Wireless link modeling using complex FIR filters

This paper presents a method for wireless link modeling, based on complex coefficient finite-impulse-response (FIR) digital filters. The complex coefficients of the proposed FIR model representation allow easy, unconstrained least-squares estimation of the target transfer function in the frequency domain, which is performed using the Moore-Penrose matrix pseudoinverse. With the reduction of FIR models, a time-gating effect can be obtained, which is not restricted to the usage of the equidistant samples in the frequency-domain measurements.