Hybrid Random Forest Classifier Adaptive Modulation Scheme For Under Water Optical Communication For Non-Gaussian Channels

Abstract

Back Ground:The goal of this research is to create an Hybrid adaptive modulation system that maximizes optical communication in non-Gaussian channels, especially in the demanding underwater environment. In optical communication, the same conventional modulation schemes e.g. BPSK, QPSK, QAM are often employed, though these schemes work poorly under practical conditions with varying channel and non-Gaussian noise conditions.

Method:To solve this limitation, we propose a dynamic system that will use the current signal-to-noise ratio (SNR) combined with the noise profile of the optical channel to select the most suitable modulation technique (e.g. BPSK, QPSK, 16-QAM, 64-QAM). The technology has the benefit of maximizing communication performance by adapting to variable channel conditions and ensuring reliable and efficient connection in diverse conditions. To be able to formulate optical communication scenarios, such as underwater communication where turbidity, scattering, and absorption have to be considered, the proposed system includes the non-Gaussian channel model where both Gaussian and Weibull noise distributions are used in a mixture.

Result:An AI model based  Random Forest classifier is trained to forecast the suitable modulation scheme for effective communication, measured with parameter metrics such as Bit Error Rate (BER) and spectral efficiency across different modulation schemes and noise situations, the performance of this adaptive system is assessed.

Conclusion:By increasing noise resilience, power efficiency, and overall system performance in non-Gaussian environments, simulation results show that the adaptive modulation system performs better than static modulation techniques, increasing the dependability and efficacy of optical communication systems.