We present a methodology for designing a specialized hardware structure of the LMS adaptive filter with a
very high sampling rate and many taps, where real-time processing requirements can not be met with
conventional off-the-shelf DSP processors. W...
We present a methodology for designing a specialized hardware structure of the LMS adaptive filter with a
very high sampling rate and many taps, where real-time processing requirements can not be met with
conventional off-the-shelf DSP processors. We show that the excessive critical path delay of the conventional
LMS filter can be reduced and pipelined efficiently by introducing a bulk of delays in the input and the error
signals. And, to reduce the hardware requirements further, filter taps are divided into groups and hardware
resources are shared between groups by interleaving technique.