CSI-Based Multi-Antenna and Multi-Point Indoor Positioning Using Probability Fusion

Abstract

Channel state information (CSI)-based fingerprinting via neural networks (NNs) is a promising approach to enable accurate indoor and outdoor positioning of user equipments (UEs), even under challenging propagation conditions. In this paper, we propose a positioning pipeline for wireless LAN MIMO-OFDM systems which uses uplink CSI measurements obtained from one or more unsynchronized access points (APs). For each AP receiver, novel features are first extracted from the CSI that are robust to system impairments arising in real-world transceivers. These features are the inputs to a NN that extracts a probability map indicating the likelihood of a UE being at a given grid point. The NN output is then fused across multiple APs to provide a final position estimate. We provide experimental results with realworld indoor measurements under line-of-sight (LoS) and non-LoS propagation conditions for an 80 MHz bandwidth IEEE 802.11ac system using a two-antenna transmit UE and two AP receivers each with four antennas. Our approach is shown to achieve centimeter-level median distance error, an order of magnitude improvement over a conventional baseline.

Feature Learning for Neural-Network-Based Positioning with Channel State Information

“…Feature extraction also serves the purpose of preparing the data for subsequent DNN processing [14]. While such architectures are widely used in state-of-the-art CSIbased positioning pipelines [8], [13], [14], [16], [17], [22]- [25], such hard-coded CSI feature extraction stages are unable to exploit the learning capabilities of DNNs.…”

“…The dataset is generated with a robot that follows a random path in a 3.2×4.2 m 2 area under LoS conditions, where we first record the training set and then a test set that have different UE locations. This new dataset is more challenging than the two other datasets from [16] as the test set contains many locations that were not previously fingerprinted in the training set. We obtain CSI measurements for W = 234 subcarriers and at M R = 4 receive antennas, and we use a VICON [1] precision positioning system to collect the groundtruth location information.…”

Improving Channel Charting using a Split Triplet Loss and an Inertial Regularizer

“…Channel charting is a self-supervised pseudo-localization framework that applies dimensionality reduction to measured CSI [6]. In brief, channel charting takes a dataset of measured CSI, converts the CSI measurements into CSI features that are resilient to small-scale fading and hardware impairments [9], [11], and applies dimensionality reduction in order to learn a low-dimensional embedding: the channel chart. Our goal is to learn a channel chart that preserves local geometry: two nearby charted points should also be close in real space.…”

“…In particular, wireless channels are subject to (i) system impairments, mostly small-scale fading caused by moving objects between the transmitter and receiver, and (ii) hardware impairments, caused by varying transmit power, timing synchronization errors, as well as residual carrier frequency and sampling rate offsets. Following the work of [9], [11], we use an autocorrelationbased strategy to extract a feature from a raw CSI matrix H i .…”

“…, where * is the discretetime convolution operator. This step renders our CSI features independent of time synchronization offset and global phase changes [11]. Note that the 2D autocorrelation of a matrix is rotationally symmetric, so we can truncate the redundant second half of the rows of H acorr .…”