Sensor array processing : localisation of wireless sources

• Main Author: Fang, Zexi
• Other Authors: Manikas, Athanassios
• Published: Imperial College London 2017
• Subjects: 621.3
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.745291

In this thesis, various subspace array processing techniques for wireless source localisation are presented and investigated in the following three aspects. First, in the environment of indoor optical wireless communications, the paths of different sources and/or from different reflectors may impinge on the receiver from closely spaced directions with a high probability. In this case, the ranges of the paths, together with their directions, are important especially for isolating the desired source from the interferers. A blind multi-source localisation approach, which can be used as a channel estimator in the receiver of a communication system, is proposed for direction, range, and path gain estimation. Utilising the above channel parameter estimates, two subspace multibeam beamformers are also presented to achieve complete interference cancellation. Second, in applications such as wireless sensor networks and ubiquitous computing, both the location and orientation of an array are important parameters of interest to be estimated. Hence, array localisation and orientation estimation approaches are proposed for two cases. In the first case, a number of sources of known locations are employed to estimate these parameters of a receiver array. In the second case, a receiver array is utilised to estimate these parameters of multiple sources with each one being a transmitter array. Last, when sources operate in the near field of an array, the spherical wave propagation model needs to be considered. A problem associated with such a scenario is source localisation under the wideband assumption, where the wavefront of a baseband signal varies when traversing through the sensors of the array. Two novel approaches with the employment of the subcovariance of the received signal and the rotation of the array reference point are proposed to localise multiple sources under the wideband assumption. Throughout the thesis, computer simulation studies are presented for evaluating the performance of the proposed approaches.