Optimal Direction Finding in Unknown Noise Environments Using Antenna Arrays in Wireless Sensor Networks

With the advancements in wireless sensor network (WSN) platform architecture and cost-effective smart antennas, it is feasible to integrate antenna arrays on the sensor node in the same dimensions with slightly additional cost, and some integrated platforms have been reported. In this paper, we consider the challenging problem of direction finding in unknown noise fields with the onboard antenna array, arising from the desire to better exploit the spatial diversity in the harsh WSN deployment environments for various network-level benefits. We present an optimal algorithm based on the maximum likelihood (ML) criteria, and computed using particle swarm optimization (PSO) for accurate and fast direction estimation. The ML criterion function is derived using parameterization of noise covariance, and the PSO is incorporated with newly introduced features and properly selected parameters to enhance its convergence. Simulation results demonstrate that the proposed algorithm produces excellent bearing estimates, even in unfavorable scenarios involving few antenna elements, low signal-to-noise ratios and short data samples, which are the typical WSN working conditions due to the power, space and cost constraints.