Abstract: Ultra-Wideband(UWB) is a newly developing high-speed wireless communication technology. It is difficult to sample it directly as its wider band width. However, compressed sensing(CS) provides a feasible way with lower sampling speed. Considering the poor performance of the gradient projection for sparse reconstruction(GPSR) algorithm which has been used in UWB channel estimation, an improved algorithm is proposed in this paper. The improved algorithm adopts the objective function form of original algorithm. It cancels the alternating searching method including search at negative gradient direction or search after projecting the gradient direction onto feasible set, change it to the single method of searching straight at negative gradient direction. This method avoids to solve the high dimensional compute problem of original algorithm and get rid of the confinement of strict restrict condition. Comparing with the gradient pursuit(GP), it adds an restrict condition to the sparse noise and translates into a l₁-norm optimization problem. The experiment results show that this improved algorithm can reduce the computation complexity and enhance the speed compared with GPSR algorithm remarkable, while it can also promote the reconstruction performance compared with GP algorithm.