Performance comparison of supervised and unsupervised equalization algorithms for frequency selective channels
Fazal-E- Asim, Sajid Bashir, Shafayat Abrar, Syed Ismail Shah · 2016
Adaptive equalization are mostly used in every wireless communication systems because the wireless channel is very unpredictable and time varying in nature. This fast randomly changing channel creates many issues in wireless communication system e.g. inter symbol interference (ISI) which degrade the overall performance of wireless communication system. To mitigate these types of channel affects, adaptive algorithms are the most powerful signal processing tool which helps in the removal of ISI. In this paper we compare the performance analysis of supervised/training based equalization and unsupervised/blind equalization algorithms for a wireless communication channel having exponentially decaying power delay profile (PDP). The algorithms selected for comparison are least mean square (LMS), recursive least square (RLS), Multimodulus algorithm (MMA) and Square Contour algorithm (SCA) respectively. The performance measuring parameters are ISI and bit error rate (BER). It is observed that training based equalization algorithms i.e LMS and RLS perform efficiently in terms of removal of ISI as compared to blind equalization algorithms i.e MMA and SCA respectively. The training based algorithms tends to converge at very fast rate as compared to blind algorithms and hence suitable for fast fading channels but with the cost of additional bandwidth required for training. The blind algorithms show more computational complexity as compared to training based algorithms. Achieving the same performance level in terms of BER, an SNR loss of 7 dB is noted if we use blind algorithms instead of training based algorithms.