Resilient Distributed Information Fusion Under Multiple Malicious Attacks

Currently, distributed information fusion technology is being deployed to monitor a parameter of interest in numerous electronic systems. To attain accurate parameter, ensuring the security of the distributed network forms a crucial foundation for information fusion. However, the stability of many detection algorithms remains inadequate when the network simultaneously contends with multiple attacks, including Byzantine attack, data manipulation attack, and false data injection attack. Moreover, some detection algorithms focus on resilience in both normal and compromised nodes, which increases severe restrictions on the network. To solve these problems, this article considers these three attacks simultaneously, and then proposes a resilient distributed estimation algorithm to achieve robust distribution information fusion. In the proposed resilient distributed estimation algorithm, a secure detection mechanism with double data verifications is first designed to detect the secure and compromised nodes. Afterwards, a resilient distributed data fusion method is developed to isolate the compromised nodes, decreasing network restrictions and providing the optimal estimation by secure data fusion. Statistical analyses of the proposed algorithm are carried out to assess its stability and mean-square behaviors, and the algorithm complexity is also analyzed. These analyses indicate that the proposed algorithm offers the better time and space complexities. Simulation experiments further illustrate that the proposed algorithm with the resilient information fusion is robust and effective under multiple attacks.