Joint LOS Identification and Data Association for 6G-Enabled Networked Device-Free Sensing

TL;DR

Numerical results verify that the proposed two-phase protocol can achieve high performance of networked sensing in the multipath environment.

Abstract

This paper considers networked device-free sensing in an orthogonal frequency division multiplexing (OFDM) cellular system with multipath environment, where the passive targets reflect the downlink signals to the base stations (BSs) via non-line-of-sight (NLOS) paths and/or line-of-sight (LOS) paths, and the BSs share the sensing information extracted from their received echoes to jointly localize the targets. A two-phase localization protocol is considered. In Phase I, we design an efficient method that is able to accurately estimate the range of any path from a transmitting BS to a receiving BS via a target, even if the transmitting and receiving BSs are separated and not perfectly synchronized. In Phase II, we propose an effective method that is able to jointly identify the ranges of the LOS paths between the targets and the BSs as well as associate the ranges of LOS paths with the right targets, such that the number and the locations of the targets both can be accurately estimated. Numerical results verify that our proposed two-phase protocol can achieve high performance of networked sensing in the multipath environment.

Figures (7)

• Figure 1: System model of our considered networked device-free sensing architecture. The BSs are connected to the central processor via fronthaul links to share the range estimations. For wireless propagation, there are Type I paths (e.g., the path from BS 1 to BS 3), Type II paths (e.g., the path from BS 1 to Target 1 to BS 2), Type III paths (e.g., the path from BS 2 to the building to Target 1 back to BS 3), and LOS blockage (Target 2 cannot be detected by BS 1 due to the obstacle).
• Figure 2: Range estimation error probability versus the number of targets.
• Figure 3: The comparison of location estimation error probability between the benchmark and the proposed algorithm with different $r$ when $B=400$ MHz, $P_{\rm b}=0.1$, and $P_{\rm nl}=0.5$.
• Figure 4: The comparison of the average running time for the benchmarks and the proposed algorithm when $B=400$ MHz, $P_{\rm b}=0.1$, and $P_{\rm nl}=0.5$.
• Figure 5: The cardinalities of $\mathcal{G}^{(l)}$, $\bar{\mathcal{G}}^{(l)}$, and $\tilde{\mathcal{G}}^{(l)}$ for the proposed algorithm when $B=400$ MHz, $P_{\rm b}=0.1$, and $P_{\rm nl}=0.5$.

Theorems & Definitions (1)

• Remark 1