MRNQ: Machine learning-based reliable node quester for reliable communication in underwater acoustic sensor networks

Abstract

Ensuring effective and reliable communication within underwater sensor networks (UWSNs) is a formidable challenge due to their unique characteristics, which include offshore exploration, underwater surveillance and monitoring. UWSNs have proven to be a promising approach in various fields, including research investigations, surveillance operations and underwater disaster response. To advance this field, numerous researchers have dedicated themselves to developing new protocols tailored to UWSNs or refining existing protocols, all with the goal of improving research. One important aspect that continues to attract the attention of researchers is the reliability factor in the underwater environment, leading to constant efforts to improve the overall efficiency of the network and optimize energy consumption. In this work, a machine learning based node reliability calculation algorithm (MRNQ) has been proposed, which takes into account numerous parameters such as the success rate, transmission time, node efficiency, and the network efficiency. The proposed approach outperforms CSLT and TMHCV across key metrics with notable percentage improvements. It achieves a 5.16% higher packet delivery rate, a 22.06% reduction in packet drop rates, a 42.4% extension in network lifetime, and a 0.87676% improvement in malicious node detection.