A novel scheduling algorithm for improved performance of multi-objective safety-critical wireless sensor networks using long short-term memory

The multiple objective optimisation (MOO) challenges encountered in the context of wireless sensor networks (WSNs) present a formidable NP-hard problem. These issues primarily arise from the constraints imposed by critical factors such as connectivity, coverage, and, most notably, energy consumption. Simultaneously fulfilling these three requirements is no longer considered the standard approach for enhancing system dependability. To illustrate, a prospective solution may optimise one or two of these requirements while bolstering overall network energy efficiency. Nonetheless, prior endeavours documented in the extant literature reveal unexplored avenues for enhancement. Hence, this paper introduces a new methodology aimed at alleviating MOO concerns and thereby enhancing the quality of service (QoS) in WSNs. A long short-term memory (LSTM) model is proposed as an analytical tool to deliver an energy-efficient scheduling solution that aligns and optimises WSN parameters, striving to attain the most favourable system performance. The LSTM algorithm’s effectiveness is assessed through the iterative application of periods, confirming the desired QoS levels. The unique feature of LSTM lies in its capability to observe specific event sequences and subsequently establish them as the system’s default configuration for its entire operational lifespan. Once these favourable parameters are identified, LSTM automatically ensures consistent service availability and reliability throughout the network’s lifespan. The results obtained demonstrate the superiority of the proposed LSTM-based scheduling algorithm in comparison to the self-organising map (SOFM)-based node scheduling algorithm. The LSTM-based approach outperforms the SOFM-based alternative by a remarkable 75% in terms of coverage and exhibits a 20% enhancement in network lifetime, all while maintaining equivalent levels of connectivity (i.e., 99%) in both algorithms.

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/)