RELAXED LOOK-AHEAD PIPELINED LMS ADAPTIVE FILTERS AND THEIR APPLICATION TO ADPCM CODER

The relaxed look-ahead technique is presented as an attractive techniq ue for pipelining adaptive filters. Unlike conventional look-ahead, th e relaxed look-ahead does not attempt to maintain the input-output map ping between the serial and pipelined architectures but preserves the adaptation characteristics. The use of this technique results in a sma ll hardware overhead which would not be possible with conventional loo k-ahead. The relaxed look-ahead is employed to develop fine-grained pi pelined architectures for least mean-squared (LMS) adaptive filtering. Convergence analysis results are presented for the pipelined architec ture. The proposed architecture achieves the desired speed-up with mar ginal or no degradation in the convergence behavior. Past work in pipe lined transversal LMS filtering are shown to be special cases of this architecture. Simulation results verifying the convergence analysis re sults for the pipelined LMS filter are presented. The pipelined LMS fi lter is then employed to develop a high-speed adaptive differential pu lse-code-modulation (ADPCM) codec. The new architecture has a negligib le hardware overhead which is independent of the number of quantizer l evels, the predictor order and the pipelining level. Additionally, the pipelined codec has a much lower output latency than the level of pip elining. Theoretical analysis indicates that the Output signal-to-nois e ratio (SNR) is degraded with increase in speed-up. Simulations with image data indicate that speed-ups of up to 44 can be achieved with le ss than 1 dB loss in SNR.