Network Topology Inference Based on Timing Meta-Data

A set of low-cost sensors is deployed to infer the network topology a self-organizing wireless network. The operate in non-invasive fashion, extracting only timings data packets and acknowledgment (ACK) from all nodes meta-data also reports source node each packet, but not destination or contents packets. central processor collects sensors, goal for based solely on such information. Prior work leveraged causality metrics identify which links are active. If ACK two nodes– say 1 2, respectively– causally related, this may be taken as evidence that communicating 2 (which sends back 1). This paper starts with observation packet losses can weaken relationship between timing streams. To obviate problem, new Expectation Maximization (EM)-based algorithm introduced– EM-causality discovery (EM-CDA)– treats latent variables. EM-CDA iterates estimation evaluation metrics. method validated through extensive experiments sensor networks NS-3 simulation platform.