Mitigating byzantine attacks in data fusion process for wireless sensor networks using witnesses

Data fusion techniques are a promising solution to increase the lifetime of wireless sensor networks, because it allows the reduction of data that needs to be transmitted back to the network sink. However, data fusion schemes are vulnerable to Byzantine attacks. Specifically, a Byzantine (dishonest) data fusion node can send falsified fusion data to mislead the sink. Hence, many witness-based schemes have been proposed to rectify such vulnerabilities by making use of additional “witness” nodes to monitor the main fusion node. The witnesses then submit a “vote” to the sink which decides whether the received data is valid. This paper proposes a dynamic witness concept together with distributed forwarder node monitoring to validate the transmitted fusion data. The distributed monitoring system detects any modification of the fusion data in transit while a randomly chosen witness is used to monitor the fusion node. We show through analysis and simulation that our scheme offers better resilience against Byzantine nodes and improved energy efficiency compared to existing witness-based approaches. More importantly, we reduce the overheads and limitations of existing witness-based schemes to make such schemes more practical.