Evaluation of Hybrid Distributed Least Squares for Improved Localization via Algorithm Fusion in Wireless Sensor Networks

Abstract

Wireless Sensor Networks (WSNs) have been a growing research domain during the past couple of years. One of the most challenging tasks of WSN research is still location estimation. As a well performing fine grained localization approach, Distributed Least Squares (DLS) was introduced, splitting the localization process in a complex global precalculation and a simple local postcalculation. Nevertheless, as size of precalculation and cost of computation and communication are increasing with the WSN dimensions, it was shown that this algorithm is unsuitable for large ones. This constraint has been overcome by scalable DLS (sDLS). Further, the computational costs of sDLS have been improved by using sDLSne. Unfortunately, sDLSne comes along with decreased localization accuracy and thus represents a tradeoff. The presented hybrid solution combines sDLSne with various coarse grained localization techniques to avoid this drawback. The resulting localization accuracy overcomes the efficient sDLSne approach as well as the more precise sDLS approach, while computational costs still outperforms sDLS.

Keywords

• Wireless sensor networks
• Localization
• Scalability