Elastic Localization: Improvements on Distributed, Range Free Localization for Wireless Sensor Networks

Numerous wireless sensor network algorithms assume that individual sensors possess location information. However, state of the art localization algorithms often achieve acceptable performance only under restrictive assumptions. For instance, some algorithms necessitate regular sensor deployment or centralized computations. Other algorithms require a high proportion of position aware nodes or the ability to accurately infer emission distance or emission direction of received radio signals. We propose the Elastic Localization Algorithm (ELA), a distributed, scalable, robust and efficient localization algorithm. ELA only presumes that a few percent of the sensors know their location and that an estimation of the maximum communication range is available. We provide extensive simulation data describing the precision, the convergence speed, and the communication load of ELA, using networks composed of thousands of sensors. In addition, we submit ELA to testing considering the influence of maximum range and beacon position misestimation, irregular radio patterns, asynchronous nodes, packet losses, particular topologies, and sensor mobility. Our results indicate that ELA performs better than four state of the art, distributed, range-free localization algorithms, in terms of accuracy. Its low communication and execution costs make its implementation possible on resource constrained sensor nodes. The key advantage of ELA is its robustness to phenomena occurring in real wireless sensor networks, such as asymmetry, packet loss, and irregular radio ranges.