Abstract

This study focuses on removing the possibility of malicious data manipulation in wireless sensor networks (WSN) by utilising a deep learning method. When training deep neural networks, datasets that have been the subject of an attack that alters the data are used as the building blocks. This is done in preparation for putting the networks to the test in the real world. We find out through simulation with a 70:30 cross-validation across a 10-fold sample size that the proposed technique is superior to the current state of the art in terms of the packet delivery rate, latency and throughput.