Angle-constrained IIR filter pipelining for reduced coefficient sensitivities

Lookahead pipelining methods can be efficiently used for high sample rate or low-power applications of digital recursive filters. However, pipelined recursive filters cannot be implemented using the conventional cascade or parallel structures with second-order sections, since pipelining is achieved by introducing additional cancelling poles and zeros. Thus, if the poles of the pipelined filters are tightly crowded, it is possible that the pipelined filters suffer from large coefficient sensitivities. In this work, an angle-constrained filter design approach using a modified Remez exchange algorithm is proposed to avoid tight pole-crowding in pipelined filters by scattered lookahead. In addition to the reduced coefficient sensitivities, the constrained poles satisfy the pipelining property without pole-zero cancellation, which leads to hardware-efficient design.