Robust Localization in OFDM-Based Massive MIMO through Phase Offset Calibration

TL;DR

This work tackles robust localization in OFDM-based massive MIMO under phase incoherence caused by frequency-dependent and antenna-dependent offsets. It develops an analytical framework combining the Cramér-Rao Lower Bound (CRLB) and a Spatial Ambiguity Function (SAF) model to quantify how phase offsets degrade localization, and defines a PMSR metric to characterize ambiguity patterns. A practical CSI calibration framework is proposed and validated with real-world measurements on a 64-antenna testbed, showing substantial gains in coherence and precision. Specifically, localization RMSE improves from about $5\,\mathrm{m}$ in uncalibrated data to about $1.2\,\mathrm{cm}$ after calibration, aligning with CRLB and SAF insights. The results highlight the critical role of joint frequency and spatial synchronization for centimeter-level localization in massive MIMO-OFDM systems.

Abstract

Accurate localization in Orthogonal Frequency Division Multiplexing (OFDM)-based massive Multiple-Input Multiple-Output (MIMO) systems depends critically on phase coherence across subcarriers and antennas. However, practical systems suffer from frequency-dependent and (spatial) antenna-dependent phase offsets, degrading localization accuracy. This paper analytically studies the impact of phase incoherence on localization performance under a static User Equipment (UE) and Line-of-Sight (LoS) scenario. We use two complementary tools. First, we derive the Cramér-Rao Lower Bound (CRLB) to quantify the theoretical limits under phase offsets. Then, we develop a Spatial Ambiguity Function (SAF)-based model to characterize ambiguity patterns. Simulation results reveal that spatial phase offsets severely degrade localization performance, while frequency phase offsets have a minor effect in the considered system configuration. To address this, we propose a robust Channel State Information (CSI) calibration framework and validate it using real-world measurements from a practical massive MIMO testbed. The experimental results confirm that the proposed calibration framework significantly improves the localization Root Mean Squared Error (RMSE) from 5 m to 1.2 cm, aligning well with the theoretical predictions.

Figures (9)

• Figure 1: Considered scenario: A static transmitting signals with $K$ subcarriers to a with $N$ receiver antennas.
• Figure 2: Distributions of the frequency and spatial phase offsets from the measured data: (a) Frequency phase offset: Gaussian distribution. (b) Spatial phase offset: uniform distribution.
• Figure 3: -based localization versus phase incoherence: (a) Frequency phase offset: minor impact on localization accuracy. (b) Spatial phase offset: severe degradation of localization accuracy.
• Figure 4: of the simulated ideal signal with $(x_{\text{true}}, y_{\text{true}})=(-2, 1)$: (a) ; (b) X- and Y-cuts of the at the sensed location. = 29.45dB.
• Figure 5: of the simulated signal with frequency phase offset ($\sigma_\phi^f=\frac{\pi}{4}$): (a) ; (b) X- and Y-cuts of the at the sensed location. = 28.69dB.