Power efficient folding of pipelined LMS adaptive filters with applications to wireline digital communications

In this paper, a new two-step algorithm is introduced for power efficient i mplementation of the folding algorithm transformation of the LMS adaptive F IR filters. The first step handles the scheduling within the folding sets t o reduce the switching activity using a greedy algorithm solution for the t raveling sales person (TSP) NP-hard problem. The greedy algorithm gives a s ub-optimal solution of the TSP problem with low implementation cost. For a typical word-length, and for large folding factors, the projected reduction in switching activity can be as large as 50%. HEAT tool was used to simula te the effect of this reduction using a typical Baugh-Wooley multiplier and the reduction in power consumption for a wireline equalization implementat ion was found to lie between 25% and 60%. The new algorithm is based on a s imple breadth-first search approach and can be easily performed for one tim e only when the filter is geared to the steady-state mode. The second step involves optimal pipelining of the folded hardware elements for minimum pow er. The tradeoff between the number of pipelining levels and the power cons umption is evaluated. To compensate for the LMS degradation due to pipelini ng, a 1-pole IIR compensator is introduced in the error loop for relaxed LM S. The IIR based relaxation outperforms the relaxed lookahead pipelining by 2-3 dB of output error. Another feature of the IIR relaxation is the "smoo thing" nature of the filter, thus reducing the effective switching activiti es on the multiplier input. The combined effect of the two techniques can l ead to power savings up to 65% compared to a normal folded structure. Simul ations for the combined techniques are carried out for the digital subscrib er loop channel and significant savings in power are demonstrated.