Underdetermined source separation of finite alphabet signals via l1 minimization

This paper addresses the underdetermined source separation problem of finite alphabet signals. We present a new framework for recovering finite alphabet signals. We formulate this problem as a recovery of sparse signals from highly incomplete measurements. It is known that sparse solutions can be obtained by ℓ₁ minimization, through convex optimization. This relaxation procedure in our problem fails in recovering sparse solutions. However, this does not impact the reconstruction of the finite alphabet signals. Simulation results are presented to show that this approach provides good recovery properties and good images separation performance.