Semi-blind collision resolution in random access wireless networks

Signal processing based collision resolution for random access wireless networks with unknown propagation channels is considered. As a performance measure for receivers with a collision resolution capability, the resolvability function is defined as the probability of resolving at least k packets from K colliding ones given m known data symbols embedded in the data packets. The resolvability performance of the zero-forcing and the semi-blind least squares smoothing receivers is analyzed, which demonstrates significant improvement of semi-blind approaches over trained algorithms that rely entirely on known symbols in the data packets