Spatial Spectrum Estimation Method Based On One Bit Sampling

Spatial spectrum estimation play an important role in array signal processingresearch, is widely used in radar, navigation, and other areas of life. In practiceapplication, the observed value need to be sampled and quantified, and that would tobring irreversible quantization error. This paper is focus on a special samplequantization, the1-bit sampling quantization.1-bit sampling only need a binary number to represent the information, which notonly reduces the memory space, that required to store information, but also decrease thenumber of bits which required in the transmission process. From the hardwareimplementation, Sample complexity is decreased because we use less number of bits toexpress information. The1-bit quantize can be accomplished with a simple, high-speedcomparator, therefore, reduce the cost.This paper research mainly includes the following aspects:1) Through the spatial spectrum estimation system, elaborates several aspectsabout spatial spectrum estimation, including the mathematical model, factors that canaffecting the estimation, parameters which could measure the performance of thealgorithm. Besides this, details of the classic estimation algorithm do simulation andanalysis with MVDR, MUSIC, and ESPRIT algorithm. It lays the theoretical foundationfor one bit spatial spectrum estimation.2) Studying the1bit sparse reconstruction, which means when the observed valueis sampled to1bit, only keep the symbolic information to reconstruct the sparse signal.It is prepared for complete the spectrum estimation based on one bit sampled.3) Doing the1-bit sampling of the observed value with different sampling models,the data lost the amplitude information, only left the symbol information. In this case,use the classic spatial spectrum estimation algorithm could get good performance withmultiple snapshots and array element.4) With the appropriate division of the space angle, we can establish sparse modelfor Spatial spectrum estimation, Basing on this model, use series expansion to correctthe model, considering the concept of a consistent recovery, using observed values which is sampled, to perform DOA estimation with1-bit sparse reconstructed method.Simulation results shows that our method, which compared with the MUSIC, MVDR,can provide better estimation when there is few array elements and snapshots.5) Use greedy algorithm to complete the DOA estimation with the one bit sampledvalues. Simulation results show that the matching pursuit algorithm can give betterestimation than MUISC, MVDR when there are few snapshots.