Adaptive filter application in underwater acoustics and implementation on FPGA

In this study, the performance of the Least Mean Square (LMS) and Recursive Least Square (RLS) algorithms, which are adaptive filter algorithms instead of traditional digital filtering methods used in the field of underwater acoustics, is investigated and the application method is shown. Field Programmable Gate Array (FPGA) hardware, which is widely used in signal processing applications, was chosen as the implementation environment. In the study, firstly, LMS and RLS algorithms were analyzed mathematically for their implementation in hardware, then both algorithms were implemented in FPGA hardware using communication signals recorded in the underwater environment. According to the results obtained, the LMS algorithm increased the Signal to Noise Ratio (SNR) value from 8.85 dB to 34.71 dB, and the RLS algorithm increased it to 28.69 dB. In terms of FPGA resource usage, the LMS algorithm used 2.68% Lookup Table (LUT), 4.7% Flip-Flop (FF), while the RLS algorithm used 1.06% LUT and 1.63% FF. In other resource usage, both algorithms used 0.42% Digital Signal Processing (DSP), 1.9% Input-Output (IO) and 3.13% Global Buffer Memory (BUFG). According to these results, the LMS algorithm provided more SNR improvement while RLS algorithm consumed less resources. In this study, it has been shown that LMS and RLS algorithms improve the SNR value in underwater communication data, can be used in applications in the field of underwater acoustics and can contribute to the studies in this field.