Energy Conservation in Wireless Sensor Networks Using Partly-Informed Sparse Autoencoder

Abstract

The energy in wireless sensor networks is considered a scarce commodity, especially in scenarios where it is difficult or impossible to provide supplementary energy sources once the initially available energy is used up. Even in cases where energy harvesting is feasible, effective energy utilization is still a crucial step for prolonging the network lifetime. Enhancement of life-time through efficient energy management is one of the essential ingredients underlining the design of any credible wireless sensor network. In this paper, we propose a sensor selection method using a novel and unsupervised neural network structure referred to as partly-informed sparse autoencoder (PISAE) that aims to reconstruct all sensor readings from a select few. The PISAE comprises three submodules, namely: the gate (which selects the most important sensors), encoder (encodes and compresses the data from select sensors), and decoder (decodes the output of the encoder and regenerates the readings of all initial sensors). Our approach relies on the premise that many sensors are redundant because their readings are spatially and temporally correlated and are predictable from the readings of a few other sensors in the network. Thus, overall network reliability and lifetime are enhanced by putting sensors with redundant readings to sleep without losing significant information. We evaluate the efficacy of the proposed method on three benchmark datasets and compare with existing results. The experimental results indicate the superiority of our approach compared with existing approaches in terms of accuracy and lifetime extension factor.

Hybrid Sensor Selection Technique for Lifetime Extension of Wireless Sensor Networks

“…The partly informed sparse auto encoder (PISAE) that primarily based energy conservation is brought in [12] for prolonging the WSNs lifetime. PISAE aimed to improve lifespan by inserting sensors with redundant readings to sleep without dropping big information.…”

Snapshot of Energy Optimization Techniques to Leverage Life of Wireless Sensor Network

“…Apart from the above discussed frequently used ML approaches, there are various other approaches too viz. Deep reinforcement learning (Ashiquzzaman et al [59], Ke et al [60], Nguyen et al [61]), deep neural network with sparse autoencoder (Ayinde [62]), opportunistic routing (Dinh et al [63]), energy saving using simulated annealing (Kang et al [64]), and energy harvesting using artificial neural network combined with linear regression (Kwan et al [65]), Hence, it can be said that there are various dedicated research attempt towards using ML approach over WSN; however, not all the approaches are found to directly address energy problems in WSN.…”

Minimization of Energy Consumption for Routing in High‐Density Wireless Sensor Networks Based on Adaptive Elite Ant Colony Optimization

“…And this path should meet multiple QoS constraint requirements such as delay jitter, delay, packet loss rate, and link bandwidth. These QoS routing conditions are used as the criteria for QoS routing considerations [6]. Due to energy limitations, how to maximize the reduction of routing energy consumption and extend the lifetime of sensor networks have become a bottleneck problem faced by HDWSNs.…”