, which is expected to increase significantly. As a
result, it is necessary to develop solutions to improve energy
efficiency while maintaining performance. This paper
introduces the basic energy-saving technology of 5G networks

Keywords—5G, technology, energy, efficiency, optimization

I. INTRODUCTION
Acording to authors of [11] the mobile telephony is the
product of telecommunication engineering which started at
the end of 19th century.
Figure 2: 5G Demands - Higher capacity, lowest latency
Cellular networks were originally designed for voice only and more consistent experience
application, using analogue transmission channels.
When digital technology evolved and more voice channels 5G comes with new requirements for saving energy: 5G
were required, digital 2G systems emerged in the 1990s. communication modules to be optimized for low-energy
Digital transmission enabled new services such as text consumption in idle modes. Energy management protocols at
messaging and circuit switched data access. 5G base station levels to optimize energy consumption based
on real traffic. Efficient cooling technologies and alternative
The low data rate services provided by 2G systems did not power sources to reduce environmental impact. Defining
fulfill the need for mobile Internet access. This lead to a standards for machine-to-machine (M2M) communications
demand for new 3G standards, which evolved to provide fast
that allow devices to communicate efficiently while
data services and more capacity for voice. The recent (4G)
mobile communications system LTE was developed to minimizing energy consumption.
provide high capacity and highest rate data service for mobile
5G Use Cases for Energy Saving: Smart Grids and Energy
multimedia. Seen from a historical point of view, each of the
cellular standards has evolved around a set of key use cases Management by Using 5G networks to facilitate fast and
presented in Figure 1: efficient communications within smart energy distribution
networks. Intelligent Transportation by Developing 5G
networks to support vehicle-to-vehicle (V2V) and vehicle-to
infrastructure (V2I) communications to improve traffic flow
and avoid congestion. Smart Buildings, Agriculture and
Industry: Integrating 5G networks for the efficient
connection and management of devices in smart buildings
and automation control for optimizing energy usage.

This paper is organized as follows:

a) Section 2 discuss about reduced power consumption
from User Equipament and GnodeB.
Figure 1: 5G: symbiotic integration of existing and new b) Section 3 presents Sleep Modes.
technologies (“5G use cases and requirements,” Nokia
c) Section 4 discuss power saving algorithms.
d) Section 5 provides the conclusion of the paper.
According to [3,6], 5G subscriptions in 2020 were
forecasted at 220 million, while by the end of 2026 they
predict 3.5 billion 5G subscriptions, making around 40% of all
mobile subscriptions.
 II. UE & GNODEB POWER CONSUMPTION 6) Dynamic Network Selection: UEs can dynamically
The traffic of cellular network has great non-uniformity select and switch between different network types (e.g., 5G,
and volatility in space and time, which brings opportunities LTE, Wi-Fi) based on factors such as signal strength, data
and challenges to the planning and management of cellular rates, and power consumption. This ensures the UE is
network. connected to the most energy-efficient network available.
The authors in [4] have surveyed the efforts of different
academic and industry projects on energy efficiency, B. GnodeB power consumption
emphasizing how to make use of daily network traffic The authors of [4] highlight that BSs consume about 60%
variations to save energy at the RAN level. to 80% of the energy used in cellular networks.
In particular, the RAN accounts for about 73% of the total The BSs have two power consumption components: node
power usage of a typical cellular network. power consumption and communication power consumption.
In this regard, some studies examined reducing the BS The node power consumption includes the power
transmit power, others concentrated on the hardware’s consumption of signal processing, cooling, and backup
efficiency, and many others considered the concept of batteries. Communication power includes the power
deactivating the BSs, putting them into sleep mode, consumption to transmit a signal with a certain coverage,
whenever it is feasible. which depends on the distance of each UE: a distant UE
A. UE battery consumption consumes more energy in data transmission than a closer one.
As the authors of [10] say, the smartphone receives the Some specific strategies and techniques employed to
required energy, to run itself, from the battery within it. achieve energy savings on gNodeBs presented by authors of
Typically, a rechargeable battery is used for this purpose, [2,6,7,9 ]:
which may be implanted within the device or may be 1) Sleep Mode
detachable. The battery needs to be recharged periodically 2) Cell Densification and Small Cells
and most often in quick succession We need to understand 3) Massive MIMO
the energy consumption patterns in a smartphone and 4) Energy-Efficient Hardware
optimize and regulate its architecture and usage accordingly. 5) Advanced Cooling Systems
A mobile device is architected in a way that it is 6) Predictive Analytics and Machine Learning: improve
constantly connected to a cellular network.Therefore, the efficiency and lengthen sleep periods, realize intelligent
power dissipation in a smartphone is highly dependent on the collaboration of 5G network, continuous optimization of
quality of the signal being received by the device at a radio resource scheduling and improvement of user
particular location. Weak signal strength leads to an increase
experience.
in power consumption of the Tx/Rx module, causing high
7) Hybrid Power Solutions: Integrating hybrid power
energy drainage.
solutions, such as combining grid power with renewable
Optimizing energy consumption in User Equipment (UE) energy sources (solar, wind), can help gNodeBs operate more
is crucial for extending battery life, improving device sustainably. This is particularly relevant in off-grid or remote
efficiency, and promoting overall sustainability in 5G locations.
systems. 8) Network Slicing: Implementing network slicing allows
operators to allocate resources more efficiently by tailoring
Authors in [10] presents several strategies and techniques
network slices to specific services or applications. This helps
employed to achieve energy savings on UEs:
in optimizing the use of gNodeB resources and reducing
1) Idle Mode Optimization: Implementing techniques to unnecessary energy consumption.
optimize power usage during idle periods is crucial. UEs can 9) D2D technology: brings the largest gains in macro
periodically enter low-power states and reduce the frequency cells, up to 70% energy reductions. Small cells are already
of background processes to extend battery life when the more energy efficient, but energy savings can still be
device is not actively in use. significant, even 40% in the case of femto cells.
2) Adaptive Modulation and Coding: UEs can
dynamically adjust modulation and coding schemes based on
the quality of the wireless channel.
3) Low-Power States and Doze Mode: UEs can enter
low-power states or doze mode during periods of inactivity.
This involves shutting down or reducing power to certain
components while maintaining essential functions, allowing
for energy savings.
4) Background Task Optimization: Managing
background tasks and services efficiently helps prevent
unnecessary power consumption.
5) Battery-Aware User Interfaces: Providing users with
information about the impact of different settings on battery
life encourages energy-conscious usage.
 III. SLEEP MODES IV. POWER SAVING ALGORITHMS
Implementing sleep modes during periods of low traffic The authors of [1, 2,4,5,6,7,8] presented the Q-learning,
helps reduce power consumption. Certain components of the Cell Zooming, Machine and Reinforcement Learning
gNodeB, such as radio transceivers, can be put into a low- algorithms that address the software and hardware energy
power state when not actively serving users. saving.
The sleep mode strategy presented by the authors of [4] For software energy saving, it is based on the software to
assumes the deactivation/activation of the BS hardware schedule base station resource according to the service load
components. These components can be grouped by similar to keep the base station to run effectively.
deactivation/activation time and assigned to different sleep
mode levels: As stated in [5], the following sleep mode levels A. Q-learning
can be assumed: The authors of [2] point out the importance of using Q-
a) Sleep Mode (SM)1 level: the power amplifier and learning to understand the request likelihood and the statistics
of random arrival and departure of mobile users because we
some processing components of the digital baseband and
don’t know the probabilities that the users would request the
analogue front-end are disabled. This is the fastest level with
files.
0.071 ms of (Orthogonal Frequency Division Multiplexing
(OFDM) symbol) deactivation/activation time. Selection: A strategy based on Q-learning in combination with
Highlight all author and affiliation lines. logarithmic cooling (LC) is proposed to optimize the
connection of user equipment (UE) with a base station or
b) SM2 level: needs 1 ms (1 sub-frame of Transmission access point (AP) to address energy efficiency issues of
Time Interval (TTI)) to deactivate/activate additional device-to-device (D2D) communications in heterogeneous
c) SM3 level: the power amplifier, all the components networks
of the digital baseband, and almost all the parts of the
analogue front-end (except the clock generator) are switched B. Predictive Analytics and Machine Learning
off. The deactivation/activation time is 10 ms Leveraging machine learning algorithms and predictive
d) SM4 level: is the standby mode where a large part of analytics can help forecast traffic patterns and dynamically
the components of BS is deactivated. The wake-up adjust gNodeB parameters to optimize energy consumption.
This proactive approach contributes to efficient resource
functionality will take 1s, as this is the minimal sleep
duration. According to the authors of [4,7] Using machine learning
As the system enters a deeper sleep mode, less energy is algorithms it is possible to predict the category of small cells,
consumed, however, the transition latency to activate the BS specifically and make the decision whether to set cells to sleep
takes longer, which degrades the end user QoS. or to gradually deactivate or activate components of BSs and
decrease BS energy consumption.
Different sets of hardware can be grouped by sleep mode,
being an example, table 1 presents the power consumption Reinforcement Learning: a subset of machine learning, an
measurements for 4 different types of BSs where the authors agent that learns using a trial-and-error approach to map
have grouped the power consumption and transition times by situations into actions. A reward or penalty is received after
sleep mode. each action, allowing action-based learning to improve. Is
used to enable a scheduling solution for data traffic
management. The scheduling framework allows different
scheduling rules to be selected as the RL agent maps the rules
to each state and learns when to apply each. This framework
allows minimizing packet delays and packet drop rates for
strict QoS requirement applications.
C. Cell zooming algorithm: is traditional based on load
BB concentration algorithm. algorithm based
TABLE 1. Base station power consumption by base station type as
defined in [5]. It’s more efficient when it’s based on load balancing
algorithm, can improve EE in a heterogeneous cellular
network. The system model used is a heterogeneous cellular
network which includes one macro BS (CBS) and many
small BSs (SBS). The CBS only provides coverage and the
SBSs dynamically adjust transmit power to achieve cell
zooming. The core of our LB algorithm is to move the
traffic from the heavily loaded SBSs to the less loaded
SBSs.
Power Consumption algorithms enable mobile devices to
consume far less energy while maintaining high connection
speeds. However, by reducing the amount of energy required
for communication, these algorithms can significantly extend
device battery life. By optimizing Internet connection speeds,
users can also consistently experience faster web downloads
and fewer macro outages.
 V. CONCLUSIONS REFERENCES
Contribution: In this paper, we make an overview about
energy-saving strategies: a review of research papers, [1] S. Malta, P. Pinto and M. Fernández-Veiga, "Using Reinforcement
addressing different challenges of energy consumption in 5G Learning to Reduce Energy Consumption of Ultra-Dense Networks
With 5G Use Cases Requirements," in IEEE Access, vol. 11, pp. 5417-
networks and a base analysis of the various entities and factors 5428, 2023
that are responsible for the majority amount of the total power [2] A. Aggarwal and D. Selvamuthu, "Energy Optimization of a Base
consumption in a smartphone and in a gNodeB. Station using Q-learning Algorithm," 2023 17th International
Conference on Telecommunications (ConTEL), Graz, Austria, 2023,
In summary, the success of 5G systems relies on the pp. 1-5.
effective implementation of these optimization and energy- [3] Allayiotis. F, Bouras. C, Kokkinos. V, ”Α Machine Learning
saving strategies. Collaboration among network operators, Mechanism for improved Energy Consumption in 5G Small Cells, “ in
equipment manufacturers, and regulatory bodies is crucial to IEEE Symposium on Computers and Communications, Rhodes,
advancing these strategies, reducing the environmental impact Greece, 2022
of telecommunications, and ensuring the long-term [4] S. Malta, P. Pinto and M. Fernández-Veiga, "Using Reinforcement
sustainability of 5G networks. Learning to Reduce Energy Consumption of Ultra-Dense Networks
With 5G Use Cases Requirements," in IEEE Access, vol. 11, pp. 5417-
By combining these strategies, device manufacturers and 5428, 2023.
software developers can contribute to the overall energy [5] B. Debaillie, C. Desset, and F. Louagie, ‘‘A flexible and future-proof
efficiency of 5G systems by ensuring that UEs operate in an power model for cellular base stations,’’ in Proc. IEEE 81st Veh.
Technol. Conf.(VTC Spring), May 2015, pp. 1–7.
optimized and sustainable manner.
[6] Z. Zhang, F. Liu and Z. Zeng, "The cell zooming algorithm for energy
The advancement and popularity of smartphones have efficiency optimization in heterogeneous cellular network," 2017 9th
made it an essential and all-purpose device. But lack of International Conference on Wireless Communications and Signal
Processing (WCSP), Nanjing, China, 2017, pp. 1-5.
advancement in battery technology has held back its optimum
[7] T. Rumeng, W. Tong, S. Ying and H. Yanpu, "Intelligent Energy
potential. Therefore, considering its scarcity, optimal use and Saving Solution of 5G Base Station Based on Artificial Intelligence
efficient management of energy are crucial in a smartphone. Technologies," 2021 IEEE International Joint EMC/SI/PI and EMC
For that, a fair understanding of a smartphone’s energy Europe Symposium, Raleigh, NC, USA, 2021, pp. 739-742.
consumption factors is necessary for both users and device [8] M. S. Khalaf, A. A. A. Wahab, H. A. Al-Wzwazy, A. S. Al-Khaleefa,
manufacturers. M. Almohamadi and A. M. Dakhil, "Review of Beamforming’s Energy
Consumption Algorithms for (5G) Wireless Networks," 2023 Al-Sadiq
International Conference on Communication and Information
Technology (AICCIT), Al-Muthana, Iraq, 2023, pp. 288-291.
[9] Höyhtyä, Marko, Olli Apilo and Mika Lasanen. “Review of Latest
Advances in 3GPP Standardization: D2D Communication in 5G
Systems and Its Energy Consumption Models.” 2018 Future Internet
no. 1: 3.
[10] P. K. D. Pramanik et al., "Power Consumption Analysis, Measurement,
Management, and Issues: A State-of-the-Art Review of Smartphone
Battery and Energy Usage," in IEEE Access, vol. 7, pp. 182113-
182172, 2019.
[11] Nokia Oyj, Karaportti “ 5G use cases and requirements,” in Nokia.com,
2016.