Topological Information Embedded Convolution Neural Network–Dependent Energy Alert‐Cluster Head Selection in WSN

ABSTRACT Energy efficiency is a major challenge in developing wireless sensor networks (WSNs). The cluster head (CH) can be selected at random or depending on one or more criteria that leads to increase the network lifespan directly. Nevertheless, the CH selection creates an optimization problem. For this purpose, a number of researches have been presented so far to select the optimum CH with the help of various optimization methods, but none of them effectively solves this problem. Therefore, a topological information embedded convolution neural network based energy alert‐cluster head selection in wireless sensor network (TIECNN‐EAC‐WSN) is proposed in this paper. In this method, cluster formation and CH selection is performed by topological information embedded convolution neural network (TIECNN). The CH selection is carried out by three features: energy stabilization, minimization of distance among nodes, and minimization of delay during data transmission. Then, the optimal route is selected by using improved manta ray foraging optimization (IMFO). The TIECNN‐EAC‐WSN approach is evaluated with some metrics, such as network lifetime, number of alive sensor node, and energy consumption with different scenrios. The stimulation results show that the proposed TIECNN‐EAC‐WSN method attains 23.20%, 27.22%, and 26.28% higher number of alive sensor node when compared with the existing models: energy‐aware optimization clustering for hierarchical routing in WSN (EAC‐HR‐WSN), energy‐aware clustering depending on fuzzy modeling in WSN utilizing modified invasive weed optimization (FMEAC‐WSN‐IWO), and quantum tunicate swarm approach–dependent energy‐aware clustering mode for WSN (QTSA‐EAC‐WSN), respectively.