Towards location-awareness in next generation wireless networks : a new approach based on channel state information

Abstract

Location-awareness in next generation wireless networks will be a key enabler for numerous emerging applications. Recently, a novel approach to localization based on soft information (SI), exploiting all positional information inherent in measurement and contextual data, has been proposed. This thesis further develops SI-based localization by establishing a new approach relying on channel state information (CSI) measurements. In particular, we design an efficient joint message-passing (MP) localization algorithm, which consists of two layers: the transformation layer and the estimation layer. The transformation layer extracts SI of the channel impulse response (CIR) from CSI measurements using a sparsity promoting prior model, which addresses the difficulty of unknown number of multipath in estimating the CIR. The estimation layer infers node positions based on the SI of the CIR using a delay-origin uncertainty model, which describes the conditional distribution of the delays in the CIR given node positions. Simulation results using QuaDriGa channel simulator show that our localization algorithm achieves decimeter-level localization accuracy for both Wi-Fi and mmWave signals, which outperforms conventional algorithms.