Gaussian Mixture Model with Bayesian approach for maximizing RSS-based localization in underwater Wireless Sensor Networks

Source localization is a highly challenging and complex task in underwater environments due to uncertainties and unknown sound propagation speed profiles i n underwater channels, as well as increased Doppler effects and constraints on the energy sources of the sensor nodes. To address these issues, we propose an energy-efficient J oint G aussian M ixture Model with a Bayesian approach for localization algorithms, aiming to improve Received Signal Strength (RSS) accuracy. In this article, we represent the additive noise using a Gaussian Mixture Model to calculate the maximum likelihood estimation. The Bayesian statistical approach solves the convex optimization problem to find effective globally optimal solutions. These joint methods help mitigate the underwater Doppler spread effects and improve the estimation of sensor node positions. The simulated results are analyzed, and the performance metrics show that the proposed GMM-Bayesian approach is very close to the Cram´er-Rao Lower Bound and this method also outperforms other existing localization algorithms in terms of lower Root Mean Squared Error (RMSE) relative to anchor nodes and a better Cumulative Distribution Function (CDF) for localization errors. From the simulation results, it is evident that the proposed approach achieves substantial performance gains in the localization of underwater wireless sensor networks.