| Content : Localization is one of the most intrinsic issues in Wireless Sensor Networks. Many protocols and services are based on the provided localization information. As a comprehensive issue, many algorithms have been suggested in recent years. Though holding the graceful flexibility, the range-free localization algorithms may fail to reach the desired accuracy in the situation of low density of network or anchors. The high cost of calibration device hinders the large-scale and low-cost applications of range-based algorithms with better accuracy.In this paper, the ontological description of localization, and the taxonomical analyses including concept, classification, evaluation schemes, and basic mathematical tools and so on are made. Some recent in-depth subjects, in particular the Cramer-Rao bound, position forwarding, and random walk, are also discussed. Based on those analyses, the hybrid algorithm HILS with high accuracy is therefore proposed, with the utility of nodes with and without calibration capability. The "Hybrid" characteristic of HILS holds in two aspects: being able to take advantage of both the range-based information and range-free information, e.g. connectivity and shortest path, and calculating the initial coarse coordinates, e.g. using DV-Hop, to utilize the error-controllable localization algorithm in heterogeneous networks. Results show that compared with classical range-free algorithm, the proposed algorithm can mitigate the local minima problem in the general iterative algorithms. Though the time consumption referred to the parameter of desired accuracy, it can be reduced by exchanging the ontology description of constrains between nodes to rank the order to localize, and by using the optional initial estimation with some lightweight range-free algorithm. In sum, simulation represents that HILS can achieve the ideal accuracy and can solve the problem of uncontrollable error in range-free algorithm, as well as be applied to the distributed situation. |
PDF Full Text Request