Bounded-Error Subset Selection for Sparse Channel Estimtation

We consider the channel estimation problem in a wireless code division multiple access (CDMA) system. Under the assumption that the channel gains vector is sparse, we propose a bounded error subset selection (BESS) technique for estimating the channel gains. We assess the performance of the channel estimate via the mean square error (MSE) criterion. We compare the proposed algorithm with a well-known benchmark for solving sparse problems, namely, the least absolute shrinkage and selection operator (LASSO) algorithm. We show that our proposed algorithm provides superior performance to the LASSO algorithm. In particular, the proposed algorithm provides a significant reduction in MSE, reduced complexity, and sparser channel gain estimate compared to the LASSO algorithm.