Detecting Malicious Data Injections in Event Detection Wireless Sensor Networks

摘要: Wireless sensor networks (WSNs) are vulnerable and can be maliciously compromised, either physically or remotely, with potentially devastating effects. When used to detect the occurrence of events such as fires, intruders, heart attacks, malicious data injected create fake events, thus trigger an undesired response, mask actual events. We propose a novel algorithm identify injections build measurement estimates that resistant several compromised sensors even when they collude in attack. also methodology apply this different application contexts evaluate its results on three datasets drawn from distinct WSN deployments. This leads us tradeoffs design algorithms how influenced by context.