Abstract
In ubiquitous environment, wireless sensor networks are increasingly being deployed to collect information from real world. For sharing huge amount of sensor data efficiently with diverse users, an information dissemination (INFOD) system is very important. In our previous work, we proposed and INFOD system integrated with sensor network and Data Grid as a publish/subscribe system for a ubiquitous-Healthcare application. To disseminate sensed data to diverse users, a matching algorithm called Classed Group Index Matching (CGIM) was proposed in our previous work to match published events with subscriptions efficiently. CGIM used statistical grouping method with equivalent intervals for each group which showed poor performance in case of non-uniform subscription distributions. We need to decide proper number of groups to improve the performance. In this paper, we propose efficiency based dynamic re-grouping (DRG) method for CGIM to improve its performance for our INFOD system. Experimental results clearly demonstrate the performance improvement of CGIM algorithm.