Simulation Of Seafloor High Resolution DOA Estimate And Realization On FPGA

Multi-beam bathymetric sonar(MBBS) is an important equipment of marine topography detection, but most of the existing MBBS are based on conventional beam-forming algorithm. The main reason is the computational load of the element-space multiple signal classification(MUSIC) algorithm which is the typical representation of high resolution algorithm is too large to be real-time implemented. This paper studies the possibility of applying MUSIC algorithm to MBBS from two sides.On the one hand, the element-space MUSIC algorithm is transformed to beam-space MUSIC algorithm. Through simulation,this paper compares and analysis the performance and computational load of three typical beam-space MUSIC algorithms.Although the root mean square error(RMSE) of the MSB-MUSIC(Multiple Sub-array Beam-space MUSIC) algorithm is slightly larger than the others', the computation load is more less and the MSB-MUSIC algorithm is still effective for the coherent signal.On the other hand, this paper combines the MSB-MUSIC algorithm with the fast subspace estimate(FSE) which is based on multistage wiener filter(MSWF) and proposes an MM-MUSIC(MSWF and MSB-MUSIC) algorithm. Compared with MSB-MUSIC algorithm, this algorithm reduces the computational load greatly and has high parallel capability at a cost of a little loss of estimation performance.In addition, based on Xilinx AccelDSP synthesis tool and TI C6713DSK platform, the MM-MUSIC algorithm with eight sub-arrays and snapshots is realized and researched in this paper. The results show that the computational speed of two stage FSE is much faster than the singular value decomposition's(SVD), which proves that the MM-MUSIC algorithm has obvious speed advantage of MSB-MUSIC algorithm.