New MODE-based techniques for direction finding with an improved thresholdperformance

We propose two MODE-based direction finding techniques with a drastically i mproved SNR threshold as compared to the conventional MODE method. Our algo rithms preserve all good properties of MODE, such as asymptotic efficiency, excellent performance in scenarios with highly correlated sources, as well as a reasonable computational cost, Similarly to conventional root-MODE, o ur techniques do not require any multidimensional search since they are bas ed on a combination of polynomial rooting and a simple combinatorial search . The first algorithm referred to as MODEX (MODE with EXtra roots) makes us e of a certain polynomial with a larger degree than that of the conventiona l MODE-polynomial, The second algorithm referred to as resampled MODEX is a dditionally based on a statistical resampling procedure to resolve subspace swap problems. In both our techniques, the source directions of arrival (D OAs) are estimated via checking a certain (enlarged) number of candidate DO As using either the stochastic or the deterministic maximum likelihood (ML) function. To reduce the number of candidate DOAs and the whole computation al cost in our procedures, a priori information about source localization s ectors can be exploited straightforwardly. Computer simulations with uncorr elated and coherent sources show dramatic improvements of the threshold est imation performance that can be achieved by our techniques relative to conv entional MODE. (C) 1999 Elsevier Science B.V. All rights reserved.