We propose a new way to estimate the sparse multipath channel of an Amplify and Forward Two Way Relay Network (AF-TWRN) based on Compressive Sensing (CS), using Compressive Sampling Matching Pursuit (CoSaMP). MSE based performance of CoSaMP and OMP methods and corresponding BPSK based BER performance of the system using the estimated channels are plotted. CoSaMP based channel estimation is performing similar to OMP based channel estimation, with lesser time complexity, in both MSE and BER measures of the system, when the sparsity/worst case sparsity of the channel is known.