Implementation and performance analysis of low complex beamforming algorithms

Abstract

The work comprises of low complexity and cost effective technique referred as RAB (Robust Adaptive Beamforming) with the use of two algorithms i.e. LOSCME (Low complexity shrinkage based mismatch estimation) and OKSPME (Orthogonal krylov subspace projection mismatch estimation). The LOSCME is used to estimate the steering vector based on the correlation of the observed input data and beamformer output. This algorithm also uses OAS (oracle approximating shrinkage) in order to estimate the input data covariance matrix and INC (interference noise covariance) matrix that only requires a prior knowledge of angular sector in which the actual steering vector is located. It is not cost effective and need not to know any extra information that is related to interferers which keeps away from finding the direction of all interferers. Simulation result of LOCSME technique shows very close to optimum. The OKSPME algorithm is based on cross-correlation estimation between the observed input data array and output of the beamformer. In this technique the steering vector mismatch is estimated by considering the larger dimension linear equation and FOM (full orthogonalization method) is used to decrease the dimensional subspace. In addition to this, adaptive algorithm is implemented, which is based on the SG (stochastic gradient) & CG (conjugate gradient) searches used to update the beamforming weights, leads to low complexity of the system and avoid any costly matrix inversion. Major advantage of this mismatch estimation with low rank proposed algorithm is cost efficient when using with number of sensor arrays. The result of OKSPME technique shows excellent performance in terms of output SINR (signal to interference noise ratio), increase in the directivity and also increase in the antenna gain, when compared with other RAB algorithms.