Process noise parameters of beamforming nodes in wireless sensor networks (WSNs)

Abstract

Signals from collaborative beamforming (CB) nodes are always arriving out of phase at the intended receiver due to unsynchronized clock frequencies of different oscillators of these nodes. Beamforming nodes have to synchronize their carrier frequencies in order to eliminate any phase offset of the received signal. To do this, estimation, correction and prediction of phase offsets due largely to these drifts needs be carried out on Software Defined-Radio (SDR) while employing non-linear filters like the Extended Kalman Filter (EKF) so as to obtain a near zero (0) phase offset. This paper presents a method of computing the NI USRP-2920 process noise parameters of the phase noise (dBc/Hz) data obtained with keysight N9320B spectrum analyzer (SA). The effects of phase offset measurements at various offset frequencies of the carrier are shown. Power-law noise model is applied to compute the Allan variance and later the process noise parameters that will be applied with the EKF. It will be noticed that the Allan variance curve truly depicts the standard slope characteristics of white frequency and random walk frequency noises which are special considerations when dealing with USRP's oscillators.