Energy Efficient Routing Mechanism Based on Similarity Guided Graph Neural Network and Decision Tree with Golden Eagle Optimization in VSN

Abstract

Recent developments in low power sensors have prompted the creation of Visual Sensor Network (VSN). Coverage, availability, network life duration, and energy usage are important issues that arise in VSN. However several energy-efficient protocols have been developed, but those protocols have transmission collisions and energy loss due to increased data redundancy. In order to overcome these challenges, an Energy Efficient Routing based Sleep Scheduling Mechanism (EERSSM) is developed. Camera nodes are randomly deployed in the Visual Sensor Network, and the data is received from the network through a relay node. Energy, distance, and node stability are taken into account while identifying the relay node. The next step is to use a Similarity Graph guided Neural Network (SGGNN) to determine whether neighboring nodes detect similar data. If similar information is detected, a similarity measure is computed. When the value of the similarity measure exceeds the threshold, the node goes into the sleep stage while the other nodes are in the wakeup stage. A decision tree is used to calculate the sleep cycle depending on a few factors. The decision tree has a number of hyperparameters, and those parameters are tuned using Golden Eagle Optimization (GEO). When the update cycle is over, the node awakens and joins in the transmission procedure. This proposed energy efficient routing algorithm is tested with several metrics that attain better performance, like 14.42 J average residual energy, 93% packet delivery ratio, 9.3% throughput value, and 770 s network lifetime. Thus, the techniques used in the proposed approach are the better choice for solving the availability, energy consumption, and network lifetime issues in VSN.