Low-dimensional approximation of recursive least squares algorithm applied to decision feedback equalizer on underwater acoustic communication

Underwater acoustic channels continue to be some of the most difficult channels for communication owing to the characteristics of the sea water and acoustic propagation. For actual use, single carrier modulation with a decision feedback equalizer has been utilized in underwater acoustic communication for decades. The equalizer can follow relatively fast time-varying channels when the channel estimation is updated according to the recursive least squares algorithm, which is one of the Kalman filter algorithms. However, the processing incurs high computational costs depending on the number of receivers and filter lengths in the equalizer. Recently, an efficient approximation for Kalman filter was proposed to reduce the calculation costs. In this study, the approximation is applied to the recursive least squares algorithm in the equalizer to reduce the calculation costs. Furthermore, previous experimental data is demodulated with the derived algorithm. Consequently, the demodulation performances obtained are statistically better than those obtained by conventional processing in this study. This indicates that the approximation reduces the costs without degradation of demodulation performance.