EN FR
EN FR


Section: New Results

Localization

Participants : Ibrahim Amadou, Roudy Dagher, Nathalie Mitton, Roberto Quilez, Nicola Zema.

Navigate in or based on a wireless sensor network present many advantages but it is still an open issue. We have focused on two particular cases in which navigation or WSN-based localization is needed [32] . The former aspect considers that sensors need to be visited on-demand by a mobile sink to offload data. This mobile sink thus needs to locate the data source. The second aspect feature a mobile entity that is needed to be localized.

In a event-based WSN, where is necessary a prompt response in terms of data processing and offloading, a set of mobile flying sinks could be a good option for the role of autonomous data collectors. For those reasons in [28] , we propose a distributed algorithm to independently and autonomously drive a mobile sink through the nodes of a WSN and we show its preferability over more classical routing approaches especially in the presence of a localized generation of large amount of information. Our result shows that, in the case of fairly complete coverage of the area where the nodes lie, it is possible to promptly notify a mobile sink about the presence of data to offload, drive it to the interested area and achieve interesting performances. [29] enhanced the previous approach by relaxing some GPS-use assumptions. We show that, under fairly common circumstances, it is possible to set the trajectory of the mobile sink and fulfill the offloading requests without the needs of additional equipment installed on nodes. We show how our system is preferable over more classical routing solutions especially in the presence of localized generation of large amounts of information.

[11] proposes Ubiquitous Navigation System (UNS), a WSN-based navigation system, which takes benefit from a WSN mesh deployment to provide a local navigation service. The positioning part of the system uses Angle of Arrival (AoA) measurements to estimate the vehicle position on the map. Based on a realistic network scenario, extracted from a city map using Google Maps, we study the performance of Triangulation using AoA in a smart urban environment that exhibits topology related constraints. Simulations results show that such constraints lead to particular spatial distribution of the anchor nodes that affects both positioning accuracy and beacon packets reception rate. We also propose and evaluate the use of the network communication range as a technique to mitigate the effect of geometric dilution of precision (GDOP). The simulation results show that this technique successfully detected GDOP-affected positions and thus significantly enhanced the positioning accuracy. One of the biggest strengths of UNS is that it relies on a single anchor unlike literature approaches. The different underlying studies are detailed in [38] in which we study the ambiguity of source localization using signal processing of large aperture antenna arrays under spherical wave propagation. This novel localization approach has been recently proposed, providing an estimate of the source position by means of two methods: geometrical and analytical. The former finds the source position as the estimate of circular loci, the latter as a solution of a linear system of equations. Although this method is proved to work for a general array geometry, we show that it suffers from ambiguities for a particular class of array geometries. Namely, in 2D, we prove that when the array geometry is linear or circular, there exist two possible solutions where only one corresponds to the actual position of the source. We also prove a relation of symmetry between the solutions with respect to the array geometry. This relation is very useful to assist the disambiguation process for discounting one of the estimates. By extension to 3D, planar (resp. spherical) arrays exhibit the same behavior i.e they provide two symmetrical estimates of the source position when the latter is not on the array plane (resp. sphere).

Note that UNS is currently a pending patent.