Received: 22 October 2019 Revised: 13 February 2020 Accepted: 15 February 2020 DOI: 10.1002/cpe.5720 RESEARCH ARTICLE Multihop routing protocols on wireless ad hoc sensor networks with modified MAC layer and broadcasting scheme Tung T Nguyen International School, Vietnam National University, Hanoi, Vietnam Summary Most sensor devices are equipped with local batteries, placing a limitation on available Correspondence Tung T Nguyen Research Center for Information System and Computer Engineering, Vietnam National University, Hanoi, Vietnam Email: tungnt@isvnu.vn energy, assuming that replacement of batteries is not feasible This constraint limits the operation time of sensor networks In many scenarios, sensors are operating in remote or dangerous areas, and it is impractical to replace or recharge the battery of the sensors In some scenarios, such as machine monitoring or medical monitoring, it is more convenient to install new sensors than to replace the battery of the sensor nodes The focus of this article is to design routing methods to achieve the maximum operation time of sensor networks (the number of rounds to transmit data) under the constraint of battery sources Therefore, the main contribution of this research is not to change any characteristic of the network infrastructure, but to define the interactions between nodes to achieve better energy efficiency KEYWORDS battery sources, computing, multihop routing, operation time, power transmission, sensor networks INTRODUCTION Some applications, for example in military, require sensors to be small in size and have short transmission ranges to reduce the chances of being detected These requirements cause further constraints in CPU speed, memory, and battery lifetime.1 As the lifetime of a sensor node is closely related to its battery lifetime, reducing the energy usage of a sensor by two often results in a reduction of sensor installation cost by 50% This means that all aspects related to the energy usage must be designed very efficiently In this article, the energy efficiency of wireless ad hoc sensor networks (WASNs), in which traffic can be transmitted from any source to any destination and there is no base station, is examined Section first studies the energy dissipation model inside a sensor Section proves the complexity theory of the selection of energy efficient paths and then proposes heuristic algorithms for the problem Also, a broadcasting scheme to eliminate the overhearing energy of the neighboring nodes is analyzed Finally, simulation results are presented In the results, three heuristic routing methods are studied: shortest path (SP), SP including neighboring nodes (SP_N), and SP of remaining energy (SP_RE) As the energy dissipation by neighboring nodes is quite significant, a prebroadcast method is used to eliminate the energy dissipation The performance of SP and SP_RE with the prebroadcast scheme is then examined and compared to that of SP and SP_RE without the broadcast scheme This is shown in Figure ENERGY DISSIPATION SOURCES INSIDE A SENSOR Sensor nodes dissipate energy in different ways when they are in different modes: active, idle, or sleep mode.2,3 When a node is in sleep mode, it turns off its radio, and the node cannot hear any action of other nodes in the network The energy dissipation in this mode is very low which is often known as background energy (Psleep ) This mode also includes some small energy dissipation for sensing data or an event In idle mode, the node can Concurrency Computat Pract Exper 2020;e5720 https://doi.org/10.1002/cpe.5720 wileyonlinelibrary.com/journal/cpe © 2020 John Wiley & Sons, Ltd of NGUYEN of F I G U R E Several heuristic multihop routing methods with and without a broadcast scheme hear action from other nodes and is ready to transmit data or receive data when there is a request The idle energy consumption Pidle is often much higher compared to Psleep and can be almost equal to the energy dissipation while sensors are receiving data.3,4 A sensor node is on active mode if it is transmitting, receiving, or computing data When a sensor is in sleep mode, it cannot participate in any activity in sensor networks unless it changes to the idle mode or the active mode Psleep is very low and can be ignored in the total energy consumption of a sensor device For examples, the authors in Reference showed the energy consumption of their radio transceiver design, which is 13.5 mW, 24.75 mW, and 15 μW in receiving mode, transmitting mode, and sleep mode, respectively The total energy dissipation of a sensor during its operation is3 : Ptotal = tidle × Pidle + treceive × Preceive + ttransmit × Ptransmit + tcompute × Pcompute (1) where Preceive , Ptransmit , or Pcompute are the energy to transmit, receive, or compute data, respectively The radio transmission for wireless sensor networks (WSNs) is similar to the radio propagation model for wireless networks In addition, in WSNs, the energy dissipation is assumed to increase linearly with the size of data messages Most of previous work on energy efficiency in WSNs uses the energy model given below.2,5-9 The total transmission energy of a message is calculated by2,5 : Et = 𝑘𝐸 elec + 𝜀amp 𝑘𝑑 n (2) The reception energy is calculated by: Er = 𝑘𝐸 elec where Eelec is the energy dissipation of the electronic circuitry to encode or decode a bit, k is the message size, 𝜀amp is the amplifier constant, and d is the distance between the transmitter and the receiver Therefore, the energy consumption is proportional to the number of bits transmitted and received Similarly to the general radio propagation model in wireless networks, the power factor n can be determined depending on the radio propagation of the signal The amplifier energy dissipation to amplify one bit can be rewritten as2,5 : Eamp (d) = {𝜀FS d2 , d < Eamp (d) = {𝜀MP d4 , d > (3) where 𝜀FS is the free space amplifier constant, 𝜀MP is the multipath amplifier constant and is the threshold transmission distance between the two modes.2,5 √ = 𝜀FS 𝜀MP (4) NGUYEN of F I G U R E Transmission from a source to a destination drains the energy of the source and the destination and neighboring nodes, Problem (6) asks for the solution to maximize the minimum remaining energy of all sensor nodes COMPLEXITY THEORY OF THE SELECTION OF ENERGY EFFICIENT PATHS The energy dissipation model used in the previous section is repeated for this section The total transmission energy of a message is calculated by: Et = 𝑘𝐸 elec + 𝜀amp 𝑘𝑑 n (5) The reception energy is calculated by: Er = 𝑘𝐸 elec where Eelec is the energy dissipation of the electronic circuitry to encode or decode a bit, k is message size, 𝜀amp is the amplifier constant, and d is the distance between the transmitter and the receiver 3.1 Original routing problem Given a network of n sensors, in which any sensor node can connect to all other sensor nodes by adjusting its transmission power using Equation (5) Each sensor node i has the energy storage of e(i) A random source node s wants to transmit data to a destination node d Obviously, there are many possible paths from s to d Each path results in an energy reduction of all nodes on the path (including the nodes are within the transmission range of the data transmission) The routing problem is to find a path from s to d so that after the data transmission, the minimum RE storage of all sensor nodes is maximized: Maximize ∶ min(e(i)), ∀i ∈ n (6) where e(i) is the RE of node i after the path is established Figure shows that the data transmission from the source to the destination will drain the energy of seven other neighboring nodes If any node of the seven nodes has low remaining battery energy, the path is not a good choice Problem (6) seeks a better path that excludes low RE nodes Unfortunately, Problem (6) is NP-complete This problem can be polynomially reduced to the Path with Forbidden Pairs problem This is a well-known NP-complete problem Details are given in Reference 10 Therefore, there is no polynomial time algorithm to find the energy efficient path There have been numerous studies on the energy efficiency of multihop routing in literature These studies use the Dijkstra algorithm or variants of this algorithm to calculate the shortest routes to a destination with different types of energy metrics Unfortunately, only few studies mentioned about the remaining battery of sensor nodes, and it is difficult to apply the Dijkstra algorithm or its variants to the lifetime problem of WASNs In References 11 and 12, a minimum total power routing was proposed In this protocol, the route with the minimum total power consumption is selected from a set S containing all possible paths NGUYEN of Minimum battery cost routing11 was another way to approach In this protocol, the inverse of battery energy of nodes is used as the metric The path metric is the sum of the link metrics The path with the smallest metric is selected from the possible routes We propose three heuristic solutions for the problem Details are given below Shortest path of the energy dissipation (SP): Given a source node s and destination node d, find a simple path Π from s to d that minimizes the total energy dissipation by all nodes on the path ∑ (pi + r)is minimized (7) i∈Π−d where r is the reception energy consumption of any sensor node (End of algorithm) Shortest path of the energy dissipation including neighboring nodes (SP_N): Given a source node s and a destination node d, find a simple path Π from s to d that minimizes the total energy dissipation by all nodes participating in the data transmission ∑ ( ∑ pi + i∈Π−d ) rj ) is minimized (8) j∈N(i) where N(i) is the set of neighboring nodes in the transmission range of node i (End of algorithm) Shortest path of the remaining energy (SP_RE): Let us define a weight for a link on any path as: W(i) = ∑ e(j) j∈N(i) (9) where N(i) is the set of neighboring nodes in the transmission range of node i Given a source node s and a destination node d, find a simple path Π from s to d that minimizes the total weight by all links participating in the data transmission ∑ W(i) is minimized (10) i∈Π−d (End of algorithm) ENERGY DISSIPATION FOR BROADCASTING In cluster-based routing, there is a base station to support the formation of clusters at the beginning of each round of data transmission Therefore, sensor nodes can reduce or eliminate the overhead energy at each round by letting the base station the initialization job In multihop routing, there is no base station and sensor nodes have to dissipate some overhead energy to exchange information.13,14 As sensor nodes are usually fixed, it is not necessary to exchange information among the sensor nodes during the data transmission However, as sensor nodes usually experience faults, the topology of the networks can change frequently due to these updates A robust routing method requires sensors to report their status periodically in their networks Therefore, in the operation of the above routing methods, the status of all sensors is transmitted to a source node before each round of data transmission so that the node can calculate the path for its data transmission The transmission of the status information requires two overhead energies: 1) A source node needs to broadcast an ID message (16 bits) to every node in the network 2) All nodes send their energy message (32 bits) to the source node 4.1 Simulation results A number of simulators in Microsoft Visual Studio is developed to simulate the performance of SP, SP_N, and SP_RE The energy dissipation model used in the previous section is repeated for this section The total transmission energy of a message is calculated by: Et = 𝑘𝐸 elec + 𝜀amp 𝑘𝑑 n The reception energy is calculated by: Er = 𝑘𝐸 elec (11) NGUYEN of Network size (200 m × 200 m) Base station (50 m, 275 m) Number of sensor nodes 100 nodes Energy message: 20 bits Position of sensor nodes: Uniform placed in the area Energy model: Eelec =50 × 10−9 J, 𝜀fs =10 × 10−12 J/bit/m2 and 𝜀mp =0.0013 × 10−12 J/bit/m4 Broadcast ID message: 16 bits Broadcast energy message: 32 bits FIGURE networks Number of rounds over 100 random 100-node Lifetime versus network topologies 700 Rounds 600 SP 500 SP_N 400 SP_RE 300 200 20 40 60 80 100 Network topology Energy per round versus network topologies Energy per round (m J) F I G U R E Average energy dissipation per round (mJ) over 100 random 100-node networks 14 12 SP 10 SP_N SP_RE 20 40 60 80 100 Network topology where Eelec is the energy dissipation of the electronic circuitry to encode or decode a bit, k is message size, 𝜀amp is the amplifier constant and d is the distance between the transmitter and the receiver The network settings for the simulations in the section are given below In the first set of simulations, the lifetime performance of the above routing methods is studied for the above 100 random 100-node sensor networks Each node begins with 50 mJ of energy The operation of each sensor network is divided into rounds In each round, a random source node transmits a unit of data to a random destination node The process is repeated until the first sensor node dies and the lifetime for each routing method in each network topology is recorded On average, SP, SP_N, and SP_RE perform 268, 363, and 519 rounds, respectively These lifetimes are the time until the first sensor fails This is shown in Figure It is also of interest to evaluate the total energy consumption of the routing methods Figure shows the performance over the 100 topologies On average, SP, SP_N, and SP_RE dissipate 11.7, 6.7, and 7.8 mJ per round, respectively As expected, SP_N provides the minimum energy dissipation per round among the three routing methods This is because SP_N selects a route to minimize the total energy dissipation of all sensor nodes in the path including neighboring nodes Although SP_RE provides the best lifetime, it spends more energy per round than SP_N This is because SP_RE needs to preserve the residual energy of all sensor nodes so it does not always select the minimum energy path MODIFIED MAC LAYER When a source node transmits data to a destination, neighbors that are inside the transmission range of the source node also receive the data As the reception energy at each neighbor is the same to that of the destination and is more than half of the transmission energy, the total energy NGUYEN of Data Ad hoc Networks t s Wakeup Channels Ad hoc Networks t Data Sensor Networks e t Sensor Networks Channels FIGURE Wakeup Wakeup Sleep Wakeup Sleep t Sleep Data communication in ad hoc and sensor networks consumption becomes much higher than the actual energy needed for the data transmission Therefore, in order to achieve a better lifetime for the sensor networks, it is desirable to reduce the unnecessary energy spent at the neighbors.15,16 Also, data transmission in sensor networks requires much lower bandwidth (on the order of 1-100 kb/s) than that of mobile ad hoc networks (on the order of 1-100 Mbps) As can be seen from Figure 5, in sensor networks, nodes send data periodically and the interval between two subsequent data transmission in the networks is usually very long, while in mobile ad hoc networks, data are transmitted continuously and randomly so mobile device has to wake up to listen to the channel Therefore, unlike mobile ad hoc networks, the medium access control design for sensor networks can use time-division multiple access (TDMA)-based protocols that conserve more energy than contention-based protocols like carrier sense multiple access (eg, IEEE 802.11).10,13 TDMA protocol allows sensors to turn off their radio when not necessary As a result, a prebroadcast scheme is proposed to reduce the energy dissipation of neighboring nodes In more detail, when a source node wants to send data to a destination, it will calculate a routing path using any routing method presented in the previous section The source node then sets its transmission power to reach the destination node and broadcasts a message that contains the list of ID of nodes that are involved in the data transmission Any node in the broadcast area will receive the message If a node's ID matches the ID of any node in the path, it will turn on its radio during the data transmission Otherwise, the node turns off its radio until the end of this data transmission All nodes wake up at the beginning of the next round for new data transmission This is shown in Figure The prebroadcast scheme requires two overhead energies: 1) A source node broadcasts a list of ID of nodes on the path to all sensors in the broadcast area 2) All sensor nodes in the area receive the message from the source node However, as the actual data message size is bigger than the broadcast message, it is expected that the prebroadcast scheme will reduce the total energy consumption significantly In the next set of simulations, SP method was run with the prebroadcast message of 100 bits (SP_100) and 500 bits (SP_500), respectively The lifetime (the number of rounds) for each case is recorded Figure shows that new SP method performs much better than the original SP method On average, SP with the broadcast message of 100 bits achieves the lifetime of about two times longer than the original SP This is because the new SP method only spends energy of 60% of the original SP method as shown in Figure The same simulations were repeated for SP_RE with the prebroadcast message of 100 bits (SP_RE_100) and 500 bits (SP_RE_500) Unlike the new SP method, the new SP_RE method only performs slightly better than the original SP_RE This is shown in Figure Figure 10 also shows that there is not significant improvement of the energy dissipation per round by the new SP_RE method This is because the original SP_RE already avoids the energy drain of neighboring nodes in the data transmission, so the broadcast process does not significantly help We not consider SP_N method as without the reception energy at neighboring nodes, SP_N is the same as SP method NGUYEN of F I G U R E Transmission from a source to a destination tells neighboring nodes to turn off their radio during the transmission Rounds Lifetime versus network topologies FIGURE networks 800 700 600 500 400 300 200 100 SP SP_100 SP_500 Number of rounds over 100 random 100-node 20 40 60 80 100 Network topology Energy per round (mJ) Energy per round versus network topologies 14 12 10 F I G U R E Average energy dissipation per round (mJ) over 100 random 100-node networks SP_100 SP_500 SP 20 40 60 80 100 Network topology Lifetime versus network topologies 900 800 Rounds 700 SP_RE 600 SP_RE_100 SP_RE_500 500 400 300 FIGURE networks Number of rounds over 100 random 100-node 20 40 60 Network topology 80 100 NGUYEN of F I G U R E 10 Average energy dissipation per round (mJ) over 100 random 100-node networks Energy per round versus network topologies Energy per round (m J) 8.5 7.5 SP_RE_100 6.5 SP_RE_500 SP_RE 5.5 4.5 20 40 60 Network topology 80 100 CONCLUSIONS This article focused on multihop routing protocols on wireless ad hoc sensor networks In multihop routing, data are transmitted from a random source node to a random destination node Wireless transmission media in sensor networks is different to the transmission of wire-line networks in that transmission from a source node to a destination node can cause neighboring nodes to dissipate energy when hearing the transmission This energy dissipation degrades the lifetime of the sensor networks and makes the energy optimization get more complicated It is shown that the problem of locating a simple path that maximizes the minimum RE of all sensor nodes is NP-complete Therefore, there is no polynomial time algorithm for the problem and heuristic solutions are required to achieve reasonable energy efficiency By reducing the unnecessary energy dissipation of the neighbors, the lifetime is increased significantly When the broadcast scheme is implemented, SP_RE only slightly increases system lifetime of the original SP_RE (1.1 to 1.3 times), while new SP method can extend the lifetime 1.8 to 2.1 times longer than that of the original SP Although the new broadcast scheme improves the lifetime of SP method significantly, the SP_RE method is still the best routing method for the lifetime until a specified proportion of nodes fail Therefore, it is recommended to use SP_RE to prolong the network lifetime of WASNs This is because SP_RE balances the RE of all sensor nodes so all sensor nodes run out of energy at the same time ORCID Tung T Nguyen https://orcid.org/0000-0003-1695-8902 REFERENCES Al-Karaki JN, Kamal AE Routing techniques in wireless sensor networks: a survey IEEE Wirel Commun 2004;11(6):6-28 Heinzelman WB Application-specific protocol architectures for wireless networks [PhD dissertation] Massachusetts Institute of Technology; June 2000 Ye W, Heidemann J, Estrin D Medium access control with coordinated adaptive sleeping for wireless sensor networks IEEE/ACM Trans Netw 2004;12(3):493-506 Feeney L, Nilsson M Investigating the energy consumption of a wireless network interface in an ad hoc networking environment IEEE INFOCOM 2001;3:1548-1557 Heinzelman WB, Chandrakasan AP, Balakrishnan H Energy-efficient communication protocol for wireless microsensor networks Paper presented at: 33rd Hawaii International Conference Systems Sciences; January 2000 Heinzelman WB, Chandrakasan AP, Balakrishnan H An application-specific protocol architecture for wireless microsensor networks IEEE Trans Wirel Commun 2002;1(4):660-670 Zhang J, Huang B, Tu L, Zhang F A cluster-based energy-efficient scheme for sensor networks Paper presented at: Proceedings of the Sixth International Conference on Parallel and Distributed Computing, Applications and Technologies (PDCAT'05); 2005 Chang R, Kuo CJ An energy efficient routing mechanism for wireless sensor networks Paper presented at: Proceedings of the 20th International Conference on Advanced Information Networking and Applications (AINA'06); 2006 Voigt T, Dunkels A, Alonso J, Ritter H, Schiller J Solar-aware clustering in wireless sensor networks Paper presented at: The Ninth International Symposium on Computer and Communications (ISCC2004); 2004 10 Tung NT Energy-efficient routing algorithms in wireless sensor networks [PhD thesis] Australia: Monash University; 2009 11 Toh CK Maximum battery life routing to support ubiquitous mobile computing in wireless ad hoc networks IEEE Commun Mag 2001;39(6):138-147 12 Safwat A, Hassanein H, Mouftah H Energy-aware routing in MANETs: analysis and enhancements Paper presented at: Proceedings of the 5th ACM International Workshop on Modeling Analysis and Simulation of Wireless and Mobile Systems; 2002:46-53 13 Nguyen TT, Binh HTT Base station location—aware optimization model of the lifetime of wireless sensor networks Mobile Netw Appl 2016;21:10-17 https://doi.org/10.1007/s11036-015-0614-3 NGUYEN of 14 Nguyen TT, Nguyen VD Optimizing the operating time of wireless sensor network EURASIP J Wirel Commun Netw 2012;348 https://doi.org/10.1186/ 1687-1499-2012-348 15 Lourthu Hepziba Mercy M, Balamurugan K, Vijayaraj M Maximization of lifetime and reducing power consumption in wireless sensor network using protocol Int J Soft Comput Eng 2013;2(6):89-95 16 Paschalidis IC, Wu R Robust maximum lifetime routing and energy allocation in wireless sensor networks Int J Distrib Sens Netw 2012;2012, Article ID 523787:14 How to cite this article: Nguyen TT Multihop routing protocols on wireless ad hoc sensor networks with modified MAC layer and broadcasting scheme Concurrency Computat Pract Exper 2020;e5720 https://doi.org/10.1002/cpe.5720 ... 100 CONCLUSIONS This article focused on multihop routing protocols on wireless ad hoc sensor networks In multihop routing, data are transmitted from a random source node to a random destination... allocation in wireless sensor networks Int J Distrib Sens Netw 2012;2012, Article ID 523787:14 How to cite this article: Nguyen TT Multihop routing protocols on wireless ad hoc sensor networks with modified. .. mobile ad hoc networks, the medium access control design for sensor networks can use time-division multiple access (TDMA)-based protocols that conserve more energy than contention-based protocols