Indoor Localization In Wireless Sensor Networks

Indoor Localization in Wireless Sensor Networks

This thesis is dedicated to solve the localization problem in mobile wireless sensor networks. It works mainly with fingerprints features and inertial movements information. The former tackles the RSSIs values between sensors while the latter deals with the objets movement attitude by using accelerometer and gyroscope. The combination of both information is performed in terms of interval analysis, or Kalman filtering. The proposed work introduces three orders mobility models to approximate nodes trajectories using accelerations, combined then to the weighted K nearest neighbors algorithm in a centralized scheme. Then the mobility models are extended up to the inertial information taking into consideration the rotations of the nodes. A decentralized localization method is also proposed in the following in view of the working mechanism of large scale sensor networks. Finally, this thesis proposes a zoning localization method aiming at determining the zones in which the nodes reside. The proposed method addresses the zoning problem by using both the belief functions theory and the interval analysis.

Indoor Localization in Wireless Sensor Networks

We considered the issue of indoor localization through the use of wireless sensor networks (WSN). We value not necessary the algorithm that provides the best accuracy but the one that provides a good enough level of accuracy in a simple and efficient manner. In the first part of our work we examined some state-of-the-art localization techniques that are deployable on wireless sensor motes. These techniques are evaluated to a set of criteria that an indoor WSN-based localization application must consider. In our investigation, we considered not only accuracy but many other factors that determine a suitable indoor WSN-based localization system. These factors among other things determine energy efficiency. We broadly separate the criteria list into two categories: efficiency-based and accuracy-based. We discovered that one of the techniques, Ecolocation, evaluates quite well to the efficiency-based criteria, but the evaluation of the accuracy-based criterions is not as promising. However, the inherent simplicity and potential for good performance (based on open environment results) make the algorithm quite attractive. We proceeded to modify the algorithm to improve its accuracy while maintaining its positive qualities. Our Weighted-Constraints algorithm, named as such due to the nature of the modification, performs in terms of average error 13.1% better than the original Ecolocation algorithm in an open environment. Furthermore, our modified algorithm shows it is more robust to noise compared to the original algorithm by perform on average 21.2% better in a noisy environment.

Indoor Localization Algorithm for Wireless Sensor Networks