Semi-blind Sparse Channel Estimation and Data Detection by Successive Convex Approximation

The aim of this paper is to propose a semi-blind solution, for joint sparse channel estimation and data detection, based on the successive convex approximation approach. The optimization is performed on an approximate convex problem, rather than the original nonconvex one. By exploiting available data and system structure, an iterative procedure is proposed where the channel coefficients and data symbols are updated simultaneously at each iteration. Also an optimized step size, introduced according to line search procedure, is used for convergence improvement with guaranteed convergence to a stationary point. Simulation results show that the proposed solution exhibits fast convergence with very attractive channel and data estimation performance.