Modified linear prediction algorithm with low bias estimator

An improved version of the linear prediction (LP) algorithm is proposed in this paper with lower bias of direction of arrival (DOA) estimation. The new algorithm is derived by modifying the cost function in spatial power calculation. The performance of both LP and proposed algorithms is compared in a scenario of two sources of signal using Monte Carlo approach. The benchmark of performance is the bias of estimation, which is studied against the variation of five parameters; signal-to-noise ration (SNR), number of snapshot, source separation, number of elements, and the choice of activated element. From the simulation results, it is shown that the proposed algorithm has better bias of DOA estimation compared to LP algorithm.