Wideband Cooperative Localization Through Generalized Cross-Correlation

Abstract

This work proposes a novel algorithm for robust cooperative localization using wideband time difference of arrival (TDoA) measurements and information coupling between nodes. The algorithm leverages a generalized cross-correlation framework to exploit the similarity of wideband measurements within the geometric vicinity of nodes. Alternatively, the algorithm uses a computationally efficient method based on perceptual hashing, a signal compression algorithm adapted from image-processing, which gives a distance-dependent metric between nodes. The algorithm is embedded within a message-passing framework, enabling the estimation of the joint-posterior probability density function of all the nodes. The paper evaluates the localization performance, using both simulated and real measurements in various challenging scenarios, and compares it with a cooperative localization method that uses additional received signal strength measurements inbetween nodes. The results demonstrate that the proposed algorithm can achieve similar or superior localization accuracy, with the same computational complexity and without the need for additional measurements. The proposed algorithm shows significant improvement over non-cooperative measurements. It achieves an improvement of root-mean-square error from 2.4 m to 1.3 m for a large-scale scenario with approximately 1000 nodes, using a signal bandwidth of 80 MHz for the TDoA measurements and six fixed, dual-antenna, anchor nodes.

Keywords

• Indoor localization
• wideband
• generalized cross-correlation
• cooperative localization
• information coupling
• dense multipath component