Minimizing the energy depletion in wireless rechargeable sensor networks using bi-level metaheuristic charging schemes

TL;DR

This work tackles energy depletion in Wireless Rechargeable Sensor Networks by formulating a bi-level optimization problem that jointly optimizes the charging path and charging time under partial charging constraints. It introduces two metaheuristics, MLSGA and MTBCS, to explore multiple upper-level paths and to optimize lower-level charging times via GA and multitasking CMA-ES, respectively. The methods are grounded in a detailed network model with realistic energy dynamics and are validated across diverse network scenarios, showing significant reductions in dead sensor nodes compared with state-of-the-art on-demand and periodic schemes. The findings demonstrate the advantage of simultaneous path-time optimization and bi-level, hybrid search strategies for prolonging network lifetime under energy and bandwidth constraints, with practical implications for scalable WRSN deployments.

Abstract

Recently, Wireless Rechargeable Sensor Networks (WRSNs) that leveraged the advantage of wireless energy transfer technology have opened a promising opportunity in solving the limited energy issue. However, an ineffective charging strategy may reduce the charging performance. Although many practical charging algorithms have been introduced, these studies mainly focus on optimizing the charging path with a fully charging approach. This approach may lead to the death of a series of sensors due to their extended charging latency. This paper introduces a novel partial charging approach that follows a bi-level optimized scheme to minimize energy depletion in WRSNs. We aim at optimizing simultaneously two factors: the charging path and time. To accomplish this, we first formulate a mathematical model of the investigated problem. We then propose two approximate algorithms in which the optimization of the charging path and the charging time are considered as the upper and lower level, respectively. The first algorithm combines a Multi-start Local Search method and a Genetic Algorithm to find a solution. The second algorithm adopts a nested approach that utilizes the advantages of the Multitasking and Covariance Matrix Adaptation Evolutionary Strategies. Experimental validations on various network scenarios demonstrate that our proposed algorithms outperform the existing works.

Figures (9)

• Figure 1: A wireless rechargeable sensor network system.
• Figure 2: Energy fluctuation of a sensor in a charging cycle $T$, where $t_0$ is the timing that $MC$ starts at the depot. Energy of the sensor decreases in the first phase ($t_0$ to $t_1$) and the third phase ($t_2$ to $T$). The second phase ($t_1$ to $t_2$) is the timing that the sensor's battery is replenished.
• Figure 3: An example of calculating similarity between two charging paths. There are 5 pair of sensors: $\langle1, 2\rangle$, $\langle1, 4\rangle$, $\langle3, 2\rangle$, $\langle3, 4\rangle$, and $\langle2, 4\rangle$ that appear in both $P_1$ and $P_2$. Thus, $\epsilon_{P_1,P_2} = 5$.
• Figure 4: Comparison of the node failure ratio and the number of sensor nodes in different sensor distribution.
• Figure 5: Comparison of node failure ratio and average energy consumption rate for different sensor distributions