Least mean squared error-design of complex FIR filters with quantized coefficients

The design of complex finite-impulse response filters with quantized coeffi cients according to a discrete mean squared error (MSE) criterion in the fr equency domain is revisited. In the first approach, the coefficients are qu antized step-by-step by recursively solving a system of linear equations. T he second approach is a generalized form of the first one: The MSE is repea tedly minimized considering a growing number of fixed, quantized coefficien ts. The first approach requires the same computational burden as the design of unquantized coefficients, in contrast to the second where the higher fl exibility is paid with a higher expenditure. For a considerable number of t est cases the first approach and one variant of the second achieves a small er MSE than the direct quantization method.