Développement d'un environnement vocal virtuel par communication sans fil et positionnement par puissance du signal reçu en environnement interne

Abstract

Wireless communication allows workers wearing hearing protection devices to communicate in a noisy work environment. However, typical communication systems send voice information to all users on the same channel regardless of their physical proximity. The introduction of a virtual voice environment concept allows to take into account the proximity of each user and decide which users should receive and hear the communication. A virtual voice environment can be obtained using a wireless communication and the knowledge of the position of each user. To retrieve this information, the received signal strength indicator (RSSI) of the communication may be associated with a distance. However, indoor multipath fading affects RSSI measurements by creating large variations in those measurements. In addition, the environment in which the system will be used does not allow the deployment of a fixed infrastructure. Although there are several indoor positioning systems in the literature, they depend on a fixed infrastructure. In order to meet this need, a new solution for indoor positioning was developed. It is based on the average of RSSI measurements at multiple frequencies. It addresses the problem in two stages. The first step is to characterize the environment with a parametric model based on the propagation loss and RSSI measurement variations for a given distance. The second step is the decision algorithm for the positioning. It is based on the probability that a user is within a predetermined area based on the RSSI measurements. The user can be located in three different areas : an area where the voice is being activated, an area where it is deactivated and a hysteresis zone in which it remains unchanged. These are defined by thresholds that must be computed against functionality criterions and the model. This paper presents the results of the proposed solution with a 80 MHz bandwidth in the 2.4 GHz ISM band. The results show a reduction of 2.0 dB in the average standard deviation of RSSI measurements by using the frequency diversity technique. In addition, two different activation distance scenarios were studied, i.e. 0.5 m and 1 m. In the 1 m scenario, the deactivation distance imposed by the model found was 3.0 m. Cross-validation of the decision algorithm gave an average deviation of 0.18 m between the activation distance provided by the model and the decision and an average deviation of 0.225 m for the deactivation distance.

Date	20 Jan 2014
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Ghyslain Gagnon (Supervisor) & Jérémie Voix (Co-supervisor)