If P antennas measure a K-multipath channel with N uniformly sampled measurements per channel, the algorithm possesses an OpKP N log N q complexity and an OpKP N q memory footprint instead of OpP N 3 q and OpP N 2 q for the direct implementation, making it suitable for K ! N . The sparsity is estimated online based on the PER, and the algorithm therefore has a sense of introspection being able to relinquish sparsity if it is lacking.The estimation performances are tested on field measurements with synthetic AWGN, and the proposed algorithm outperforms non-sparse reconstruction in the medium to low SNR range (ď 0dB), increasing the rate of successful symbol decodings by 1{10 th in average, and 1{3 rd in the best case. The experiments also show that the algorithm does not perform worse than a nonsparse estimation algorithm in non-sparse operating conditions, since it may fall-back to it if the PER criterion does not detect a sufficient level of sparsity.The algorithm is also tested against a method assuming a "discrete" sparsity model as in Compressed Sensing (CS). The conducted test indicates a trade-off between speed and accuracy.