Research On Angle-of-Arrival Estimation Using Sparse Approximation Methods With Sensor Arrays

AOA (Angle-of-Arrival) estimation is a key technology in radar, sonar, communication, seismic prospecting, radio astronomy and etc. It has a long history which can dates back to earlier than the second world-war. Traditional AOA methods are simply understandable and applied widely while they are not so accurate and have low resistance to noise. When have to be confronted with the more and more complicated electromagnetic environments, their performance can hardly satisfy any engineer. And at the same time, most of their mutations impose more complexities on systems. In order to solve these problems, a new concept called antenna array has arisen. Except for its high resolution and good performance, its drawbacks are obvious as well, such as constrained by the number of elements, behaving badly when correlativity of sources grows and when confronted with complicated circumstances and etc. It also makes the measurement unit more complicated and larger. In this paper we applied sparse approximation methods into the issue. These non-parameters methods perform well under simulation environments, main works on them are showed as follows:1. Depict the history of AOA estimation techniques, explain the reason of the advent of phase ambiguity and design algorithms to solve the problem.2. Summarize existing sparse approximation methods called basis selection methods (forward sequential basis selection methods: Basic Matching Pursuit (BMP), Order Recursive Matching Pursuit (ORMP) and Modified Matching Pursuit (OMP), and parallel basis selection method: the FOCal Under-determined System Solver (FOCUSS) and l~1-SVD algorithm).3. Apply MP (Matching Pursuit) method into electronic reconnaissance and derive estimation equations which fuse frequency and space information together and perform well under various conditions; find the AOA estimation structure of the equations and discuss its performance under multiple sources and snapshots, additive noise regime. 4. Design AOA estimation algorithms using Matching Pursuit: sectors thinning, beam-space MP (BSMP), rotational invariance MP (RIMP), and non-ambiguous MP; show their advantages and drawbacks as well; compare with CBF, ML, MUSIC, B-MUSIC, SSMUSIC, FOCUSS , BMP and theoretical bounds (CRB) to demonstrate its advantages.5. Derive the relationship between MP, CBF, MVM, MSM, Sub-Space methods, and sparse approximation methods (BMP, OMP, FOCUSS and etc.).6. Design three dimensional parameters (frequency contents, elevation, and azimuth) estimation algorithm and analyze its error distribution.7. Show the performance of algorithms under complicated electromagnetic environments (non-Gaussian noise, colored noise and clutter) and propose optimization measures which is proved through numerical simulations.According to the comparison, good performance and robustness of MP have been proved:1) Resolve more sources;2) Resolve coherent sources;3) More accurate;4) do well in low SNR regime;5) More resistant to complicated electromagnetic environment.At the same time, their drawbacks are obvious as well:1) Lower resolution than sub-space methods;2) Slower than sub-space methods.Numerical simulations have proved the effectivity of proposed algorithms, if there is somewhere in this paper can not be self-proved or is not clear, I'd like to listen.