Abstract: We consider multiple-input multiple-output (MIMO) radar system with widely spaced antennas, whose parameters of estimation under compressed sensing algorithms are always influenced by false peak or wrong localization. A new algorithm for parameters estimation based on Group lasso is present, which utilizes SOCP to improve the precision under low SNR, since the RCS of target meet with the block sparse. By contrasting with conventional CVX, the proposed algorithm gets better performance of estimation. In addition, by use of the estimator of reflectivity, one strategy of transmitting energy allocation is also proposed by transfer MIMO radar into MISO radar. At the same time, we would double the energy of one transmitter, which has the bigger value of RCS. MISO radar with energy allocation gains an advantage over MIMO radar by using parameter metric, especially in the case for RCS of target is fluctuant. The analytical and empirical results establish that the proposed algorithm is better than classical method for estimation in MIMO radar. Simulation results verify the usefulness of the proposed algorithm.