The sampling speed for the ultra wideband (UWB) channel is too high to realize with the existing sampling technology. To solve the problem, a novel blind channel estimation algorithm was presented based on the theory of compressive sensing. Firstly, some measurements are obtained which are linear combinations of the received signals multiplied by a random incoherent measurement matrix. Then, the mathematical model is established by exploiting the first statistics of the measurements. Finally, the orthogonal matching pursuit (OMP) algorithm is utilized to get the estimating channel parameters. With the proposed algorithm, the number of the measurements need for channel estimation is much smaller than that of the samples needed for the existing algorithms, which reduces the ADC resources greatly. The simulation result shows that the estimation performance of the algorithm is good, while the bit error rate (BER) is only 2~3dB higher than that of the exact channel.