Abstract
In this paper, we focus on wireless sensor network-based localization for user devices (UDs). Prior to UD localization, signal propagation model (SPM) is required to calibrate using training samples from a number of location grids. However, SPM usually suffers from measurement noise and inevitable error in calibration grid (CG) locations. This will significantly degrade UD localization performance. Nevertheless, the impact of CG location error, CG layout and the number of CGs on SPM calibration performance has not been characterized. Furthermore, the effect of SPM calibration error on UD localization performance has not been developed. In this paper, we aim to provide a unified framework for performance analysis of SPM calibration and UD localization. Firstly, we establish a closed-form Cramér-Rao lower bound on SPM calibration error and UD localization error, respectively. Secondly, the impact of measurement noise, CG location error and the number of CGs on SPM calibration performance is revealed. Thirdly, the influence of SPM calibration error, CG location error and measurement noise on UD localization performance is studied. The effect of modeling mismatch is also studied. The obtained analysis framework builds a theoretical basis for the design of efficient system optimization strategies, including resource allocation and CG deployment optimization, for UD localization performance enhancement.
| Original language | English |
|---|---|
| Pages (from-to) | 3254-3268 |
| Journal | IEEE Transactions on Wireless Communications |
| Volume | 20 |
| Issue number | 5 |
| Online published | 12 Jan 2021 |
| DOIs | |
| Publication status | Published - May 2021 |
Research Keywords
- Calibration
- Cramer-Rao lower bound
- internet-of-things localization
- Location awareness
- Loss measurement
- Measurement uncertainty
- Path loss model calibration
- performance limit
- Training
- Wireless communication
- Wireless sensor networks