Automated Detection of Rule Conflicts for Enterprise IoT
Mari Abe, Gaku Yamamoto, Sanehiro Furuichi, Kazuhito Akiyama · 2019
In this paper, we propose a method for automated detection of rule conflicts for enterprise IoT that improves quality of rules. The detection process starts with extracting the IoT context descriptor from the runtime-specific rules to normalize conditions and actions. Then the optimization step is conducted on the normalized model so that the rule conflicts can be detected and duplications of rules can be eliminated. We conducted an experiment to evaluate the feasibility of our approach in a connected vehicle service. In a preliminary experiment, we showed how to detect rule conflicts with our sample rules. We then confirmed that although the two real applications did not include rule conflicts, our algorithm could reduce the number of rules from 92 to 60, which had a significant impact on the performance of the system.