High Performance Flexible Adaptive Filtering for Digital Communication Systems

摘要

The purpose of the research was to investigate the design and finite precisionimplementation of high performance fast recursive least squares adaptive filtering algorithms for high speed digital communications systems. The multi-channel general order Fast Transversal Filter (FTF) algorithm was investigated because, besides its efficiency and fast convergence, it can be configured for various adaptive filtering tasks. The performance of the algorithm was evaluated using a detailed block diagram simulation of a full duplex 9600 bits per second voice-band modem. All major components of the system were modeled and simulated, including accurate models for realistic channels. The adaptive filter was used both for adaptive complex equalization with decision feedback and in echo cancellation. The performance of the algorithm for both floating point and fixed point implementations was compared with the conventional normalized LMS algorithm. Although the LMS algorithm failed to perform in typical and worst case channels, the multi-channel general order FTF algorithm performed flawlessly, effectively removing channel distortion and interference.