In this paper we propose using algorithm aggregation to improve signal strength based localization performance. Prior work comparing a spectrum of received signal strength (RSS) based localization algorithms concluded that their performance is strikingly similar and none of them manage to localize objects accurately all the time. We, however, show that despite such similarity in average localization accuracy, simple algorithm aggregation can improve performance. Specifically, comparing with point-based algorithms, the performance improvement ranges from 23% (when aggregating 2 algorithms) to 51% (when aggregating all 12 algorithms we consider); comparing with area- based algorithms, aggregation offers comparable accuracy with much higher and directly controllable precision. As a guideline for the practical use of aggregation, our experimentation shows that aggregating any 3 or 4 algorithms out of our 12 algorithms will be able to achieve a good performance gain.
Title
Simple algorithm aggregation improves signal strength based localization
Li, X. and R. P. Martin (7-9 May 2008) "Simple algorithm aggregation improves signal strength based localization." Proceedings of the International Symposium on Wireless Pervasive Computing, 540-544.