Mauricio A. Caceres

Hybrid Cooperative Positioning Based on Distributed Belief Propagation

We propose a novel cooperative positioning algorithm that fuses information from satellites and terrestrial wireless systems, suitable for GPS-challenged scenarios. The algorithm is fully distributed over an unstructured network, does not require a fusion center, does not rely on fixed terrestrial infrastructure, and is thus suitable for ad-hoc deployment. The proposed message passing algorithm, named hybrid sum-product algorithm over a wireless network (H-SPAWN), is described and analyzed. A novel parametric message representation is introduced, to reduce computational and communication overhead. Through simulation, we show that H-SPAWN improves positioning availability and accuracy, and outperforms hybrid positioning algorithms based on conventional estimation techniques.

Cramér-Rao Bound for Hybrid GNSS-Terrestrial Cooperative Positioning

In this contribution we derive an expression of the Cramér-Rao bound for hybrid cooperative positioning, where GNSS information is combined with terrestrial range measurements through exchange of peer-to-peer messages. These results provide a theoretical characterization of achievable performance of hybrid positioning schemes, as well as allow to identify critical network configurations and devise optimized node placement strategies.

Hybrid GNSS-Terrestrial Cooperative Positioning Based on Particle Filter

We propose a novel hybrid GNSS-terrestrial localization algorithm based on particle filter that fuses ranging data from both satellites and terrestrial receivers. The proposed positioning approach, named hybrid-cooperative particle filter (HCPF), is fully distributed and allows both increased positioning availability and accuracy compared to GNSS-only localization in challenged scenarios. Moreover, simulation results based on a realistic indoor scenario show that the proposed solution outperforms several state of the art algorithms such as unscented Kalman filter and an approach based on belief propagation.

Hybrid GNSS-ToA localization and tracking via cooperative unscented Kalman filter

Cooperative localization algorithms have been recently introduced to overcome the limitations of systems relying on GPS or other terrestrial infrastructure. The novel hybrid cooperative unscented Kalman filter (hcUKF) approach, which fuses ranging data from both satellites and terrestrial receivers, allows increased availability, robustness and accuracy compared to GNSS-only localization in challenged scenarios. Simulation results show that the proposed solution outperforms traditional algorithms such as extended Kalman filter.

Adaptive Location Tracking by Kalman Filter in Wireless Sensor Networks

The increasing availability of ubiquitous, small, low-cost devices using wireless communications to build wireless sensor networks calls for autonomous solutions and algorithms capable of calculating the location where the information is gathered, processed, used. The requirements are particularly strict when sensor nodes are mobile; in fact, mobile applications demand more accurate locating and real time tracking, with limited impact on hardware complexity, network load and latency. Despite of remarkable research efforts put into this field by the scientific community, a common unique solution has not been adopted yet, due to the great variety of scenarios and application requirements. This paper focuses on the design and performance evaluation of Kalman filters for tracking a mobile target moving at low dynamics and for smoothing range estimations from noisy measurements. Some adaptive techniques to self-tune the filter and estimate the propagation model parameters are presented. One among the most promising adaptive algorithm presented is implemented in a real-life wireless sensor network test-bed including a mobile node. The resulting experimental results illustrate the benefits of this approach with respect to traditional multilateration based on least mean squares estimators.

WLAN-Based Real Time Vehicle Locating System

This paper presents a wireless local area network (WLAN)-based real time system for indoors and outdoors ve- hicle localization. The proposed solution uses a neural network trained with a map of received power fingerprints from WLAN Access Points (APs) surrounding the vehicle. The practical implementation of the system is described and results from an outdoor experimental testbed are analyzed to address system performance in terms of calibration, estimation complexity, and location accuracy. Pre- and post-processing approaches aimed at improving system accuracy are also discussed. Index Terms—positioning, neural network, fingerprint, real time locating system, wireless local area network ∗

Hybrid GNSS-Terrestrial Cooperative Positioning via Distributed Belief Propagation

Cooperative positioning algorithms have been recently introduced to overcome the limitations of traditional methods, relying on GNSS or other terrestrial infrastructure. In particular, SPAWN (Sum- Product Algorithm over a Wireless Network) was shown to provide accurate position estimate even in challenged indoor environments, thanks to exchange of local information among peers. In this paper we extend the SPAWN framework by considering a hybrid scenario, where agents combine satellite and peer-to-peer terrestrial measurements. The novel hybrid SPAWN (H-SPAWN) approach allows increased availability and robustness compared to GNSS- only positioning in light and deep indoor scenarios, while keeping the advantages of a distributed implementation of the original SPAWN. A parametric message representation is proposed to reduce the communication overhead, and to improve the estimation accuracy. Simulation results show that the proposed solution outperforms traditional algorithms such as cooperative least squares and the extended Kalman filter.

Hybrid WSN-RFID cooperative positioning based on extended kalman filter

In this paper we propose a novel hybrid and cooperative positioning approach based on extended Kalman filter (EKF) to localize mobile targets in indoors. The algorithm fuses both received signal strength (RSS) measurements performed by nodes of a wireless sensor network (WSN) and proximity information from radio frequency identification (RFID) devices. Simulation results prove that the proposed cooperative approach outperforms the non-cooperative version of the algorithm.

A simulation tool for hybrid-cooperative positioning

We propose a simulation tool able to test hybrid GNSS-terrestrial and cooperative positioning algorithms that fuse both pseudorange measurements from satellites and range estimates between terrestrial receivers. In particular, the tool simulates devices belonging to a peer-to-peer (P2P) network where peers cooperate by exchanging data in order to improve both positioning accuracy and availability. More specifically, the simulation tool generates range and pseudorange measurements by employing realistic propagation models and GPS orbits. Furthermore, the simulation tool allows the user to test different algorithms based on Bayesian and least squares methods in two realistic GNSS challenged scenarios namely indoor and urban canyon. Finally, the simulator take into account communication and mobility aspects allowing the user to evaluate their impact on the final positioning accuracy.

Distributed-weighted multidimensional scaling for hybrid Peer-to-Peer localization

This paper presents a hybrid localization algorithm for a Peer-to-Peer (P2P)-Positioning scenario, where “peers” are devices equipped with both a GNSS (Global Navigation Satellite System) receiver and a wireless communications interface. Through their GNSS receiver, peers obtain pseudorange estimates from satellites while they use their wireless interface both to communicate with each other and to obtain terrestrial range estimates. The proposed algorithm is an extended version of the distributed-weighted multidimensional scaling (dwMDS) [1] algorithm to be compatible with the hybrid P2P scenario where unknown peers infer their own positions by fusing both GNSS pseudorange and terrestrial range estimates. Simulation results show that the proposed hybrid dwMDS (hdwMDS) algorithm out-performs the cooperative version of the least squares algorithm. Moreover, they show the related benefits in term of positioning performance in a cooperative scenario in comparison to a non cooperative one based on GNSS-only positioning.