Deep Learning-Enhanced Soft Output Viterbi Turbo Decoding for Robust EEG Signal Transmission in Telemedicine

Abstract

The accurate transmission of electroencephalogram (EEG) signals is an essential requirement for the proper diagnosis and treatment in telemedicine. This paper presents the potential of Deep Learning-Enhanced Soft Output Viterbi Turbo Decoding (DLSOVTD) to improve the robustness of EEG signal transmission and its bit error rate (BER) performance with respect to Soft Output Viterbi Turbo Decoding (SOVTD). The BER vs. signal-to-noise ratio (SNR) characteristics were investigated for several code rates that proved highly advantageous for DLSOVTD. The results demonstrate the superiority of DLSOVTD over SOVTD across all SNR values, especially at lower SNR, which corresponds to noisy channels. The performance difference is significant, demonstrating deep learning's ability to reduce noise and correct errors for accurate data transfer. Both schemes can achieve low BER at high SNRs, but DLSOVTD still has a bit of a head start, with the ability to provide even cleaner signal transmission. The results suggest that DLSOVTD has the potential to greatly advance diagnostic accuracy, increase the efficacy of remote monitoring, and increase confidence in the use of telemedicine.