Systolic implementation for 2-D unitary ESPRIT
Tomoki Abe, Kenji Araki · 2002
We discuss the systolic array implementation of 2-D unitary ESPRIT. By exploiting the idea of Unitary ESPRIT we modify the algorithm into totally real valued process, which can reduce the computational burden of each processor. An important issue in extending the 1-D systolic ESPRIT into its 2-D case is to achieve the simultaneous upper-triangularization of 3 compressed matrices which we call SGSD. For this aim we employ the Jacobi-type method incorporated with iterative shifted QR method to obtain an approximate decomposition. Finally a simulation result shows that the proposed algorithm can track the azimuth and elevation of moving sources.