W. Khlebnikov, S.-H. Kim, and D.-H. Kim (Korea)
Adaptive arrays, numerical stability, singular values, and systolic arrays
A novel fast stable algorithm of adaptive beamforming is derived using the conjugate-direction decomposition (CDD) of the optimal weight vector. To provide the highest numerical stability of computations, a stabilized version of the modified Gram-Schmidt (MGS) orthogonalization with suppressed propagation of numerical errors is applied. The new algorithm demonstrates the same fast convergence rate as many of RMS-algorithms, but features with reduced computational cost and higher numerical stability that makes possible its effective implementation in fixed-point real-time adaptive beamformers.
Important Links:
Go Back
