Estimation of fading channels with a parallel matching pursuit structure

A parallel matching pursuit followed by cancellation (PMPC) algorithm is presented for estimation of multipath channels with large possible delay spreads. In the transient phase of the parallel MP, an increasing number of channel coefficients is chosen in each iteration. The selected coefficients are next accurately estimated using a cancellation step in the algorithm. Unlike a serial MPC (SMPC) approach, the PMPC algorithm enables faster convergence in the transient phase, and allows readjustment of the selected channel coefficients. The performance of PMPC, however, depends on a set of weighting coefficients. Our study shows that PMPC can perform significantly better than conventional least squares under sparse channel conditions.