Performance Analysis of MSFRLS-VFF Based Real-Time Adaptive Noise Canceller with RLS and APA Algorithms Using TMS320C6713 Processor

摘要

In this paper, the real-time adaptive noise canceller (ANC) system is implemented using modified sigmoid function variation based recursive least square with variable forgetting factor (MSFRLS-VFF) algorithm. The experiment is performed on DSP TMS320C6713. The performance of MSFRLS-VFF algorithm is evaluated and comparison is made with conventional RLS and affine projection algorithm (APA). In the experimental setup, the different types of noises are artificially added into the clean signals to make the signal noisy at different input SNR levels (-5dB to 10dB). The result shows that MSFRLS-VFF algorithm provides superior performance than RLS and APA algorithm (the best performance achieved for SNR improvement is 2.2dB over RLS and 2.1dB over APA at -5dB input SNR with filter order 10). Also, the MSFRLS-VFF algorithm provides minimum mean square error than RLS and APA algorithms, however computational complexity of APA algorithm is less as compared to MSFRLS-VFF and RLS algorithms. The output error-free signal obtained after MATLAB simulation and from TMS320C6713 DSP shows the similar result that proves the correctness of the setup.