Joint AOA and TOF Estimation for Multipath OFDM Signals Based on MSCNN

Abstract

Indoor localization has aroused great interest due to its wide range of application scenarios such as home automation and emergency rescue. Among many localization technologies, the positioning methods based on the angle of arrival (AOA) and time of flight (TOF) have always been researched hotspots due to high real-time performance. However, the multipath effect seriously affects the positioning accuracy. To accurately estimate the AOA and TOF of the multipath OFDM signal, this paper proposes a Multi-scale Convolutional Neural Network (MSCNN) with a multi-scale convolution kernel. In the data preprocessing stage, the real and imaginary parts of the CSI matrix are reconstructed according to the Hadamard matrix composition. Then, a cyclic window interception method is proposed to generate multiple sub-matrices, then generate a multidimensional matrix from these sub-matrices. The information of AOA and TOF is contained in every sub-carrier. This matrix provides different sub-carrier combinations. AOA and TOF features can be extracted from these different sub-carrier combinations by convolution kernels of different scales in MSCNN. The simulation results show that the estimation accuracy of this method is higher than that of the traditional method.