Mobile Wireless Localization through Cooperation

This paper considers N mobile nodes that move together in the vicinity of each other, whose initial poses as well as subsequent movements must be accurately tracked in real time with the assist of M(>=3) reference nodes. By engaging the neighboring mobile nodes in a simple but effective cooperation, and by exploiting both the time-of-arrival (TOA) information (between mobile nodes and reference nodes) and the received-signal-strength (RSS) information (between mobile nodes), an effective new localization strategy, termed cooperative TOA and RSS (COTAR), is developed. An optimal maximum likelihood detector is first formulated, followed by the derivation of a low-complexity iterative approach that can practically achieve the Cramer-Rao lower bound. Instead of using simplified channel models as in many previous studies, a sophisticated and realistic channel model is used, which can effectively account for the critical fact that the direct path is not necessarily the strongest path. Extensive simulations are conducted in static and mobile settings, and various practical issues and system parameters are evaluated. It is shown that COTAR significantly outperforms the existing strategies, achieving a localization accuracy of only a few tenths of a meter in clear environments and a couple of meters in heavily obstructed environments.

💡 Research Summary

The paper addresses the problem of real‑time tracking of a group of mobile nodes that move together in close proximity. It proposes a cooperative localization scheme called COTAR (Cooperative TOA and RSS) that jointly exploits time‑of‑arrival (TOA) measurements from a set of at least three fixed reference nodes and received‑signal‑strength (RSS) measurements among the mobile nodes themselves.

The authors first formulate a realistic system model. The RSS follows a log‑distance path‑loss law with Gaussian shadowing (standard deviation σ_g = 8 dB, path‑loss exponent η = 3.086). The TOA error is modeled using a two‑ray Rician channel that captures the fact that the direct path may not be the strongest. Two representative environments are considered: clear line‑of‑sight (K‑factor = 5, mean excess delay ≈ 25.8 ns, TOA std ≈ 8.8 ns) and heavily obstructed (K = 2, mean excess delay ≈ 76.9 ns, TOA std ≈ 40.2 ns). These parameters are derived from measurements in real buildings and are used throughout the simulations.

A maximum‑likelihood estimator (MLE) for the joint set of positions is derived. Because the MLE leads to a highly nonlinear optimization problem, the authors propose a low‑complexity iterative algorithm. The algorithm linearizes the RSS‑derived distance constraints around the current estimate, combines them with the TOA equations, and solves a weighted least‑squares problem at each iteration. The Jacobian of the measurement functions is used to update the estimate in a Gauss‑Newton‑like fashion. Convergence is typically achieved within a few iterations, and the resulting root‑mean‑square error (RMSE) closely approaches the Cramér‑Rao lower bound (CRLB) derived analytically for the COTAR system.

The cooperation protocol is simple yet effective. Mobile nodes broadcast beacons sequentially. When node i transmits, all reference nodes record its TOA, while all later‑transmitting mobile nodes measure the RSS from node i. Node i also embeds in its beacon the RSS values it previously measured from earlier nodes, so that each subsequent node receives a growing set of inter‑node RSS data. Consequently, the network collects N TOA measurements (one per node) and N(N‑1)/2 RSS measurements (pairwise distances) without requiring extra hardware or large bandwidth.

Extensive simulations evaluate the impact of several key parameters: number of cooperating nodes N, number of reference nodes M, geometry of the reference anchors, measurement range, node mobility speed, and the presence of missing RSS data. Results show:

• Increasing N from 2 to 6 reduces the average RMSE by roughly 20–30 % because of the added geometric diversity.
• Adding more reference nodes (e.g., from 3 to 5) improves accuracy especially in the obstructed scenario, where the extra anchors help mitigate large TOA errors.
• For short ranges (< 5 m) RSS dominates and the system achieves sub‑0.3 m accuracy; for longer ranges (> 10 m) TOA dominates and the error stays below 1 m in clear LOS and below 2 m in heavy NLOS.
• Even when a fraction of RSS links are unavailable, the cooperative framework still outperforms non‑cooperative baselines, with performance degradation limited to a few percent.
• With node speeds up to 2 m/s the iterative estimator converges within 5–10 ms, satisfying real‑time tracking requirements.

The authors emphasize that COTAR works with inexpensive omnidirectional antennas and narrowband radios (e.g., ZigBee at 2 MHz), making it suitable for GPS‑denied environments such as indoor facilities, forests, or disaster zones. The method requires only a single low‑precision clock per node, keeping hardware cost and power consumption low.

In conclusion, COTAR integrates a realistic multipath channel model, cooperative RSS sharing, and a computationally efficient MLE‑based iterative solver to deliver localization accuracy that is a few tenths of a meter in clear conditions and a couple of meters in heavily obstructed environments—significantly better than traditional TOA‑only, RSS‑only, or naïve hybrid schemes. The paper suggests future work on hardware prototyping, asynchronous beacon scheduling, and the incorporation of machine‑learning‑based channel compensation to further enhance robustness and applicability.