Detecting Node Failures in Mobile Wireless Networks: A Probabilistic Approach

Abstract

The detection of contract failures in mobile wireless networks is a major challenge because network topology can be very dynamic, the network may not always be connected, and resources are limited. In this paper, we take a probabilistic approach and propose two contract error detection systems that systematically combine local monitoring, site assessment and contract collaboration. The large-scale simulation results in both connected and non-connected networks show that our plans achieve high maximum detection rates (close to the upper limit), low false positive rates, and low communication costs. Compared to approaches that use centralized monitoring, our approach has up to 80% reduction in communication, slightly lower detection rates and slightly higher false positive rates. In addition, our approach has the advantage that it is applicable to both connected and offline networks while centralized monitoring applies only to connected networks. Compared to other approaches that use local monitoring, our approach has similar detection rates, up to 57% reduction in connection level and much lower false positive rates (eg, 0.01 versus 0.27 in some settings).