Section: New Results
Localization
Participants : Nathalie Mitton, Roudy Dagher, Valeria Loscri, Salvatore Guzzo Bonifacio.
[20] presents our approach to localize a node with the use of only one landmark. It is a passive and non intrusive cross-layer approach that relies on a signal processing of all received signals Results are evaluated by simulation and show good accuracy. To complete the previous study, we developed [11] a novel array-based method to estimate the path loss exponent (PLE). The method is designed as a part of an automatic calibration step, prior to localization of a source transmitting in the near-far field of the array. The method only requires the knowledge of the ranges between the array elements. By making the antenna elements transmit in turn, the array response model in the near-far field is exploited to estimate the current environment PLE. Simulation results show that this method can achieve good performance with one transmission round. The performance of the PLE estimation is investigated in the context of source localization with a sensitivity analysis to the PLE estimation. These works are the purpose of a pending patent (submitted in March 2015).
Alternatively, we derive similar localization schemes to enable a cooperation between mobile robots to locale a target based on RSSI [13] . Received Signal Strength Indicator (RSSI) is commonly considered and is very popular for target localization applications, since it does not require extra-circuitry and is always available on current devices. Unfortunately, target localizations based on RSSI are affected with many issues, above all in indoor environments. In this paper, we focus on the pervasive localization of target objects in an unknown environment. In order to accomplish the localization task, we implement an Associative Search Network (ASN) on the robots and we deploy a real test-bed to evaluate the effectiveness of the ASN for target localization. The ASN is based on the computation of weights, to "dictate" the correct direction of movement, closer to the target. Results show that RSSI through an ASN is effective to localize a target, since there is an implicit mechanism of correction, deriving from the learning approach implemented in the ASN.