The purpose of this proposal is to design and build a generic adaptive filtering module for digital communication systems. The design will be based on the multi-channel fast Recursive Least Squares (RLS) adaptive filtering algorithm. Designs will be introduced in which echo cancellation, equalization, automatic gain control, and timing and carrier recovery are all performed digitally on single or multiple digital signal processors. Both fixed point and floating point implementations will be developed and the finite precision issues involved in mapping these algorithms to finite wordlength DSPs will be addressed. Real time and high level simulations of these algorithms will be performed over realistic channels. These include fading in mobile digital radio and voice-band channels with various impairments. Present digital communication systems based on current adaptive filtering technology fail in worst case channels for voice-band modems and, in the case of mobile communication systems, during rapid fading. Higher speed communication over channels with more noise and distortion will become feasible by employing these new algorithms. Furthermore, the all digital DSP-based design ensures manufacturability with greater reliability and little or no time consuming manual adjustments.The potential commercial application as described by the awardee: Both high speed modem and mobile digital radio equipment manufacturers will use the proposed high performance adaptive filtering module to achieve robust and reliable communication. They will be able to penetrate a larger share of the market where equipment based on current technology fails to provide quality service.
