Data aggregation based topology inference for wireless sensor networks

Knowledge of topology in wireless sensor networks is significant for network management and maintenance. In this paper, a conditional probability of data loss theorem is proposed for wireless sensor networks based on the data aggregation paradigm.al probability of data loss theorem. It reveals the relationship between conditional probabilities of sensor data loss given different conditions. Based on this theorem, we propose a novel algorithm to infer topologies of sensor networks using end-to-end loss measurements. The algorithm does not incur any additional burden to the network. A large number of networks are simulated in different scenarios with NS-2. The results show that our proposed algorithm can identify more than 95% of topologies within a small data collection rounds.