Predicting link quality using supervised learning in wireless sensor networks

摘要: Routing protocols in sensor networks maintain information on neighbor states and potentially many other factors order to make informed decisions. Challenges arise both (a) performing accurate adaptive discovery (b) processing/analyzing the gathered data extract useful features correlations. To address such challenges, this paper explores using supervised learning techniques decisions context of wireless networks.We investigate design space offline online use link quality estimation as a case study evaluate their effectiveness. For purpose, we present MetricMap, metric-based collection routing protocol atop MintRoute that derives classifiers learned training phase, when traditional ETX approach fails. The is evaluated 30-node network testbed, our results show MetricMap can achieve up 300% improvement over delivery rate for high situations, with no negative impact performance metrics. We also explore possibility paper.