Int J Wireless Inf Networks DOI 10.1007/s10776-015-0271-2 Comparative Analysis of Routing Protocols in Wireless Sensor–Actor Networks: A Review Jagadeesh Kakarla1 • Banshidhar Majhi1 • Ramesh Babu Battula2 Received: 6 February 2014 / Accepted: 15 May 2015 Ó Springer Science+Business Media New York 2015 Abstract Wireless sensor–actor networks (WSANs) are applicable in versatile domains ranging from very common to those which demand reliable actions in the event area. The unique characteristics of WSANs and the resourceconstrained nature of the constituent sensor nodes give rise to design energy and delay efficient routing protocols. Many researchers have made considerable efforts to meet these challenges by designing energy and delay efficient protocols. In this paper, a comprehensive survey is presented on existing routing protocols for WSAN to analyze and list out their merits and demerits. The study broadly segregates the existing routing protocols into two categories i.e., cluster based and non-cluster based protocols. The cluster based protocols structure the physical network into virtual groups, whereas the non-cluster based protocols either use flooding or broadcasting mechanisms for communication. All the routing protocols under consideration have been simulated in a common simulation platform. Their performances are analyzed with respect to average end-to-end delay, packet delivery ratio, and average energy dissipation in the network individually. In addition, the best protocols among both the categories are compared to derive an overall conclusion. & Jagadeesh Kakarla jagadeesh0826@gmail.com Banshidhar Majhi bmajhi@nitrkl.ac.in Ramesh Babu Battula ramsbattula@gmail.com 1 Department of CSE, NIT Rourkela, Rourkela, Odisha, India 2 Department of CSE, MNIT Jaipur, Jaipur, Rajasthan, India Keywords Actor  Delay  Packet delivery ratio  Energy  Sensor 1 Introduction Wireless Sensor Networks (WSNs) have been considered as one of the most important technologies in the twenty first century [1, 2]. These are tiny, cheap, and smart sensors communicating with wireless technologies deployed in a physical area. WSN is widely used in various applications such as environmental monitoring, battlefield surveillance, industrial process control, and home applications [3–5]. These networks are distinguished from traditional wireless networks because of its unique characteristics namely, the dense node deployment, unreliability of sensor nodes, severe energy consumption, and storage constraints [6–9]. WSN is a passive network, where sensors gather and forward environmental information to the sink. It does not perform any actions in the environment. The realization of intelligent interaction in the physical area has led to the evolution of a new class of wireless networks called as wireless sensor–actor networks (WSAN) [10]. They are capable of monitoring the environment and perform necessary actions on it [11, 12]. Application areas of WSANs comprise but are not limited to habitat monitoring [13, 14], health environment monitoring [15, 16], military applications [17, 18], industrial and consumer applications [19, 20], and preventing chemical, biological, or nuclear threats in an area [21]. WSAN consists of both resource rich actors and resourceimpoverished sensors. An actor is a transducer that accepts a signal and converts it to a physical action. It is also a network entity that performs network-related functionalities such as receive, transmit, process, and relay data. The 123 Int J Wireless Inf Networks actors are mobile elements with high processing, communication capabilities, and are less constrained to energy resources than the static sensors [22–24]. Typical examples of actors are robots, electrical motors, etc. The resource constrained sensors gather the environmental information and forward it to the actors. The actors process the data and perform application dependent actions on the physical world. Consider a battlefield application, where image sensors are used to detect the presence of enemy targets and tasks. The smart weapons and ambulance can be considered as actors for destroying the enemy targets and rescuing the injured soldiers. WSAN supports three types of data communication modes, namely event-driven, periodic, and on-demand [25–27]. In the event-driven mode, sensors report their sensing data to the sink or actors on occurrence of an event. The data transmission delay is an important parameter in event driven mode [28–30]. Target detection and tracking applications use this mode of communication. In the periodic mode, sensors sense the environment at predefined intervals and send the data to the sink or actor. Data gathered in periodic mode does not require quick delivery to the sink. In the on-demand mode, users gather the event information based on their interest. They send instructions to the sink consisting of their interests to know the environmental conditions [31–34]. WSAN supports three types of architectures namely automated architecture, semi-automated architecture, and cooperative architecture [35–39]. In the automated architecture, sensors sense the environment and report directly to the actor. Based on the received data, actors coordinate among themselves and perform appropriate actions in an event area. The automated architecture minimizes the communication delay and energy consumption as the sensing data is directly communicated to the nearest actor. In semi-automated architecture sensors first transfer the data to the sink node. Then sink process the data and issue commands to an actor based on the sensors information [40]. The working principle of semi-automated architecture is similar to that of traditional WSN [41–43]. In cooperative architecture sensors transfer the sensed information to an actor. The actor processes the data and consults the sink before performing any action in the event area. In WSAN, the sensors are deployed more densely as compared to resource rich actors [44–47]. The sensorsensor, sensor–actor, and actor–actor coordination is important in WSAN and as a result routing in WSAN is more complex as compared to WSN [48–53]. WSAN includes the combined characteristics of WSN and ad-hoc networks [54, 55]. Hence existing routing protocols for WSN and ad-hoc networks cannot be directly applied to WSAN [56–60]. In this paper, an attempt has been made to survey the plethora of routing protocols for WSAN by 123 simulating them in a common platform. The objective is to understand their relative performances and guide the researchers in WSAN to utilize the best protocols suitable. The rest of the paper is organized as follows: Section 2 presents the simulation setup used for carrying out the simulations. Section 3 describes the design goals of routing protocols for WSAN. Section 4 discusses the performance of existing cluster based routing protocols with their merits and demerits. Section 5 provides analysis of cluster based routing protocols. Section 6 discusses the performance of existing non-cluster based routing protocols and list out their relative issues. Section 7 analysis non-cluster based routing protocols. Section 8 provides comparison between HEROP and DEARP. Finally, Sect. 9 concludes the paper. 2 Simulation Setup To understand the routing protocols better we use NS2 discrete simulator which supports wired, wireless, and sensor networks to study various performance parameters [61]. For simulation analysis, 100–1000 static sensors and 8–12 mobile actors are deployed in a 1000 m  1000 m network area. The other network parameters like duration of simulation, traffic flow, etc. are listed in Table 1. In the simulation, a radio model is assumed for sensor node energy dissipation for transmission and receiving as shown in Fig. 1. The free space ðEfs Þ and multi-path fading ðEmp Þ channel models are used based on the distance between the transmitter and receiver [62, 63]. The free space model is used, if distance between transmitter and receiver is less than threshold do , otherwise multi-path Table 1 Simulation parameters Parameters Values Network area 1000 9 1000 m2 Simulation duration 200 s Traffic flow MAC layer CBR IEEE 802.15.4 Number of sensors 100–1000 Number of actors 8–12 Actor’s mobility speed 0–16 m/s Mobility pattern Random waypoint Sensor’s transmission range 100 m Actor’s transmission range 300 m Packet size 512 b Sensor’s initial energy 2J Eelec 50 nJ/bit Efs 10 pJ=bit=m2 Emp 0:0013 pJ=bit=m4 Int J Wireless Inf Networks protocols uses flooding mechanism to learn about their neighbors. These protocols do not structure the physical network into virtual groups. The working principles of cluster as well as non-cluster based routing protocols are discussed in the following sections. Fig. 1 Radio energy dissipation model 4 Cluster Based Routing Protocols model is used. The energy required to transmit a b  bit message over a distance d ðETX ðbÞÞ and to receive the message ðERX ðbÞÞ are represented defined as, ETX ðbÞ ¼ ETXelec ðbÞ þ ETXamp ðb; dÞ ( bEelec þ bEfs d2 ; d\d0 ¼ bEelec þ bEmp d4 ; d  d0 ð1Þ ðbÞ ¼ bEelec ð2Þ ERX ðbÞ ¼ ERXelec rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi . where, d0 ¼ Efs Emp , electrical energy ðEelec Þ depends on digital coding, modulation, and filtering mechanism of the signal. The amplifier energy, Efs d 2 or Emp d4 depends on the distance between transmitter and receiver, and the acceptable bit-error rate. Parameters such as average end-to-end delay (AEED), packet delivery ratio (PDR), and average energy dissipation (AED) in the network are used to analyze the performance of the existing routing protocols. Average packet delay is denoted as average time required for transferring a data packet from source sensor to the destination actor. Packet delivery ratio is defined as the number of packets that are successfully transferred from source sensor to the destination actor. Average energy dissipation is the amount of energy consumed for network establishment and to transfer the event information from source sensor to destination actor. 3 Routing in Sensor–Actor Networks Routing is an important issue in the network layer. The essential function of this layer is to forward the data from source to destination in a multi-hop fashion [64]. In WSAN, sensors sense the physical area and forward the event information to an actor. The actor processes the information and performs reliable actions on the event area. The design goal of any routing protocol in WSAN need to be simple, energy efficient, and distributed. The existing routing protocols of WSAN are broadly classified into cluster based and non-cluster based protocols. The cluster based protocols virtually divide the nodes into groups using their physical properties. The key idea of these protocols is to exploit the capabilities of actors to reduce the overhead on the sensors. The non-cluster based Clustering can be defined as the virtual partitioning of the nodes into various groups based on the distance between them [65]. In WSAN, cluster head acts as a local coordinator for its cluster member in inter-cluster and intracluster routing for proper utilization of resources. The gateway node works as the intermediate node for two cluster heads. The process of clustering is a combination of two phases namely, cluster formation and maintenance. The cluster formation phase deals with the logical partition of the nodes into several groups and selects a suitable node as a cluster head for every group. The maintenance phase helps to preserve the existing clustering structure as long as possible. The existing cluster based routing protocols are described below with their working principles and relative merits and demerits. 4.1 Hierarchical, Reliable, and Energy Efficient Routing Protocol (HEROP) Eduardo et al. [66] have proposed a hierarchical, reliable, and energy efficient routing protocol for WSAN. It uses meta-data to create energy efficient clusters. HEROP consists of three phases namely discovery phase, joining phase, and routing phase. In discovery phase, each sensor discovers its neighbor information using location-based protocols. The discovery phase is controlled by three parameters such as discovery protocol frequency, execution time, and inactivity time. In joining phase, sensors use the routing tables created in the discovery phase to transmit the event information to the nearest cluster head. Finally, in routing phase data is transferred from sensors to a sink via cluster head. HEROP is a scalable approach, and consider sensors energy while transmitting data to them. Hence, it is a energy efficient mechanism. It also provides fault tolerance routing and reliable data transmission in the network. HEROP does not consider the node heterogeneity property. The actors mobility control, coordination among actors and sensors are also not addressed properly. Packet delivery ratio (PDR) for the HEROP is shown in Fig. 2. The number of mobile actors for each evaluation is considered to be 8, 10, and 12. It is observed that PDR decreases with increase in the number of sensors for a particular set of actors. Further with increase in number of 123 Int J Wireless Inf Networks Fig. 2 Packet delivery ratio versus number of sensors (HEROP) actors, the PDR also increases for any fixed set of sensors. Figure 3 shows the average energy dissipation (AED) decreases with increase in number of sensors. This is due to the fact that similar functional properties are assigned to sensors and actors during data transmission which causes high communication overhead in sensors. Figure 4 depicts the average end-to-end delay (AEED) which is directly proportional to number of sensors for any fixed number of actors, where as it is inversely proportional to the number of actors with fixed number of sensors. The HEROP has to address the mobility of actors and node heterogeneity in order to improve its performance. 4.2 Hierarchical Geographic Clustering Protocol (HGCP) Haidong et al. [67] have proposed a three level coordination model using hierarchical geographic clustering mechanism. In this protocol, network area is divided into fixed zones called as virtual grids which are used to optimally divide the network area and distribute the work load among actors. Fig. 3 Average energy dissipation versus number of sensors (HEROP) 123 Fig. 4 Average end-to-end delay versus number of sensors (HEROP) The main objective of sensor-sensor coordination is to gather event information in an energy efficient manner in the deployed area. The size of the virtual grid depends on the sensor range. The distance between any two sensors in adjacent grids should not go beyond the sensor range. The virtual grid is a square, where the length of each side is r units defined as, R r  pffiffiffi 5 ð3Þ where R is the range of a sensor. In every grid, one sensor acts as a cluster head based on its residual energy. The cluster head aggregates the data received from its associated cluster members and forward it to an appropriate actor. In sensor–actor coordination, each cluster head maintains a routing table which consists of nearest actor information. The cluster head is responsible for monitoring and transferring data to its closest actor. The actor–actor coordination has been divided into two schemes namely, action-first scheme and decision-first scheme. In the actionfirst scheme, the actor starts performing action and informs it to other actors. Then the actors take decision independently whether to join in the action or not. On the other hand, decision-first scheme communicates with its neighbor actors before performing action on the event area. This scheme distributes the load properly among actors and maximize the overall task performance. The reduced size of the grid area causes formation of more number of clusters that reduces the network lifetime. HGCP does not address the delay parameter properly which is important in real time application of WSAN. Finally, it assumes that both the sensors and actors are static. It is observed that HGCP does not perform well for average end-to-end delay with the increase in number of sensors as shown in Fig. 5. HGCP considers both the sensors and actors as static, which is a non-realistic assumption in many WSAN applications. When the actor Int J Wireless Inf Networks Fig. 5 Average end-to-end delay versus number of sensors (HGCP) moves far away from the cluster head, it results in the increase in packet delay. From Fig. 6, it can be observed that the value of AED is directly proportional to the number of sensors. The reduced size of the grid area causes formation of more number of clusters and also consume lots of energy of the sensors. 4.3 QoS-Aware Routing Protocol (QARP) Boukerche et al. [68] proposed a QoS aware routing protocol for WASN. It uses a publish-subscribe approach for interaction among nodes in the network. QARP comprises of three phases. In the first phase, sink initiates the hop tree construction process. The hop tree is used for flooding the subscription messages and data around the network. Similarly, every actor initiates their hop tree construction to divide the entire network into acting areas. In the second phase, sink and actor flood the subscription messages across the network. The final phase is responsible for packet delivery from sensor nodes to the sink. In QARP, whenever a sensor senses an event in its sensing range, it checks its subscription table to find whether any interest on the event is registered or not. If any Fig. 6 Average energy dissipation versus number of sensors (HGCP) node is registered, then the sensor selects a path to transfer the packet based on its priority. A queuing model is designed to transfer low priority packets, which does not consider latency in a less-expensive path to reduce the energy loss in the network. A packet scheduling approach is used to set the weights for flow types. QARP uses direct diffusion technique to transfer the event to the actors. It also uses packet prioritization technique to give priority to the delay sensitive packets. QARP assumes that both the sensors and actors are static, which is non-realistic nature for many WSAN applications. It does not properly utilize resource rich actors, which causes extra communication burden on sensors and leads to reduction in network lifetime. The AEED of QARP for 8, 10, and 12 actors are shown in Fig. 7 and it is observed that AEED increases with the increase in number of sensors and decreases with the number of actors. The packet delay increases because of mobile actors and frequent path failures. The value of AED is directly proportional to the number of sensors, because it uses sensor for complex process such as hop tree construction, maintenance, and flooding mechanism as shown in Fig. 8. Figure 9 depicts the PDR of the existing QARP. Due to the actor mobility a lot of packets are dropped before reaching the destination. Hence, the PDR of the QARP does not perform well under mobile actors and reduce with the increase in number of actors. The simulation results indicate that QARP does not perform well under node heterogeneity conditions and mobile actors. 4.4 Event Driven Clustering Protocol (EDCP) Tommaso et al. [69] have proposed an on-demand routing protocol for WSAN. In EDCP, clusters are created on-thefly when an event is triggered. It deals with both sensor– actor and actor–actor coordination. In the sensor–actor coordination, an aggregation tree is created between each Fig. 7 Average end-to-end delay versus number of sensors (QARP) 123 Int J Wireless Inf Networks events occur simultaneously or the occurrence of events are frequent in nature. The simulation is performed under concurrent event scenarios. The event waiting time is defined as the difference between occurrence of an event and the starting time of an action. Figure 10 shows the waiting time for an event under multiple events scenario. The event waiting time is increased with the increase in number of sensors. The simulation results indicate that EDCP does not perform well, when multiple events occur concurrently. Fig. 8 Average energy dissipation versus number of sensors (QARP) Fig. 9 Packet delivery ratio versus number of sensors (QARP) actor and its sensors in the transmission range. Each sensor consists of four states namely idle, start-up, speed-up, and aggregation. Initially every sensor is in idle state. A sensor enters into start-up state when it detects an event or an event information is received from any of its neighbors. If the sensor does not reach its minimum event reliability threshold value, then it enters into speed-up state. The sensor enters into aggregation state, if its event reliability is more than maximum threshold value. In the speed-up state, each sensor uses greedy routing scheme to reduce the packet end-to-end latency. The aim of the aggregation state is to reduce the energy consumption by performing data aggregation on each sensor. For actor–actor coordination a distributed solution has been proposed based on a real-time auction protocol. A localized auction takes place in every overlapping area. The actor which has more residual energy and also takes less completion time of an action wins the auction. EDCP utilizes actors properly in data communication to reduce the burden on sensors. It also uses greedy routing scheme to improve packet delay in sensor–actor coordination. EDCP is not well suited for scenarios where multiple 123 4.5 Hierarchical, Energy Efficient Routing Protocol (HEERP) Fei et al. [70] have proposed an energy efficient member recognition protocol. Each actor segregates the network area into domains consisting a set of sensors. Each domain is further divided into ripples based on the distance between sensor and the actor. Few sensors are randomly selected as masters to aggregate the received data. The network architecture of HEERP is shown in Fig. 11. Suppose master node D is not able to reach master node B (due to its battery dissipation or the unreliable wireless link), then the master node D checks the reachability of B’s neighbors i.e., C and G. If any one of them is reachable, then master node D forwards the data to it. If C and G both fail to find any neighbor, then it goes back one more ripple to master node F. If F finds D’s neighbors it delivers the packet, otherwise the packet will be dropped. HEERP constructs virtual domains and zones around an actor, which is similar to the hierarchical geographic clustering protocol (HGCP). HEERP does not perform well in sparse network and the delay parameter has also not been addressed properly. The performance of the HEERP using AEED parameter is shown in Fig. 12. When a node is not reachable, then it forwards the packet to its neighbor which causes high packet delay. The value of AEED is inversely proportional Fig. 10 Event waiting time for concurrent events scenario in EDCP Int J Wireless Inf Networks maximize its coverage area based on the density of the sensors. It uses energy constrained sensor nodes for data aggregation process that consume lots of energy. Further, WBMP does not address the end-to-end delay parameter properly. The end-to-end delay parameter is used to analyze the simulation results. Figure 13 shows the value of AEED for the WBMP under heterogeneous node scenario with 8, 10, and 12 actors respectively. It may be observed that WBMP does not address end-to-end delay parameter properly. 4.7 Sensor–Actor Coordination Protocol (SCP) Fig. 11 Network architecture of HEERP Fig. 12 Average end-to-end delay versus number of sensors (HEERP) to the number of actors. The simulation results also reveal that HEERP causes a lot of packet delay in the network. 4.6 Weighted Bi-partite Matching Protocol (WBMP) WBMP deals with a data forwarding mechanism, where sensors form clusters using weighted bi-partite matching method [71]. The distance between sensor and actor is denoted as the weight of an edge. The cluster head size is restricted to limit the traffic in a cluster. If any cluster is full, then the sensor will choose next best cluster using greedy mechanism. Whenever a new node joins the network, it gets the cluster information from its neighbors. It sends the cluster join message to its nearest actor. To improve the network lifetime, WBMP employs resource rich actors as cluster heads. An actor collects the event information from its associated cluster members and perform reliable actions in the event area. To reduce the latency between sensing and acting tasks, the actor should Zhicheng et al. [72] have proposed a sensor–actor coordination for WSAN based on Voronoi diagram and the shortest path tree. It is divided into various rounds and each round consists of two phases namely cluster set-up phase and data transmission phase. During cluster set-up phase, actor acts as a cluster head and sends its residual energy to the sink. The sink constructs a weighted actor Voronoi diagram and sends back to an actor. Finally, every actor informs its Voronoi region information to its cluster members. In the data transmission phase, the shortest path tree is calculated from sensor to the actor in a cluster. The weight of each edge is represented as the energy required to send a data packet. Sensors transmit their data to the actor using shortest path tree to reduce the packet delay. It requires complete topological information and also consumes a lot of energy to calculate the shortest path tree. SCP does not consider sensor–sensor and actor–actor coordination. It assumes that both the sensors and actors are static in nature. SCP requires complete topological information and consumes lot of energy to calculate the shortest path tree. The average end-to-end delay, average energy dissipation, and packet delivery ratio for the existing SCP under actors mobility are shown in Figs. 14, 15, 16 respectively. In case of mobile actors, the lifetime of a Fig. 13 Average end-to-end delay versus number of sensors (WBMP) 123 Int J Wireless Inf Networks path is very less as compared to the static actors scenario. The sensors around an actor should always update their neighbor list due to the mobility of actors. This mechanism consumes lot of energy and also degrades the packet delivery ratio. Hence, SCP does not perform well under static sensors and mobile actors, which is a realistic scenario for WSAN. improve the packet delay. Finding k-hop IDS is a NP hard problem and requires complete topological information of the network. Hence, it consumes lot of energy of the sensors and also decreases the network lifetime. From Fig. 17 4.8 Distributed Actor Positioning and Clustering Protocol (DAPCP) Akkaya et al. [73] have proposed a distributed actor positioning and clustering protocol. In DAPCP, actors act as cluster heads and are placed in an appropriate positions to improve the coverage and packet delay. The k-hop independent dominating set ðk  IDSÞ is used to find the actor position. In k  IDS, three different messages are exchanged among sensors namely alive, dominator, and border. Each sensor sends an alive message to its k-hop neighbors to announce its existence. If a sensor desires to become a dominator, then it sends a dominator message to its k-hop neighbors. Finally, sensor sends a border message that is at kth hop from a dominator. Each sensor calculates its suitability score to become a cluster head based on the number of received alive and the closest border messages. The suitability score of a sensor is calculated as   B þ Mi ACi si ¼ ð4Þ 2BTTL where, B indicates sensor probability to become a dominator, Mi is a weight assigned to other nodes with in a khop to the sensor i, ACi is the total number of heard alive messages by node i, and BTTL is the time-to-live of the closest border message. DAPCP uses actor as a cluster head to reduce the communication burden on sensors. It also uses node degree parameter while selecting the cluster head, it leads to Fig. 14 Average end-to-end delay versus number of sensors (SCP) 123 Fig. 15 Average energy dissipation versus number of sensors (SCP) Fig. 16 Packet delivery ratio versus number of sensors (SCP) Fig. 17 Average energy dissipation versus number of sensors (DAPCP) Int J Wireless Inf Networks also we can observe that DAPCP protocol consumes lot of energy from the sensors and reduces the network lifetime. 5 Analysis of Cluster Based Routing Protocols In the previous section, cluster based routing protocols are analyzed individually to study their performance in isolation. To derive an overall inference about the cluster based routing protocols, the performance parameters are analyzed together. The simulation results of the cluster based protocols presented in the previous section. HEROP is a scalable and fault tolerance routing protocol. It is an energy efficient approach as it uses meta data to create clusters, but it does not consider node heterogeneity and actors mobility control. Similarly, WBMP is a scalable approach. The size of each cluster is restricted to limit the traffic load in the cluster. A sensor acts as a cluster head, which causes lot of communication burden on sensor and results in degrading the network lifetime. The HEERP, SCP, QARP, and HGCP have considered both sensors and actors are static. This assumption is not feasible for many WSAN applications. EDCP creates cluster on-the-fly, but it does not perform well when multiple events occur or events are frequent in nature. The DAPCP finds optimal position of an actor based on the sensors density. It requires complete topological information and consumes lot of energy. To visualize the merits and demerits of all cluster based routing protocols, we have listed them together in Table 2. It may be observed that with respect to AED, AEED, PDR parameters under consideration, HEROP outperforms other cluster based routing protocols. Ta;n 8 < Ta;n1  a if SQ [ 0 ¼ Ta;n1 : if SQ ¼ 0 b ð5Þ where, Ta;n1 is the previous wake-up interval, and a and b are the constants. The a and b values determine the packet delay and energy consumed by the sensors. The values of a and b are defined as 2. Whenever a sensor wakes up, it broadcasts a beacon message consisting of its own id, the start time of its next sleep period, and its residual energy. In the geographical routing phase, it uses a greedy mechanism to transfer data to the forwarding candidate set. Each node forwards the packet to an active neighbor that is closer to the destination than itself with a certain threshold distance. This forwarding mechanism provides loop free routing to the destination for transferring the data, but it cannot deliver the packet to the destination in the presence of holes. Since WSAN is a dense network, there is no scope for existence of holes in the network. DEARP provides reliable data transmission and also uses greedy routing scheme to improve the packet delay. The values of AEED, AED, and PDR for the existing DEARP is shown in Figs. 18, 19, 20 respectively. DEARP does not specify how to select an actor to forward the data. It just forwards the data to a nearest actor. Hence, it creates a problem for border nodes which are having more than one closest actor. The path lifetime is less under actors mobility as compared to static actors scenario [75, 76]. The sensors around the actor should always update their neighbor list because of actors mobility. The simulation results also reveal that DEARP does not perform well under actors mobility. 6.2 Anycast Tree Based Communication Mechanism (ATCM) 6 Non-cluster Based Routing Protocols The non-cluster based routing protocols either use flooding or broadcast mechanisms for communication. They do not structure the physical network into virtual groups. Different non-cluster based routing protocols are described below with their working principles and relative merits and demerits. 6.1 Delay Energy Aware Routing Protocol (DEARP) DEARP consists of two major components namely, a random wake-up scheme and geographic routing protocol [74]. The main idea of the random wake-up scheme is that every node should wake-up once in every slot for a predetermined time. The sensor active period ðTa;n Þ depends on its queue size ðSQ Þ which is defined as, Wen et al. [77] have proposed an anycast communication mechanism for a hybrid sensor actuator network. ATCM constructs an anycast tree rooted at the sensor. A sink can dynamically join in the sink tree or leave. It uses directed diffusion mechanism, which is a data-centric and reversepath based communication for sensor networks. In ATCM, every sensor forms an anycast tree. When a sink enters into the network, a new branch leading to the sink must be added to each anycast tree. After initial setup, if a data packet arrives sensor looks up its anycast table to forward the packet to the nearest sink. Every sink periodically sends a beacon packet to refresh anycast table entries. ATCM approach is similar to the direct diffusion routing protocol. The anycast table size is controlled by storing only nearest sink information. It performs well when the updates from the sink are not frequent. ATCM mechanism has been simulated using IEEE 802.11 MAC protocol, 123 Int J Wireless Inf Networks Table 2 Comparison of cluster based routing protocols Year First author Approach Merits Demerits 2005 Tommaso Melodia Event driven clustering protocol (EDCP) Clusters are created on-the-fly It does not perform well when multiple events occur It is a state less protocol Not feasible when events are frequent in nature 2005 2005 Fei Hu Yen-Ting Lin Hierarchical, energy dfficient routing protocol (HEERP) Weighted bi-partite matching protocol (WBMP) It uses greedy routing scheme to reduce packet delay It is an energy efficient member recognition protocol Sensors are used for data aggregation process It performs well in the dense network It does not address properly the real time requirements of WSAN It is a scalable approach Sensor is used for data aggregation process The packet delay parameter is not addressed properly The size of each cluster is restricted to limit the intra-cluster traffic Actor act as a cluster head 2006 Haidong Yuan Hierarchical geographic clustering protocol (HGCP) It discuss about three level coordination mechanism in WSAN It uses virtual grids Small grid area Actor acts as a cluster head Delay parameter is not addressed properly Action-first and decision first action schemes are discussed in actor–actor coordination 2006 Boukerche QoS-aware routing protocol (QARP) It assumed that both the sensors and actors are static It uses packet prioritization scheme It assumes that both the sensors and actors are static A publish/subscribe approach is used among nodes Actor nodes are not used properly Hop-trees are constructed to flood the subscription messages and data 2007 Kemal akkaya Distributed actor positioning and clustering protocol (DAPCP) The K-IDS mechanism is used to find the actor positions Computing K-IDS is a NP hard problem Actor acts as a cluster head It requires complete topological information about a network The cluster head is selected based on its neighborhood degree 2009 ZhiCheng Dai Sensor–actor coordination protocol (SCP) Actor Constructs its Voronoi diagram It considers only sensor–actor coordination Non-overlapping clusters are created It assumed that both the sensors and actors are static Shortest path tree is calculated from sensors to actor in a cluster 2012 Eduardo Canete Hierarchical, reliable, and energy efficient routing protocol (HEROP) It uses meta data to create clusters Node heterogeneity is not considered It is a scalable approach Lack of mobility control of actor nodes Energy efficient mechanism It does not provide coordination mechanism among nodes Fault tolerance routing Reliable data transmission which is designed explicitly for WLANs. It is generally not suitable for energy constrained networks such as WSN and WSAN [78, 79]. In this paper, the ATCM is simulated and analyzed using IEEE 802.15.4 MAC protocol. The values of PDR, 123 AEED, and AED of ATCM under IEEE 802.15.4 MAC protocol are depicted in Figs. 21, 22, 23 respectively. ATCM mechanism has assumed that always sensors are in active state. This assumption leads to degrade the ATCM performance in WSAN. Int J Wireless Inf Networks Fig. 18 Average end-to-end delay versus number of sensors (DEARP) Fig. 19 Average energy dissipation versus number of sensors (DEARP) Fig. 21 Packet delivery ratio versus number of sensors (ATCM) Fig. 22 Average end-to-end delay versus number of sensors (ATCM) Fig. 20 Packet delivery ratio versus number of sensors (DEARP) Fig. 23 Average energy dissipation versus number of sensors (ATCM) 6.3 Delay Sensitive Routing protocol (DSRP) Ngai et al. [80] have proposed a reliability-centric framework for event reporting in WSAN. The reliability in WSAN not only depends on the accuracy, but also depend on the freshness of data. The network area is segregated into virtual grids for event monitoring. DSRP consists of three key phases namely, data aggregation phase, priority based event reporting, and adaptive actor allocation mechanism. In the aggregation phase, an aggregating node 123 Int J Wireless Inf Networks collects the event data and calculates the median m of it. This node compares each sensor data xi with m and removes those sensors data that has a difference greater than a threshold value. Then in a grid, the aggregating node re-calculates the mean value from the remaining sensors data. The reliability ðrÞ for the aggregated data from a grid can be computed as, r ¼ 1  fg ¼ 1  Nx X Nx ! ðfs Þi ð1  fs ÞNx i i! N ð  i Þ! x Nx ð6Þ i¼ 2 phase. Hence, DSRP creates lot of communication burden on resource conservative sensors and results in reducing the network lifetime. In this paper, the existing DSRP is simulated and analyzed under the IEEE 802.15.4 MAC protocol. The PDR, AEED, and AED in the network for the existing DSRP under IEEE 802.15.4 MAC protocol are depicted in Figs. 24, 25, 26 respectively. The values of AEED and AED increases with the increase in number of sensors, but PDR is inversely proportional to the number of sensors. From simulation results, it can be observed that DSRP does not work properly under node heterogeneity conditions and mobile actors. where, fg is the failure probability of grid g on data aggregation, Nx is the number of nodes in the grid, and fs is a ratio of the malfunctioned sensors. The aggregating node may act as a reporting node to forward the aggregated data to an actor. Each sensor decides independently whether it will serve as a reporting node or not according to the number of data reports to be transferred by it. In the priority based event reporting module, each sensor is enabled with a priority queue to assign priorities for the received data. The queue utilization acts as an index for route selection. An adaptive actor allocation mechanism forwards the sensor data to a nearest actor to reduce the packet delay. It computes event frequency freq for every grid. Then, the network area is equally partitioned into two subfields based on the frequency distribution. Actors are equally allocated to every subfield. It is a recursive process and repeats till each subfield contains only one actor. DSRP is a reliability centric framework and uses fault tolerant data aggregation mechanism to eliminate the faulty sensors in the network. DSRP was also simulated using the IEEE 802.11 MAC protocol, which is a non-realistic in nature for many WSAN applications. It assumes both sensors and actors are static. The actors are not used properly in network establishment and data transmission GBRP is a distributed approach, where nodes take local decisions while forwarding data to the destination based on geometric mechanism [81]. GBRP imposes low communication overhead as it does not require neighborhood information. The actors are utilized properly to reduce the energy consumption in sensors. GBRP uses separate protocols to handle the broadcast mechanisms for sensor and actor respectively. In GBRP, R and k  R denotes the transmission range of a sensor and actor respectively. In actor broadcast protocol, sensor transfer its data to a nearest actor. Then actor broadcasts the packet to the remaining actors and sensors in the network. In sensor broadcast mechanism, each sensor checks whether the broadcast packet is already received or not. If it is already received then sensor discards the packet, else it broadcast the packet to the remaining sensors. GBRP broadcast packets in the entire network area instead of concentrating on a specific region. From Fig. 27, It can be observed that the value of PDR increases with the increase number of actors and inversely proportional to the number of sensors. Fig. 24 Packet delivery ratio versus number of sensors (DSRP) Fig. 25 Average energy dissipation versus number of sensors (DSRP) 123 6.4 Geometric Broadcast Routing Protocol (GBRP) Int J Wireless Inf Networks Fig. 26 Average end-to-end delay versus number of sensors (DSRP) Fig. 28 Packet delivery ratio versus number of sensors (PARP) receiver. PARP requires lot of memory to store the large size routing table. It chooses an energy efficient from the routing table to transfer its data. This process increases packet delay as shown in Fig. 28. PARP is not feasible for dense network as the routing table size increases with the increase in network size. 6.6 Power Controlled Routing Protocol (PCRP) Fig. 27 Average end-to-end delay versus number of sensors (GBRP) 6.5 Power Aware Routing Protocol (PARP) Cayirci et al. [82] have proposed a power aware many-tomany routing protocol for WSAN. It has two versions. In the first version, every node transmits data using same transmission power. Actors send registration packets to the sensors consisting of their data interest. The registration message consists of node identification, actor identification, the minimum number of hops required to reach a node from an actor, minimum power, total power available, and task(s). When a sensor receives the registration packet, it builds the registration table contains actor id, uplink node, echelon, minimum power, total power, and tasks information. Each sensor constructs a routing table from the registration table consisting a record for every unique downstream node and sensing task. Each sensor forwards its data to its nearest sensor towards an actor. An actor can de-register from a sensing task by broadcasting a de-registration packet to the respective sensor. The task de-registration message consists of node identification, actor identification, and tasks. In the second version, sensor dynamically adjusts its transmission power based on the intended distance to the Yangfan et al. [83] have proposed a real-time power aware routing protocol. It forwards the packets in a stateless manner. The intermediate nodes do not maintain any path to a nearest actor. Each sensor sets its power level based on the distance to the intended neighbor. PCRP is a cross layered approach which requires physical layer information. In PCRP, sensor selects a neighbor according to the packet delay deadline and energy required to forward the packet. The packet delay at a sensor u is computed as, DelayðuÞ ¼ tprop þ tproc þ tq þ ttran ð7Þ where, tprop is the propagation delay from sensor u to its intended neighbor, tproc is the processing delay of u, tq is the queuing delay of u during which the packet is waiting to be sent out, and ttran is the transmission delay which is related to the channel bandwidth and packet size. The intermediate sensor can estimate the future hop-by-hop delay based on the delay estimation of the upstream hops. It does not require feedback from downstream sensors to estimate the future hop-by-hop packet delay. The average number of hops that a packet can go through per second is computed using the following equation: s¼ 1 davg hop where, davg davg hop ð8Þ hop is denoted as ¼ b  davg hop þ ð1  bÞdðsi Þ ð9Þ 123 Int J Wireless Inf Networks where, b is a constant, dðsi Þ is the delay calculated at sensor si . PCRP requires 2 to 3  hop neighborhood information to compute the packet delay, that produces control packet overhead in dynamic networks. The delay is estimated only from the upstream sensors. PCRP does not consider feedback from the recently visited downstream nodes. This type of estimation performs well only for periodic data transfer mode. Due to the transmitter power control, a sensor uses small transmitting power to the nearest node. This information may not be sensed by other neighbors that are far away and want to transfer the packets at the same time. This scenario is called as a hidden terminal problem. PCRP has been simulated using IEEE 802.11 MAC protocol which is specifically designed for WLAN. IEEE 802.11 MAC protocol is not feasible for energy constrained sensor–actor networks. In this paper, PCRP is simulated using IEEE 802.15.4 MAC protocol, that is specially designed for dense and energy constrained sensor networks. The values of AEED, PDR, AED for the PCRP is analyzed under IEEE 802.15.4 MAC protocol and depicted in Figs. 29, 30, 31 respectively. It can be observed that PCRP does not perform well under IEEE 802.15.4 MAC protocol. Fig. 30 Average energy dissipation versus number of sensors (PCRP) 6.7 Scalable Source Routing Protocol (SSRP) Fuhrmann has proposed a scalable source routing protocol for sensor–actor networks [84]. The working principle of SSRP is similar to overlay routing in a virtual network structure. SSRP is a reactive protocol and uses a proactive mechanism for the virtual ring construction. Each node has a location independent address and consists of physical neighbors. The physical network and virtual ring are constructed using distance metric. In the physical network, distance is computed based on the hop count, while in the virtual ring it is measured from the difference between absolute address of two nodes. If a sensor cannot retrieve Fig. 31 Packet delivery ratio versus number of sensors (PCRP) Fig. 32 Average energy dissipation versus number of sensors (SSRP) Fig. 29 Average end-to-end delay versus number of sensors (PCRP) 123 entire source route to the destination from its own cache, then it dynamically constructs a route to the intermediate node. The source selects an intermediate node that is near to the destination. This type of routing may not always produce shortest paths and also increases packet end-to end delay. From Fig. 32, it can be observed that SSRP consumes lot of energy to transfer the data. SSRP also not specified which actor is selected as a destination to transfer the data Int J Wireless Inf Networks architecture for communication, where the sink collects all the sensor data and takes decision. The semi-automated architecture incurs high end-to-end delay and rapid energy depletion on the sensors closer to the server called as funneling problem. The inclusion of backhaul nodes may increase network design complexity. Figure 34 depicts the value of AEED for the existing LMCA under node heterogeneity conditions and mobile actors. To reduce the delay the sensors should forward the event information directly to the nearest actor instead of transferring via a sink. Fig. 33 Average end-to-end delay versus number of sensors (SSRP) 7 Analysis of Non-cluster Based Routing Protocols Fig. 34 Average end-to-end delay versus number of sensors (LMCA) and it may leads to wrong selection of actor that is far away from the event area. It can be observed that SSRP degrades the average end-to-end delay parameter as shown in Fig. 33. 6.8 Routing Protocol for Light Monitoring and Control Application (LMCA) In LMCA, sensor-sensor coordination and actor–actor coordination is performed in separate channels with different capacity, cost, and reliability [85]. Both sensor and actor networks are connected to a central server, that handle the user request and also provides coordination between both the networks. The backhaul nodes are resource rich nodes, act as a mediator between sensor and actor networks. The sensor network uses data-centric routing architecture. On the other hand, the actor network uses point-to-point communication to improve the network performance. The centralized server gathers data from both sensor and actor networks to take decisions. LMCA uses semi-automated In the previous section, non-cluster based routing protocols are analyzed individually to study their performance in isolation. To derive an overall inference, the performance of all the non-cluster based routing protocols are analyzed together. The simulation results of the non-cluster based protocols presented in the previous section. DEARP uses a greedy mechanism and assures loop free path selection while transferring the data. It provides reliable data transmission and each sensor uses periodic wake-up mechanism to improve the network lifetime. The PCRP, DSRP, and ATCM protocols have been simulated using IEEE 802.11 MAC protocol. It is specially designed for WLAN and does not require any energy efficient mechanisms. SSRP may not always produce the shortest paths and requires complete topological information. It does not select destination actor properly, which may cause delay in data transmission. GBRP is useful for only query based applications. But, it biases the energy consumption and delay as it uses broadcast mechanism to transfer the data. LMCA uses a semi-automated architecture, which produces high delay in the network. DEARP did not specify how to select an destination for border sensors. It requires MAC layer information for calculating the sleep schedule of a sensor and actors mobility is also not considered properly. To visualize the merits and demerits of all non-cluster based routing protocols, we have listed them together in Table 3. It may be observed that with respect to all the three parameters under consideration, DEARP outperforms other non-cluster based routing protocols. 8 Comparison of HEROP and DEARP In the cluster based routing protocols HEROP outperforms others with respect to all the three parameters packet delivery ratio, average end-to-end delay, and average 123 Int J Wireless Inf Networks Table 3 Comparison of non-cluster based routing protocols Year First author Approach Merits Demerits 2005 Arjan Durresi Delay energy aware routing protocol (DEARP) Sensor sleep-active mechanism is used to improve network lifetime It does not specify how to select an actor for border sensors It uses a greedy mechanism to transfer the data Requires MAC layer information for calculating the traffic load of a sensor It assure no loops in data forwarding Actor mobility is not considered Reliable data transmission 2005 Wen Hu Anycast tree based communication mechanism (ATCM) It uses directed diffusion mechanism Complex routing table updating The anycast table size is controlled by storing only nearest sink information It was simulated using IEEE 802.11 MAC protocol designed for WLANs It uses broadcasting mechanism to transfer the data It is useful only for query based applications It imposes low communication overhead It does not address packet delay properly It does not provide data reliability 2005 Arjan Durresi Geometric broadcast routing protocol (GBRP) 2005 cayirci Power aware routing protocol (PARP) 2006 Ngai Delay sensitive routing protocol (DSRP) It employs transmitter power control It produces traffic load It constructs many-to-many multi-cast routing tree High overhead for updating routing table Actors are used properly in communication It creates hidden terminal problem It is a reliability centric framework It was simulated using IEEE 802.11 MAC protocol designed for WLANs It create virtual grids It assumes that both the sensors and actors are static It has designed priority based event reporting Actor nodes are not used properly in data communication Poor scalability A fault tolerant data aggregation model is designed to remove faulty sensors 2006 Thomas Scalable source routing protocol (SSRP) It is a reactive protocol It does not always produce shortest path It uses proactive mechanism for virtual ring construction It does not consider delay parameter The physical network and virtual ring are constructed using distance metric Complete topological information is required Actor selection is not specified properly in sensor–actor coordination 2006 Suet-Fei Li Routing protocol for light monitoring and control application (LMCA) It uses heterogeneous architecture Backhaul nodes increases design complexity It uses semi-automated mechanism It produces high end-to-end delay Backhaul nodes are used as mediators between two networks 2007 Yangfan Zhou Power controlled routing protocol (PCRP) It is a power aware mechanism It creates hidden terminal problem It forwards the packet in a stateless manner It was simulated using IEEE 802.11 MAC protocol designed for WLANs It is a cross layered mechanism energy dissipation in the network. Similarly, DEARP is shown to have superior performance among non-cluster based routing protocols. A comparative analysis among these two protocols is shown in Table 4 for all the three 123 parameters under consideration. It may be observed that HEROP performs better as compared to DEARP. Hence, cluster based routing protocols have better scope of application in WSAN. Int J Wireless Inf Networks Table 4 Comparative analysis of HEROP and DEARP Protocol/No. of sensors HEROP DEARP PDR AEED (ms) AED (J) PDR AEED (ms) 100 97 200 94 300 2.1 0.48 78 4.7 2.5 0. 52 76 5 0.83 92 2.8 0.6 73 5.3 0.87 400 89 3.2 0.65 71 5.6 0.9 500 85 3.7 0.72 68 5.8 0.94 600 82 4 0.79 66 6.2 0.99 700 80 4.5 0.87 63 6.7 1.02 800 77 4.9 0.98 61 7 1.1 900 75 5.5 1.07 58 7.2 1.21 1000 72 6.1 1.16 54 7.6 1.3 9 Conclusion In this paper, we have made a comparative analysis of wireless sensor–actor network (WSAN) routing protocols. The protocols are divided into two categories i.e., cluster based and non-cluster based. Simulation has been carried out in NS2 for each protocol under similar environment. Each protocol is analyzed in isolation with respect to average end-to-end delay, packet delivery ratio, and average energy dissipation varying the number of sensors and actors. It is observed that among cluster based protocols hierarchical, reliable, and energy efficient routing protocol (HEROP) outperforms others. Similarly delay energy aware routing protocol (DEARP) has superior performance in non-cluster based protocols. Comparative analysis between HEROP and DEARP reveals that HEROP performs well with respect to all the three parameters under consideration. Hence, it can inferred that cluster based protocols are more popular among the research community of WSAN. References 1. A. Rezgui and M. Eltoweissy, ‘‘Service-oriented sensor–actuator networks: Promises, challenges, and the road ahead,’’ Computer Communications, vol. 30, no. 13, pp. 2627–2648, 2007. 2. T. Melodia, D. Pompili, V. Gungor, and I. Akyildiz, ‘‘Communication and coordination in wireless sensor and actor networks,’’ IEEE Transactions on Mobile Computing, vol. 6, no. 10, pp. 1116–1129, 2007. 3. R. Jafari, A. Encarnacao, A. Zahoory, F. Dabiri, H. Noshadi, and M. Sarrafzadeh, ‘‘Wireless sensor networks for health monitoring,’’ in The Second Annual International Conference on Mobile and Ubiquitous Systems: Networking and Services, 2005, pp. 479–481. 4. D. Trossen and D. Pavel, ‘‘Sensor networks, wearable computing, and healthcare applications,’’ IEEE Pervasive Computing, vol. 6, no. 2, pp. 58–61, 2007. AED (J) 0.81 5. I. Akyildiz, S. Weilian, Y. Sankarasubramaniam, and E. Cayirci, ‘‘A survey on sensor networks,’’ IEEE Communications Magazine, vol. 40, no. 8, pp. 102–114, 2002. 6. J. Al-Karaki and A. Kamal, ‘‘Routing techniques in wireless sensor networks: a survey,’’ IEEE Wireless Communications, vol. 11, no. 6, pp. 6–28, 2004. 7. X. Wang, L. Ding, B. Dao-Wei, and S. Wang, ‘‘Energy-efficient optimization of reorganization-enabled wireless sensor networks,’’ Sensors, vol. 7, no. 9, pp. 1793–1816, 2007. 8. N. Vasanthi and S. Annadurai, ‘‘Sleep schedule for fast and efficient control of parameters in wireless sensor–actor networks,’’ in First International Conference on Communication System Software and Middleware. IEEE, 2006, pp. 1–6. 9. R. Rajagopalan and P. Varshney, ‘‘Data-aggregation techniques in sensor networks: a survey,’’ IEEE Communications Surveys Tutorials, vol. 8, no. 4, pp. 48–63, 2006. 10. G. Shah and M. Hassan, ‘‘A reliable event response framework for wireless sensor and actor networks,’’ in IEEE Workshops of International Conference on Advanced Information Networking and Applications, 2011, pp. 396–401. 11. S.-L. Wu, Y.-C. Tseng, and S. Ivan, Eds., Energy conservation in sensor and sensor–actuator networks. Auerbach Publications, 2007. 12. V. Gungora, M. Vurana, and O. Akanb, ‘‘On the cross-layer interactions between congestion and contention in wireless sensor and actor networks,’’ Ad Hoc Networks, vol. 5, no. 6, pp. 897 – 909, 2007. 13. A. Zamanifar, M. Sharifi, and O. Kashefi, ‘‘Self actor–actor connectivity restoration in wireless sensor and actor networks,’’ in First Asian Conference on Intelligent Information and Database Systems. IEEE, 2009, pp. 442–447. 14. Y. Gao, J. Wang, and X. Song, ‘‘Data collection scheme of mobile sink in wireless sensor and actor networks,’’ in 11th World Congress on Intelligent Control and Automation. IEEE, 2014, pp. 2505–2508. 15. Z. Cai, X. Ren, G. Hao, B. Chen, and Z. Xue, ‘‘Survey on wireless sensor and actor network,’’ in 9th World Congress on Intelligent Control and Automation. IEEE, 2011, pp. 788–793. 16. C. Konstantopoulos, I. Venetis, G. Pantziou, and D. Gavalas, ‘‘An efficient event handling protocol for wireless sensor and actor networks,’’ in IEEE Symposium on Computers and Communication. IEEE, 2014, pp. 1–6. 17. A. Boukerche, R. Araujo, and L. Villas, ‘‘A wireless actor and sensor networks qos-aware routing protocol for the emergency preparedness class of applications,’’ in 31st IEEE Conference on Local Computer Networks. IEEE, 2006, pp. 832–839. 123 Int J Wireless Inf Networks 18. H. Peng, W. Huafeng, and G. Chuanshan, ‘‘Elrs: an energy-efficient layered routing scheme for wireless sensor and actor networks,’’ in 20th International Conference on Advanced Information Networking and Applications, vol. 2. IEEE, 2006, pp. 5–pp. 19. H. Shibo, J. Chen, P. Cheng, Y. Gu, H. Tian, and Y. Sun, ‘‘Maintaining quality of sensing with actors in wireless sensor networks,’’ IEEE Transactions on Parallel and Distributed Systems, vol. 23, no. 9, pp. 1657–1667, 2012. 20. W. Li, E. Chan, M. Hamdi, S. Lu, and D. Chen, ‘‘Communication cost minimization in wireless sensor and actor networks for road surveillance,’’ IEEE Transactions on Vehicular Technology, vol. 60, no. 2, pp. 618–631, 2011. 21. W. Abbas, H. Jaleel, and M. Egerstedt, ‘‘Energy-efficient data collection in heterogeneous wireless sensor and actor networks,’’ in IEEE 52nd Annual Conference on Decision and Control. IEEE, 2013, pp. 4164–4169. 22. M. Kamali, S. Sedighian, and M. Sharifi, ‘‘A distributed recovery mechanism for actor–actor connectivity in wireless sensor actor networks,’’ in International Conference on Intelligent Sensors, Sensor Networks and Information Processing. IEEE, 2008, pp. 183–188. 23. K. Muazzam, G. Shah, M. Ahsan, and M. Sher, ‘‘An efficient and reliable clustering algorithm for wireless sensor actor networks (wsans),’’ in 53rd IEEE International Midwest Symposium on Circuits and Systems. IEEE, 2010, pp. 332–338. 24. K. Selvaradjou, H. Nikhil, A. Franklin, and C. Siva, ‘‘Energyefficient directional routing between partitioned actors in wireless sensor and actor networks,’’ IET communications, vol. 4, no. 1, pp. 102–115, 2010. 25. S. Sedighian, M. Sharifi, S. Azhari, and H. Momeni, ‘‘Service requirements for actor–actor coordination through sensor nodes in wireless sensor actor networks,’’ in International Conference on Innovations in Information Technology. IEEE, 2008, pp. 475–479. 26. N. Sabri, S. Aljunid, R. Ahmad, M. Malik, A. Yahya, R. Kamaruddin, and M. Salim, ‘‘Towards smart wireless sensor actor networks: design factors and applications,’’ in IEEE Symposium on Industrial Electronics and Applications. IEEE, 2011, pp. 704–708. 27. S. Kashi and M. Sharifi, ‘‘Connectivity weakness impacts on coordination in wireless sensor and actor networks,’’ IEEE Communications Surveys and Tutorials, vol. 15, no. 1, pp. 145–166, 2013. 28. F. Senel, K. Akkaya, and M. Younis, ‘‘An efficient mechanism for establishing connectivity in wireless sensor and actor networks,’’ in IEEE Global Telecommunications Conference. IEEE, 2007, pp. 1129–1133. 29. K. Ozaki, K. Watanabe, S. Itaya, N. Hayashibara, T. Enokido, and M. Takizawa, ‘‘A fault-tolerant model of wireless sensor– actor network,’’ in Ninth IEEE International Symposium on Object and Component-Oriented Real-Time Distributed Computing. IEEE, 2006, pp. 8–pp. 30. A. Abbasi, M. Younis, and U. Baroudi, ‘‘Restoring connectivity in wireless sensor–actor networks with minimal node movement,’’ in 7th International Wireless Communications and Mobile Computing Conference. IEEE, 2011, pp. 2046–2051. 31. S. Zhang, X. Wu, and H. Wang, ‘‘Length-aware topology reconfiguration in wireless sensor–actor networks to recover from an actor failure,’’ in 33rd Chinese Control Conference. IEEE, 2014, pp. 304–309. 32. A. Abbasi, F. Younis, and U. Baroudi, ‘‘Recovering from a node failure in wireless sensor-actor networks with minimal topology changes,’’ IEEE Transactions on Vehicular Technology, vol. 62, no. 1, pp. 256–271, 2013. 33. N. Sabri, S. Aljunid, R. Ahmad, M. Malik, A. Yahya, R. Kamaruddin, and M. Salim, ‘‘Wireless sensor actor networks,’’ in 123 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. IEEE Symposium on Wireless Technology and Applications. IEEE, 2011, pp. 90–95. J. Barbar, M. Diaz, I. Esteve, D. Garrido, L. Llopis, B. Rubio, and J. Troya, ‘‘Tc-wsans: A tuple channel based coordination model for wireless sensor and actor networks,’’ in 12th IEEE Symposium on Computers and Communications. IEEE, 2007, pp. 173–178. A. Ian and H. Ismail, ‘‘Wireless sensor and actor networks: research challenges,’’ Ad Hoc Networks, vol. 2, no. 4, pp. 351–367, 2004. A. Boukerche and A. Martirosyan, ‘‘An efficient algorithm for preserving events’ temporal relationships in wireless sensor actor networks,’’ in 32nd IEEE Conference on Local Computer Networks. IEEE, 2007, pp. 771–780. A. Martirosyan and A. Boukerche, ‘‘Preserving temporal relationships of events for wireless sensor actor networks,’’ IEEE Transactions on Computers, vol. 61, no. 8, pp. 1203–1216, 2012. S. Habib, M. Safar, and N. ElSayed, ‘‘Automatic placement of actors within wireless sensor–actor networks,’’ in Telecommunication Networks and Applications Conference. IEEE, 2008, pp. 224–229. V. Rafe, H. Momeni, and M. Sharifi, ‘‘Energy-aware task allocation in wireless sensor actor networks,’’ in Second International Conference on Computer and Electrical Engineering, vol. 1. IEEE, 2009, pp. 145–148. L. Barolli, T. Yang, M. Ikeda, A. Durresi, and F. Xhafa, ‘‘A simulation system for routing efficiency in wireless sensor–actor networks: a case study for semi-automated architecture,’’ in Parallel and Distributed Systems, 2008. ICPADS’08. 14th IEEE International Conference on. IEEE, 2008, pp. 567–574. Z. Li and H. Shen, ‘‘A kautz-based real-time and energy-efficient wireless sensor and actuator network,’’ in 32nd International Conference on Distributed Computing Systems. IEEE, 2012, pp. 62–71. N. Trivedi, G. Elangovan, S. Iyengar, and N. Balakrishnan, ‘‘A message-efficient, distributed clustering algorithm for wireless sensor and actor networks,’’ in IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems. IEEE, 2006, pp. 53–58. M. Akba, M. Brust, and D. Turgut, ‘‘Sofrop: Self-organizing and fair routing protocol for wireless networks with mobile sensors and stationary actors,’’ Computer Communications, vol. 34, no. 18, pp. 2135–2146, 2011. S. Chinnappen-Rimer and G. Hancke, ‘‘Actor coordination in wireless sensor–actor networks,’’ in INDICON Conference. IEEE, 2009, pp. 1–4. H. Momeni, M. Sharifi, and S. Sedighian, ‘‘A new approach to task allocation in wireless sensor actor networks,’’ in First International Conference on Computational Intelligence, Communication Systems and Networks. IEEE, 2009, pp. 73–78. V. Narasimhan, A. Arvind, and K. Bever, ‘‘Greenhouse asset management using wireless sensor–actor networks,’’ in International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies. IEEE, 2007, pp. 9–14. A. Zamanifar, M. Sharifi, and S. Sedighian, ‘‘A distributed algorithm for restoring actor–actor connectivity in wireless sensor and actor networks,’’ in International Conference on Electronic Design. IEEE, 2008, pp. 1–6. H. Kim and J. A. Cobb, ‘‘Optimization trade-offs in the design of wireless sensor and actor networks,’’ in 37th IEEE Conference on Local Computer Networks, 2012, pp. 559–567. A. Zamanifar, M. Sharifi, and O. Kashefi, ‘‘A hybrid approach to actor–actor connectivity restoration in wireless sensor and actor networks,’’ in Eighth International Conference on Networks. IEEE, 2009, pp. 76–81. A. Alamuti, ‘‘Three protocols for actor selection in wireless sensor and actor networks,’’ in International Conference on Education and e-Learning Innovations. IEEE, 2012, pp. 1–3. Int J Wireless Inf Networks 51. V. Ranga, M. Dave, and V. Anil, ‘‘A distributed approach for selection of optimal actor nodes in wireless sensor and actor networks,’’ in International Conference on Contemporary Computing and Informatics. IEEE, 2014, pp. 312–319. 52. M. Akbas, E. M., and D. Turgut, ‘‘Localization for wireless sensor and actor networks with meandering mobility,’’ IEEE Transactions on Computers, vol. 64, no. 4, pp. 1015–1028, April 2015. 53. X. Li, X. Liang, R. Lu, S. He, J. Chen, and X. Shen, ‘‘Toward reliable actor services in wireless sensor and actor networks,’’ in 8th International Conference on Mobile Adhoc and Sensor Systems. IEEE, 2011, pp. 351–360. 54. M. Alaiwy, F. Alaiwy, and S. Habib, ‘‘Optimization of actors placement within wireless sensor–actor networks,’’ in 12th IEEE Symposium on Computers and Communications, 2007, pp. 179–184. 55. G. Shah, M. Bozyigit, and F. Hussain, ‘‘Cluster-based coordination and routing framework for wireless sensor and actor networks,’’ Wireless Communications and Mobile Computing, vol. 11, no. 8, 2011. 56. M. Imran, N. Haider, and M. Alnuem, ‘‘Efficient movement control actor relocation for honing connected coverage in wireless sensor and actor networks,’’ in 37th Conference on Local Computer Networks Workshops. IEEE, 2012, pp. 710–717. 57. M. Dong, K. Ota, S. Du, H. Zhu, and S. Guo, ‘‘Ants: Pushing the rapid event notification in wireless sensor and actor networks,’’ in International Joint Conference on Awareness Science and Technology and Ubi-Media Computing. IEEE, 2013, pp. 753–758. 58. N. Dinh and Y. Kim, ‘‘Directional anycast routing in wireless sensor and actor networks,’’ in International Symposium on Communications and Information Technologies. IEEE, 2012, pp. 251–255. 59. C. Tuan and Y. Wu, ‘‘Event ordering by double confirmation in wireless sensor and actor networks,’’ IEEE Sensors Journal, vol. 11, no. 3, pp. 829–836, 2011. 60. T. Melodia, D. Pompili, and I. Akyldiz, ‘‘Handling mobility in wireless sensor and actor networks,’’ IEEE Transactions on Mobile Computing, vol. 9, no. 2, pp. 160–173, 2010. 61. T. Issariyakul and H. E, ‘‘Introduction to network simulator 2 (ns2),’’ in Introduction to Network Simulator NS2. Springer US, 2009, pp. 1–18. 62. H. Kim and J. Cobb, ‘‘Optimal transmission range for multi-hop communication in wireless sensor and actor networks,’’ in 36th Conference on Local Computer Networks. IEEE, 2011, pp. 223–226. 63. P. Lameski, E. Zdravevski, A. Kulakov, and D. Davcev, ‘‘Architecture for wireless sensor and actor networks control and data acquisition,’’ in International Conference on Distributed Computing in Sensor Systems and Workshops. IEEE, 2011, pp. 1–3. 64. P. Bose, P. Morin, and J. Urrutia, ‘‘Routing with guaranteed delivery in ad hoc wireless networks,’’ Wireless Networks, vol. 7, no. 6, pp. 609–616, 2001. 65. W. Chen, J. Hou, and L. Sha, ‘‘Dynamic clustering for acoustic target tracking in wireless sensor networks,’’ IEEE Transactions on Mobile Computing, vol. 3, no. 3, pp. 258–271, 2004. 66. C. Eduardo, D. Manuel, L. Luis, and R. Bartolom, ‘‘Hero: A hierarchical, efficient and reliable routing protocol for wireless sensor and actor networks,’’ Computer Communications, vol. 35, no. 11, pp. 1392 – 1409, 2012. 67. Y. Haidong, M. Huadong, and L. Hongyu, ‘‘Coordination mechanism in wireless sensor and actor networks,’’ in First International Multi-Symposiums on Computer and Computational Sciences, vol. 2, 2006, pp. 627–634. 68. A. Boukerche, R. Araujo, and L. Villas, ‘‘A wireless actor and sensor networks qos-aware routing protocol for the emergency preparedness class of applications,’’ in 31st IEEE Conference on Local Computer Networks, 2006, pp. 832–839. 69. M. Tommaso, P. Dario, C. Vehbi, and I. F. Akyildiz, ‘‘A distributed coordination framework for wireless sensor and actor networks,’’ in Proceedings of the 6th ACM international conference, 2005, pp. 99–110. 70. H. Fei, C. Xiaojun, S. Kumar, and K. Sankar, ‘‘Trustworthiness in wireless sensor and actuator networks: towards low-complexity reliability and security,’’ in IEEE Global Telecommunications Conference, vol. 3, 2005. 71. L. Yen-Ting and S. Megerian, ‘‘Low cost distributed actuation in large-scale ad hoc sensor–actuator networks,’’ in International Conference on Wireless Networks, Communications and Mobile Computing, vol. 2, 2005, pp. 975–980. 72. D. ZhiCheng, W. Bingwen, L. Zhi, and A. Yin, ‘‘Vdspt: A sensor–actor coordination protocol for wireless sensor and actor network based on voronoi diagram and shortest path tree,’’ in International Symposium on Computer Network and Multimedia Technology, 2009, pp. 1–4. 73. B. McLaughlan and K. Akkaya, ‘‘Coverage-based clustering of wireless sensor and actor networks,’’ in IEEE International Conference on Pervasive Services, 2007, pp. 45–54. 74. A. Durresi, V. Paruchuri, and L. Barolli, ‘‘Delay-energy aware routing protocol for sensor and actor networks,’’ in 11th International Conference on Parallel and Distributed Systems, vol. 1, 2005, pp. 292–298. 75. S. Yahiaoui, M. Omar, A. Bouabdallah, and Y. Challal, ‘‘Multiactuators based anycast routing protocol for wireless sensor and actuator networks,’’ in International Conference on Advanced Networking Distributed Systems and Applications. IEEE, 2014, pp. 31–34. 76. D. Reina, S. Toral, P. Johnson, and F. Barrero, ‘‘An improvement of route duration in wsan based on nodes mobility and rss,’’ in 37th Annual Conference on Industrial Electronics Society. IEEE, 2011, pp. 2986–2991. 77. H. Wen, N. Bulusu, and J. Sanjay, ‘‘A communication paradigm for hybrid sensor/actuator networks,’’ in 15th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, vol. 1, 2004, pp. 201–205. 78. J. Haapola and N. Bui, ‘‘Towards dynamic application-dependent protocol stacks for wsans,’’ in Future Network and Mobile Summit. IEEE, 2010, pp. 1–8. 79. K. Jagadeesh, B. Majhi, and B. Ramesh, ‘‘A voronoi diagram based efficient coordination mechanism for wsan,’’ in First International Conference on Networks and Soft Computing. IEEE, 2014, pp. 226–230. 80. Z. Ngai, E.C.H Yangfan, M. Lyu, , and L. Jiangchuan, ‘‘Reliable reporting of delay-sensitive events in wireless sensor–actuator networks,’’ in IEEE International Conference on Mobile Adhoc and Sensor Systems, 2006, pp. 101–108. 81. A. Durresi and V. Paruchuri, ‘‘Geometric broadcast protocol for sensor and actor networks,’’ in 19th International Conference on Advanced Information Networking and Applications, vol. 1, 2005, pp. 343–348. 82. E. Cayirci, T. Coplu, and O. Emiroglu, ‘‘Power aware many to many routing in wireless sensor and actuator networks,’’ in Proceeedings of the Second European Workshop on Wireless Sensor Networks, 2005, pp. 236–245. 83. Z. Yangfan, L. M. Ngai, E.C.H, and L. Jiangchuan, ‘‘Powerspeed: A power-controlled real-time data transport protocol for wireless sensor–actuator networks,’’ in IEEE Conference on Wireless Communications and Networking, 2007, pp. 3736– 3740. 84. F. T, ‘‘Scalable routing in sensor actuator networks with churn,’’ in 3rd Annual IEEE Communications Society on Sensor and Ad Hoc Communications and Networks, vol. 1, 2006, pp. 30–39. 123 Int J Wireless Inf Networks 85. L. SuetFei, ‘‘Wireless sensor actuator network for light monitoring and control application,’’ in 3rd IEEE Conference on Consumer Communications and Networking, vol. 2, 2006, pp. 974–978. Jagadeesh Kakarla is a Ph.D student, Department of Computer Science, NIT Rourkela, Rourkela, India. He has obtained his M.Tech in the field of Computer Science, Pondicherry University, India. He has obtained his B.Tech in the field of Information Technology, Jawaharlal Nehru Technological University, India. His research areas include Wireless Sensor Networks, Adhoc Networks and Image Processing. Banshidhar Majhi is working as a professor in Computer science department. NIT Rourkela, India. Has 23 years of teaching and 3 ears of industry experience. Has published 50 journal articles in referred journals and 100 articles in reputed international conferences. Research interests include image processing, computer vision, security protocols and wireless sensor networks. 123 View publication stats Ramesh Babu Battula is working as an assistant professor and pursuing Ph.D in Department of Computer Science, MNIT Jaipur, India. He has obtained his M.Tech in the field of Computer Science, IIT Guwahati, India. He has obtained his B.Tech in the field of Information Technology, Nagarjuna University, India. His research areas include Network Security, Computer Networks, Information Security, Android Security, Wireless Mesh Networks, and Wireless Sensor Networks.