Optimization of binary transversal filters with constrained sampled impulse response

The design of binary transversal filters (BTF's) suitable for shaping binary data signals is considered. It is shown that the approximation problem for such BTF's can be solved with respect to minimum mse or min—max time and/or frequency domain error, under the constraint that the impulse response exhibits a specified intersymbol interference. The BTF's considered employ only the lowpass filtering inherent in the non return-to-zero input data and exhibit the property that for a fixed number of taps there is an optimum BTF clock rate which minimizes the approximation error. The design of BTF's with quantized coefficients is also considered and the approximation problem solved with respect to min—max error by using mixed integer programming.