ABSTRACT

As the scale of sensor networks increases, the complexities are inevitably enhanced, which gives rise to the urgent necessity for developing new distributed state estimation technologies in order to meet the need of practical engineering. It is not surprising that, in the past few years, the distributed state estimation problem with network-enhanced complexities has become an interesting and imperative yet challenging topic. Different from the traditional central state estimation techniques, the difficulty in designing distributed state estimation algorithms stems from both the complicated coupling between the sensor nodes according to a given topology and the effects from network-enhanced complexities. In addition, due to physical constraints or technological limitations, data among sensors are usually transmitted over common networks without proper security protections. Specifically, the interconnection of low-cost sensor nodes makes it complicated to protect against inherent physical vulnerabilities therein. Typical attacks include denial of service (DoS) attacks, replay attacks and deception attacks. It is worth mentioning that deception attacks in different scenarios can also be called false data-injection attacks and malicious attacks, to just name a few. Therefore, it is of great importance to determine the impact of distributed filtering.