A Novel Energy Replenishment Algorithm to Increase the Network Performance of Rechargeable Wireless Sensor Networks Authors: Tariq, Vishwanath Eswarakrishnan, Adil Hussain, Zhu Wei, Muhammad Uzair Published: 2024 ABSTRACT AbstractThe emerging wireless energy transfer technology enables sensor nodes to maintain perpetual operation. However, maximizing the network performance while preserving short charging delay is a great challenge. In this work, a Wireless Mobile Charger (MC) and a directional charger (DC) were deployed to transmit wireless energy to the sensor node to improve the network’s throughput. To the best of our knowledge, this is the first work to optimize the data sensing rate and charging delay by the joint scheduling of an MC and a DC. We proved we could transmit maximum energy to each sensor node to obtain our optimization objective. In our proposed work, a DC selected a total horizon of 360° and then selected the horizon of each specific 90∘ area based on its antenna orientation. The DC’s orientation was scheduled for each time slot. Furthermore, multiple MCs were used to transmit energy for sensor nodes that could not be covered by the DC. We divided the rechargeable wireless sensor network into several zones via a Voronoi diagram. We deployed a static DC and one MC charging location in each zone to provide wireless charging service jointly. We obtained the optimal charging locations of the MCs in each zone by solving Mix Integral Programming for energy transmission. The optimization objective of our proposed research was to sense maximum data from each sensor node with the help of maximum energy. The lifetime of each sensor network could increase, and the end delay could be maximized, with joint energy transmission. Extensive simulation results demonstrated that our RWSNs were designed to significantly improve network lifetime over the baseline method. 1. Introduction There are many applications for wireless sensor networks (WSNs), such as environmental detection, civil infrastructure, target tracking, information security, and intelligent media care [1,2]. In WSNs, energy replenishment is becoming the bottleneck to improving network utility. Currently, most sensor nodes are powered by batteries. However, the high manufacturing and swapping costs of batteries hinder WSNs from fully realizing their potential. Wireless charging is one of the most promising technologies for tackling the energy limitation problem of sensor nodes. A sensor node’s lifespan can be extended without battery replacement with the use of a wireless charger. The sensor node’s capabilities include sensing, computation, data gathering, and data aggregation [3,4]. The sensor node transmits data to a sink node in real time and receives control instructions from a cloud server using a wireless connection module. As sensor nodes analyze and send substantial information, the energy of the RWSN progressively diminishes over time [5]. When the battery energy gets low and cannot be restored [6], this results in the breakdown of the sensor nodes. Consequently, the energy consumption of sensor nodes has emerged as a critical concern and is considered a primary determinant in the large-scale implementation of RWSNs. To enhance the longevity of RWSNs, scheduling algorithms for sensor nodes are presented to optimize energy utilization. Most rechargeable batteries are nonremovable, resulting in sensor nodes being inoperative because of battery energy depletion. Recent studies [7,8] have proposed using mobile chargers to transmit wireless charging power to sensor nodes [9]. The weakness of previous studies is that mobile chargers cannot transmit proper energy because they have limited battery capacity and select limited routing paths to transmit charging power to the area of the RWSN. Different sensor nodes cannot maintain their working power because of the limited energy. In our proposed research, a mobile charger and a directional charger were used jointly to transmit wireless charging power to each sensor node to overcome these challenges. To the best of our knowledge, we transmitted maximum energy to sense maximum data from the maximum number of sensor nodes. The predetermined recharging scheduling of mobile chargers provided a near-optimal charging path because of the chargers’ dynamic energy consumption. In RWSNs, a directional charger transmits energy to sensor nodes from different orientations. Each directional charger selects its total 180° area, and each orientation covers a 90° area to fulfill the energy demand of each sensor node. Directional chargers cover the network area in two directions; each orientation covers 90 areas. Mobile chargers cover the nodes that are out of the coverage area of the directional chargers. The critical challenge of our designed research was coupling directional chargers and mobile chargers to fulfill the energy demand without delay. In our proposed research, the Mix Integral Programming method was used to find the optimal stopping point of mobile chargers. After selecting the optimal stopping point, the mobile and directional chargers jointly started energy transmission. At each time slot, the mobile and directional chargers selected their numbers of sensor nodes for energy transmission. We had to transmit maximum energy at each time slot to overcome the joint energy replenishment challenge. In the RWSNs in our proposed research, a time-slot-based algorithm was used for energy transmission to each node. The joint charging process was used to transmit energy to each set of networks. We used a priority-based charging method. In this optimal algorithm, when a node has less residual energy, the priority of energy supplement is higher, and that node should be replenished with higher charging priority. After that, we selected the tour length of the mobile charger for maximum energy transmission. Some nodes far from the mobile charger’s tour could not replenish energy properly. These sensor nodes would be charged in the next tour. In this algorithm, we needed to bind the length of the mobile charger tour to transmit maximum energy within each time slot. Moreover, we used an optimal joint-based charging algorithm to charge sensor nodes jointly with the help of a mobile charger and a directional charger. This research aimed to optimize the data transmission rate while preserving charging delay. DCs alone cannot satisfy all the charging demands of each node in an RWSN because of a fair charging delay. To tackle this issue, we needed to schedule the MC and DCs jointly to fulfill the charging demands of each node. The MC started the route from the base station, and after transmitting energy to different sensor nodes, it returned to the base station for energy replenishment. The main contributions of this paper can be summarized as follows:To the best of our knowledge, this is the first work to study the joint scheduling problem of DCs and MCs for energy replenishment. We noticed that the throughput of WRSNs can be further improved by jointly working on the performance of the MC and DCs.We designed a near-optimal joint energy replenishment algorithm and proposed a central charging point selection algorithm for proper energy transmission.Extensive simulations were conducted to indicate the effectiveness and advantages of our proposed algorithms. To the best of our knowledge, this is the first work to study the joint scheduling problem of DCs and MCs for energy replenishment. We noticed that the throughput of WRSNs can be further improved by jointly working on the performance of the MC and DCs. To the best of our knowledge, this is the first work to study the joint scheduling problem of DCs and MCs for energy replenishment. We noticed that the throughput of WRSNs can be further improved by jointly working on the performance of the MC and DCs. We designed a near-optimal joint energy replenishment algorithm and proposed a central charging point selection algorithm for proper energy transmission. We designed a near-optimal joint energy replenishment algorithm and proposed a central charging point selection algorithm for proper energy transmission. Extensive simulations were conducted to indicate the effectiveness and advantages of our proposed algorithms. Extensive simulations were conducted to indicate the effectiveness and advantages of our proposed algorithms. The organization of this paper is as follows: Section 2 describes the related work. Section 3 contains the methodology. Section 4 presents the network model. Section 5 presents the charging model, and Section 6 presents the problem formulation. Section 7 presents the Mobile Charging Energy Transmission Protocol. Section 8 presents the evaluation and simulation results of the proposed scheme, the latter of which are compared with those of other schemes. Section 9 shows the conclusion of this paper with a summary of new findings. 2. Related Work Different existing research has focused on energy harvesting techniques that extract energy from solar, vibration, and wind to replenish energy in sensor nodes. Wireless energy transfer technology provides a new way of supplying high-density, stable, and sustainable energy for sensor nodes. Recently, Powercast has developed products that transmit energy to the sensor nodes [10]. The authors of [11] used Powercast chargers to extend the network lifetime in constructing a charging queue based on the greedy algorithm. 2.1. Mobile Charging Scheduling For replenishing energy to each sensor node through strong magnetic resonance, there has been a surge of interest in using the MC traveling algorithms in WRSNs [12]. Most of the reported approaches employed optimization algorithms to solve wireless charging problems. A single mobile charger can charge sufficiently a small-scale network of RWSNs. Due to the limitation of the battery capacity of the sensor nodes, multiple mobile chargers are required for large-scale areas of a network. Energy replenishment provides the stability and reliability of energy supply for sensor nodes [13]. The mobile charging energy transmission process can be evaluated from the sensor node’s and charger’s perspectives [13]. The most intuitive way is to charge sensor nodes periodically along an optimal TSP tour in RWSNs [8]. The researchers employed a mobile charger to charge sensor nodes, where the cumulative traveling time and charging time are expected to be minimized. The best charging scheme was reported [13], which jointly considers routing and charging. In a dynamic environment, the objective is to maximize network lifetime under practical constraints. In [14], the authors proposed multiple energy-constrained mobile charging schemes to charge RWSNs collaboratively. The aim was to maximize the energy efficiency of sensor nodes. In [8,15], the authors focused on a charging strategy that employed a mobile charger equipped with a high-capacity battery to replenish energy periodically in sensor nodes. In [16], the issue pertains to 2-D wireless sensor networks and the small quantity of energy-constrained mobile chargers. The objective of this endeavor is to ensure the network’s performance. The average performance is far below that of an ideal charger. In [17], the issue of reducing the quantity of mobile chargers to guarantee uninterrupted energy transfer for each sensor node was examined. The varying charging cycles of MC yield diverse outcomes for the charging tour. In [18], numerous mobile chargers were analyzed to minimize the overall trip cost of the mobile charger, ensuring that each node receives energy replenishment from the sensor node. In [19] an on-demand charging method was modeled for scheduling multi-copter vehicles assigned to charge sensor nodes. The aim was to reduce the number of mobile chargers. In [20], the scheduling of multiple MCs for incoming charging requests assessed the nearest-first and recent-rarest-first techniques. The closest scheduling technique was superior for the localization of the charge request process, as demonstrated through simulation. In [18], the node’s energy level cannot decrease from its maintaining level. Multiple mobile chargers reduced the overall travel expenses of various mobile charging units. In [19], MCs returned to their designated depots for energy replenishment. The network model addressed on-demand charging issues for sensor nodes in WRSNs according to the schedule of a mobile charger. The aim was to reduce the number of mobile charger routes to optimize energy transfer. 2.2. Directional Charging Scheduling Some works study the directional charger problem, but their solutions have not been applied in our research. In [21], the authors proposed a method to detect omnidirectional charging capability for a given topology of directional chargers. They also studied how to place directional chargers to maximize charging utility in [22] to resolve the wireless charger deployment optimization problem. In [23], chargers with directional antennas were deployed on specific grid points and designed the greedy and adaptive cone covering algorithms. To maximize the survival rate of end devices, an omnidirectional wireless charger with partial coverage has been used to maximize the survival rate of terminal devices [24]. One of the important aspects of a directional antenna is that it transmits maximum energy in different directions to cover different sensor nodes in one sector. However, these models are so complicated that they must be tractable in the performance analysis. Using directional antennas instead of omnidirectional antennas in wireless ad hoc networks or RWSNs can significantly improve the network performance since directional antennas can concentrate the transmitting/receiving capability to desired directions [25]. Figure 1 shows the WRSN. For example, some recent studies have shown that using directional antennas in WANs can improve the network capacity and reduce the end-to-end delay [26,27,28,29]. In addition, using directional antennas in WSNs can improve security, as shown in [25,30]. In WSNs with directional antennas, it is difficult for each node to obtain the location knowledge of other neighbors due to directional beam forming [31]. To solve the problem of directional neighbor discovery, complicated schemes, such as using direction of arrival estimation, and swiveling the beam from 0 to 2π, were proposed in [32]. To the best of our knowledge, this is the first work to study the mobile charger and directional charger problem in RWSNs. The existing works have contributed to optimizing the power supply through the mobile charger and directional charger. However, coupling issue in mobile chargers and directional chargers makes the rechargeable wireless sensor network charging problem quite challenging. Previous works did not consider the joint charging method using a mobile charger and directional charger jointly to transmit maximum energy and minimum charging delay. The novelty of our proposed research is that there are no data loss in our proposed work because of on-demand energy transmission to each sensor node. 3. Methodology The critical challenge of our design is the coupling of DC and MC used to fulfill the energy demand without delay. In RWSNs, each DC selects its total 360∘ area, with each 90∘ specific orientation to fulfill the energy demand of each node in every time slot. MC selects a network area that is out of the coverage area of DC. The charging process with the MC antenna is high gain, which is used for energy transmission in each set of networks. MC selects the optimal stopping point in each set of networks to charge all the sensor nodes at different time slots. After selecting a specific network area, MC and DC jointly start energy transmission to each sensor node. At each time slot, MC and DC select a different number of sensor nodes for energy transmission. MC travels from the base station to each charging point, which is obtained by solving Mix Integral Programming. When the energy level of MC is low, it returns to the base station for energy replenishment. We need to choose different MC stopping points from each set of sensor nodes from each set of networks. After selecting a specific area by MC and DC, both chargers jointly start energy transmission to the sensor node. At different times, slot MC turns its antenna orientation toward the area of DC to transmit maximum energy. To overcome the joint energy replenishment challenge, we need to transmit maximum energy to each sensor node at each time slot to collect maximum data from each node. It is quite challenging to utilize the limited energy of DC and MC to charge the lowest energy demand sensor node and maintain the network performance continuously. MC selects each set’s route from the base station for energy transmission. MC finds the optimal stop location in each set of sensor nodes with the help of Mix Integral Programming. In RWSNs, after finding the optimal location, MC starts to replenish charging power to each node that is out of the DC coverage area and in the DC coverage area. Voronoi diagram has been used to divide the network area into different sets. In RWSNs, we must allocate a time slot-based algorithm for energy transmission to each node. The joint charging process is usually used to transmit energy in each set of networks. For proper energy transmission to each node, we select the tour length of MC. We bound the length of the MC tour to maintain network operation. The proposed work’s objective is to increase each node’s sensing rate. In our proposed research, a simulation is built in the widely used MATLAB. The experiment results show that our proposed algorithm works efficiently to obtain maximum throughput. The main contribution of this paper can be summarized as follows: To the best of our knowledge, this is the first work to study the joint scheduling problem of DC and MC for energy replenishment. We noticed that the throughput of WRSNs can be further improved by jointly working on the performance of the MC and DC. We designed a near-optimal joint energy replenishment algorithm and proposed a central charging point selection algorithm for proper energy transmission. Extensive simulation tests were conducted to indicate the effectiveness and advantages of our algorithms. Figure 2 shows the methodology of the research work. This flowchart contains a multi-step algorithmic methodology to increase power consumption and charging productivity. It starts with the first algorithm (Min-Charging), which distinguishes the arrangement of nodes with the required minimum energy assumption. Depending on whether the energy condition Exy exceeds the most extreme limit of the battery, (B maximum, the cycle process into different paths. It is unlikely that Exy is more remarkable than the calculation of B maximum) the second algorithm (MaxLife) is used to decide on charging, and the cycle checks if Exy, Mcd exceeds B maximum. Assuming it does, algorithm 3 (MinDT) is used to limit rest time and increase energy productivity with the extra chance that Exy, Mcd is not precisely B maximum. Both calculations in algorithm 3 (MinDT) and algorithm 4 (MaxCE) are used to limit the rest time and expand the charging efficiency. This organized methodology is expected to improve the total energy of the system and the functionality efficiency of the overall system. Figure 3 shows the framework of the algorithms. 4. Network Model In RWSNs, it should be desirable that all the sensor nodes should be charged before their energy depletion. To achieve this goal, we need to transmit maximum charging power to the maximum sensor node at each time slot. In RWSNs, a set of rechargeable sensor nodes S=s1,s2,s3,…,sn; a set of sink stations A=a1,a2,a3,…an; and a set of directional chargers D=d1,d2,d3…dn are deployed randomly at different positions where each sensor node senses data and transmits data to the sink node. In RWSNs, we jointly used MC and DC to transmit maximum energy to maximum sensor nodes as per energy demand. When the DC cannot satisfy the recharging delay required by the nodes due to constraints such as energy distribution or scheduling limits, using an MC in addition to a DC ensures timely and efficient energy replenishment for nodes. The MC improves the DC by delivering concentrated charging to nodes with significant energy demands, ensuring load balance and redundancy, preventing node failures, and maintaining network reliability. Additionally, the MC reduces the potential for shadowing or line-of-sight constraints that could impede the DC’s effectiveness. This mixed strategy optimizes network efficiency by allowing the DC to oversee a wide range of coverage. At the same time, the MC focuses on critical energy support, thereby guaranteeing the continuous operation and resilience of the network. After replenishing energy from the basic station, MC selects its shortest routing path to charge sensor nodes that are out of coverage of DC as well as under the coverage of MC. At different time slots, MCs turn their route toward the DC coverage area and transmit energy to the sensor nodes in the DC coverage area to maintain each node’s energy level. Therefore, in the proposed scheme, we scheduled the routing path of MC to cover sensor nodes that are out of the coverage area of DC and those that are under the coverage area of DC. In this research, we applied a tree topology with a fixed routing path, where each sensor node had only a single link to the next hop. One bit of data is sensed by node i and is transmitted to the sink station through a fixed routing table. We divided nodes into different sets. In RWSNs, we must schedule the MC to transmit maximum energy to each sensor node. We divided the time series of DC into different slots T=t…,t. In each time slot, DC changes its orientation from two orientations with 90∘ to transmit charging power to the different no-node in the area of the network. At each orientation, the directional charger covers a different number of nodes with a 90∘ angle. DC covers each orientation with a 90∘ angle to increase the coverage length of the nodes for energy transmission. The total coverage area of DC is a 180∘ angle. MC M covers the remaining 180 angle area which the DC uncovers. In RWSNs, each node has a different energy consumption rate at different times. Each sensor’s energy consumption rate varies and is impacted by several factors. This fluctuation may be related to event-driven sensing, as sensors consume more energy during increased activity than idle times. Moreover, sensors performing various activities require differing energy levels. Besides the sensor’s location concerning energy sources like MCs or DCs, energy consumption may also be affected by variable data transmission demands and adaptive duty cycles that change according to event likelihood or power conservation needs. We need to select an optimal stopping point of MC from each set of sensor nodes to transmit maximum charging power to maximum sensor nodes. This research aims to maximize network lifetime with the help of maximum energy transmission. Thus, we apply the following model for optimization. (1)MAX rCijxij=+Σt=1TΣi=1N (2)s.t. ri+fitin=fitout (3)eoiri+erifiin+etifiout≥eci (4)eci=Σi=1MXijPcij (5)Xij=CijYj Equation (1) shows the sensing rate of node i, and Equation (2) shows the data conservation constraint. For each sensor node, the aggregated incoming data flow equals the aggregated data outgoing flow. Equation (3) calculates the available charging power of every sensor node. Equation (4) is the power which shows that the consumption charging power should never exceed the total charging power. Equation (5) shows the sensor node coverage area in the network area where the MC transmits charging power at the node. Table 1 provides the notations. 5. Charging Model Our design is based on the Powercast wireless charging and sensing platform. We utilized the Friis space equation, which has been experimentally proven to be a good approximation of the energy recharging and commonly applied, to represent the energy recharging model in Equation (6). We claim that our design can work on any energy-recharging model whose charging power is monotone and increases with the distance between the charger and sensor node. (6)Pr=GsGrηLpλ4πd+b2po where d is the Euclidean distance between the charger and sensor node. po is the source of power. Gs is the source antenna gain. Gr is the receive antenna gain. Lp is the polarization loss. η is a parameter to adjust the model for short-distance transmission. λ is the wavelength. b is the parameter to set the rectifier efficiency. We simplify the energy charging model as shown in Equation (7) for easy presentation. (7)P=ad+b2 6. Problem Formulation In our proposed scheme for WRSNs, we first divide the network area into different sets for proper energy transmission to each sensor node. With the help of different sets, we can transmit maximum energy to each sensor node. We used a Voronoi-based algorithm to divide the network area into different sets. In each network set, one DC and MC are used to transmit energy to the sensor nodes in RWSNs. The process of charging with a mobile charger is highly gained. We can combine nodes within different sets N, and then all the nodes in the coverage of DC and out of the DC coverage area in G can be charged at different time slots. 6.1. Time Slot-Based Charging The proposed charging algorithm jointly works with directional and mobile chargers. The schedule of directional chargers and mobile chargers is identical and the charging activities of each set of nodes are repeated at different times. Due to the time-varying nature of energy demand, we divide the time for joint energy transmission in each node into fixed time intervals with length L. MC contains multiple slots for energy transmission. In each time slot, MC runs a tour to selectively recharge sensor nodes from low energy levels. The energy transmission of one node should be finished before the current slot ends to ensure that the next node energy transmission can be timely started in the next period. Sensor nodes and DC consume energy at different rates in the network area. We need to divide the time for energy transmission according to slot length. MC and DC jointly recharge sensor nodes to maintain the energy level of each node. A slot is the smallest time unit where one charger can replenish energy to each node. In the network area, MC is responsible for energy transmission at each set of network areas. In each set of networks, time is divided for different sensor nodes and directional chargers. Each set has one DC which covers sensor nodes. At each time slot, the MC antenna changes its orientation to select a node for energy transmission. The length of the slot is T=a−Tm which is the duration of each node for energy transmission. The required charging period of each node is based on slot length. The cycle length L is the required charging period of all the sensors in each set of networks. The duration of the charging period of each node can be counted in terms of the slot. Suppose the sensor is categorized into the number of N=n1,n2,…,nn in set. The required charging period of each set is T=t1,t2,…,tn in each slot. M=m1,m2,…,mn with each node set and slot-based design. Based on the node set, we can periodically schedule the power replenishment into slots. According to the schedule, MC selectively charges a set of nodes in each period. We can give slots based on periodical charging. We first obtain nodes with equal energy demand and then finally obtain recharging nodes as they have low data sensing rates because of low energy levels. The charging tasks are issued in different slots at each time. A slot is called a working slot once the charging task issued in the slot. This algorithm aims to obtain all the working slots in one stopping point of MC. 6.2. Optimal Stopping Point with Mix Integral Programming In a RWSN, the feasible region is the region where all the constraints are satisfied. Pseudo-code for the feasible region is as B is the base station in the RWSN area. Ri is the radius, n is the number of nodes, and N in one set of network areas. Assume all the nodes use power po. The objective is to refine the continual cessation of the charger inside each network region, with target node A identified by the circle encompassing the charger’s two random stopping positions, x1 and x2. Let d=x1,x2 represent the Euclidean distance. t=i,x denotes the charging time for node i when the charger halts at point x. If t=i,x1,≤t=i,x2 then d=i,x1≤d=i,x2 with all the viable areas remaining as convex polygons. The ideal place or location for a mobile charger typically occurs in the corner of the feasible zone. The feasible zone is defined as pi representing the energy available at source i, where i=1,2,3…n. pj represents the energy accessible at destination j as j=1,2,3,…n. Xij represents energy distribution from source node i to destination node j. Cij represents the energy loss from source node i to destination j. The problem may be formally addressed by mixed integer programming as shown in Equations (8)–(11). (8) Σi=1mΣj=1nCijXij (9) Σi=1mΣj=1n=pii=1,2,3,…n (10) Σi=1mΣj=1nXij=eii=1,2,3,…n (11) xij≥0 Now consider the pair of nodes A and B as the targeted node and base node with feasible regions with R2 and R1 in each set of networks. Let the charger stop at xr, then Equations (12) and (13) are as follows (12) ∀xAϵR2 (13) dxA,xr0. As long as tm exists, even very small, the relation can be obtained as U−np>0, that is n