Geometric Constraint Based Range Free Localization Scheme For Wireless Sensor Networks (WSNs)

Abstract

Localization of the wireless sensor networks (WSNs) is an emerging area of research. The accurate localization is essential to support extended network lifetime, better covering, geographical routing, and congested free network. In this thesis, we proposed four distributed range-free localization schemes. The proposed schemes are based on the analytical geometry, where an arc is used as the geometric primitive shape. The simulation and experimental validation are performed to evaluate the performance of the proposed schemes. First, we have proposed a mobile beacon based range-free localization scheme (MBBRFLS). The proposed scheme resolved the two underlying problems of the constraint area based localization: (i) localization accuracy depends on the size of the constraint area, and (2) the localization using the constraint area averaging. In this scheme, the constraint area is used to derive the geometric property of an arc. The localization begins with an approximation of the arc parameters. Later, the approximated parameters are used to generate the chords. The perpendicular bisector of the chords estimate the candidate positions of the sensor node. The valid position of the sensor node is identified using the logarithmic path loss model. The performance of proposed scheme is compared with Ssu and Galstyan schemes. From the results, it is observed that the proposed scheme at varying DOI shows 20.7% and 11.6% less localization error than Ssu and Galstyan schemes respectively. Similarly, at the varying beacon broadcasting interval the proposed scheme shows 18.8% and 8.3% less localization error than Ssu and Galstyan schemes respectively. Besides, at the varying communication range, the proposed scheme shows 18% and 9.2% less localization error than Ssu and Galstyan schemes respectively. To further enhance the localization accuracy, we have proposed MBBRFLS using an optimized beacon points selection (OBPS). In MBBRFLS-OBPS, the optimized beacon points minimized the constraint area of the sensor node. Later, the reduced constraint area is used to differentiate the valid or invalid estimated positions of the sensor node. In this scheme, we have only considered the sagitta of a minor arc for generating the chords. Therefore, the complexity of geometric calculations in MBBRFLS-OBPS is lesser than MBBRFLS. For localization, the MBBRFLS-OBPS use the perpendicular bisector of the chords (corresponding to the sagitta of minor arc) and the approximated radius. The performance of the proposed MBBRFLS-OBPS is compared with Ssu, Galstyan, and Singh schemes. From the results, it is observed that the proposed scheme using CIRCLE, vii SPIRAL, HILBERT, and S-CURVE trajectories shows 74.68%, 78.3%, 73.9%, and 70.3% less localization error than Ssu, Galstyan, and Singh schemes respectively. Next, we have proposed MBBRFLS using an optimized residence area formation (ORAF). The proposed MBBRFLS-ORAF further improves the localization accuracy. In this scheme, we have used the adaptive mechanism corresponding to the different size of the constraint area. The adaptive mechanism defines the number of random points required for the different size of the constraint area. In this scheme, we have improved the approximation accuracy of the arc parameters even at the larger size of the constraint area. Therefore, the localization accuracy is improved. The previous scheme MBBRFLS-OBPS use the residence area of the two beacon points for approximation. Therefore, the larger size of the constraint area degrades the approximation accuracy. In the MBBRFLS-ORAF, we have considered the residence area of the three non-collinear beacon points, which further improves the localization accuracy. The performance of the proposed scheme is compared with Ssu, Lee, Xiao, and Singh schemes. From the results, it is observed that the proposed MBBRFLS-ORAF at varying communication range shows 73.2%, 48.7%, 33.2%, and 20.7% less localization error than Ssu, Lee, Xiao, and Singh schemes respectively. Similarly, at the different beacon broadcasting intervals the proposed MBBRFLS-ORAF shows 75%, 53%, 38%, and 25% less localization error than Ssu, Lee, Xiao, and Singh schemes respectively. Besides, at the varying DOI the proposed MBBRFLS-ORAF shows 76.3%, 56.8%, 52%, and 35% less localization error than Ssu, Lee, Xiao, and Singh schemes respectively. Finally, we have proposed a localization scheme for unpredictable radio environment (LSURE). In this work, we have focused on the radio propagation irregularity and its impact on the localization accuracy. The most of the geometric constraint-based localization schemes suffer from the radio propagation irregularity. To demonstrate its impact, we have designed an experimental testbed for the real indoor environment. In the experimental testbed, the three static anchor nodes assist a sensor node to perform its localization. The impact of radio propagation irregularity is represented on the constraint areas of the sensor node. The communication range (estimated distance) of the anchor node is derived using the logarithmic regression model of RSSI-distance relationship. The additional error in the estimated distances, and the different placement of the anchor nodes generates the different size of the constraint areas. To improve the localization accuracy, we have used the dynamic circle expansion technique. The performance of the proposed LSURE is compared with APIT and Weighted Centroid schemes using the various deployment scenarios of the anchor nodes. From the results, it is observed that the proposed LSURE at different deployment scenarios of anchor nodes shows 65.94% and 73.54% less localization error than APIT and Weighted Centroid schemes.