Blind Deconvolution for SISO Systems
Temel Engin Tuncer · 2006
A deterministic solution is presented for the blind deconvolution of signals with symmetric or cyclic extensions. It is shown that nonlinear equations can be converted to a form where the input or the filter coefficients are found by using a closed form expression. This solution works for non-finite alphabet inputs in general. If the input is from a finite alphabet, the solution becomes significantly robust to noise in the samples. Proposed solution works independent from the input statistics and length and does not require any assumption on the input except the symmetric or cyclic extensions.