Wireless Sensor Network (WSN) is made up of sensors that simultaneously sense, process, and transmit data. WSN routing protocols are vulnerable to the security risk and it is ineffective at managing dynamic network situations, like fluctuating node mobility or shifting ambient conditions. While routing, the early failure of node in WSN is caused by the imbalanced consumption of power and a dissimilar distribution of sensor nodes. So, to overcome these challenges Energy Efficient Tree based Routing Algorithm (EETRA) combined with deep learning algorithm is developed to achieve reliable and secure data transmission. Initially, the nodes are placed in specific locations and grouped into both local and global clusters, and identifies the cluster head using Dual tier zonal stable election protocol (DTZSEP). Energy Efficient Tree based Routing Algorithm (EETRA) is used for data transmission between cluster head to base station to minimize network energy consumption. WSNs are highly vulnerable to security attacks, so the attack detection is crucial in WSN. To perform attack detection, the network characteristics is subjected to pre-processing which is done through Attention-Based Generative Adversarial Networks (ATTN-GAN) and Max Normalization (MA) for imputing the missing values and normalizing the data. Finally for classifying the attack, the Modified Attention based on Kolmogorov–Arnold Networks (MAKAN) is implemented in which the attention layer is integrated by the shuffle attention layer to extract the features efficiently. This proposed Energy Efficient Tree-Based Routing Algorithm (EETRA) was compared with the existing models, which attains higher performance such as average residual energy has 17.4J, the throughput value is 388Gbps, 98% packet delivery ratio and 0.001ms of transmission delay. In the classification, the proposed MAKAN attains the accuracy, selectivity, NPV and error values of 98.30, 96.20, 97.70 and 1.70% respectively which are further compared with existing approaches. These proposed integrated technique enables data transmission in secure and efficient manner by accurately detecting the attack in wireless sensor networks.
Jagwani et al. (Sun,) studied this question.