Enhanced Source Detection in Incomplete Networks via Redundancy-Guided Sensing and Anchor-Bounded Inference

Despite recent advances in source detection, accurate and efficient detection in incomplete networks remains a significant challenge. Current methods, which typically rely on the entire network structure, are limited when the acquired network is structurally incomplete due to the data acquisition limits in practice, and they also incur substantial computational overhead when applied to large-scale networks. Therefore, we propose an efficient approach for source detection in incomplete networks via Redundancy-Guided sensing and Anchor-Bounded inference (RGAB). To address the bias introduced by missing edges, RGAB develops a redundant coverage metric for sensor deployment and iteratively deploys sensors based on a likelihood score that integrates coverage capability and redundancy contribution of nodes. Furthermore, to mitigate the computational burden of detection, this approach introduces two anchor sensors and restricts the candidate region based on the radial circle of anchors. Comprehensive evaluations conducted on diverse benchmark datasets confirm RGAB's superior performance over baselines in source detection in incomplete networks.