Detecting Node Collapse in Mobile Wireless Networks: An Investigative Model

Abstract

Detecting node failures in mobile wireless networks poses significant challenges due to the dynamic nature of the network topology, intermittent connectivity, and limited resources. In our research, we address this issue by adopting a probabilistic approach and proposing two innovative node failure detection schemes. These schemes intelligently combine localized monitoring, location estimation, and node collaboration. Our extensive simulations encompass both connected and disconnected networks, demonstrating the effectiveness of our schemes. They achieve remarkably high failure detection rates, nearing the upper bounds, while keeping false positive rates low and minimizing communication overhead. Compared to conventional methods that rely on centralized monitoring, our approach presents a substantial advantage. We achieve communication overhead reductions of up to 80 percent, with only marginal decreases in detection rates and slightly higher false positive rates. Furthermore, our approach caters to both connected and disconnected networks, whereas centralized monitoring is limited to connected networks. When compared to other localized monitoring techniques, our approach maintains similar failure detection rates while remarkably reducing communication overhead by up to 57 percent. Moreover, it significantly decreases false positive rates, with instances as low as 0.01 compared to 0.27 in certain settings. By combining localized monitoring, location estimation, and node collaboration in a probabilistic framework, our proposed schemes exhibit superior performance, making them promising solutions for efficient node failure detection in mobile wireless networks.