Robust spatial multicasting for randomly distributed sensors

Abstract

Beamforming is used in sensor networks to transfer data to remote stations with enhanced communication. When the position of array nodes and directions are known, it is better to form an approximate uniform linear array (ULA), uniform circular array (UCA) and uniform elliptical array (UEA). Since it is not always possible to find actual sensors matching with the ideal positions of ULA, UCA and UEA, least-square (LS) estimation is used to calculate weights of the array elements. LS estimation does not give realisable weight values and yields transmission in unintended directions. Also, if many single beams arrays are used to transfer data simultaneously to different users, co-channel interference occurs. In this study, to overcome these problems, multibeam array with restriction imposed on the weights is proposed and is reformulated as Tikhonov regularised LS (TRLS) method. This proposed spatial multicasting method is found to be robust up to 90% position variation of sensors for the ULA with weights under control improving the power efficiency and hence the spectral efficiency and data rate. Moreover, it is also observed that when a faulty node is substituted by nearby nodes, the weights calculated using TRLS method need not be recalculated.