Controller Design for Leader-Follower Systems With Hidden Markovian Jamming Attack

Transportation systems often use agent-to-agent communication technologies to enhance control performance by transmitting information across wireless networks to keep a reasonable inter-distance between agents. The agent-to-agent data transmission is influenced by the fading channel and the limited communication bandwidth. Denial of Service attack model is used to describe the attack’s influence on control system. Thus, the communication model is built as Bernoulli distribution, Markovian distribution and so on. But in practical situations, it is more complicated than we imagine. To further push the theoretical results to practical situations, we investigate the jamming attack with non-stationary stochastic process in the leader-follower systems and introduce hidden Markovian distribution model (HMM) to describe this special jamming attack. Furthermore, the controller design method is presented to achieve stochastic stability of the leader-follower system. The structure of the control algorithm is a special MPC method, which is divided into a state feedback part and a modification part. The state feedback part is a typical LMIs controller to guarantee the system’s stability, while the modification part is used to improve the influence of the model mismatch and uncertainty. The highlights of the results are the following two points. Firstly, the more complex and precise transportation system model is founded by the hidden Markovian distribution. Secondly, the controller design method is also presented for the multi-agent systems with this HMM transportation system model. The existing results always find the worst case scenario, and design the controller in this scenario. Usually, the controller is very conservative and even unstable. The results in this paper can change the controller in different scenarios to adapt to the corresponding jamming stochastic distribution model, which is the novelty of this paper. We use two examples to illustrate the proposed results. The first one is used to show the ineffectiveness of the results by the simple stochastic distribution model method and the effectiveness of the proposed results in this paper. The second one is used to show that the method can be used in a 3-vehicle-agent system.
Note to Practitioners
—The paper’s results have some potential applications in logistics industry, automated factory, and flexible automated manufacture systems. In these scenarios, the feed flow mobile robots are widely used to exchange a mass of manufacturing materials. On one hand, usually, single vehicle control theory does not have the group intelligence ability. Its actions only consider the safety and efficiency itself, which may not be the best actions for the whole system. In this paper, we only consider that the vehicles form a platooning, which is the most time of working conditions for mobile robots. On the other hand, usually, the network data exchange environment in manufacturing industry is complex, which is main reason to cause instability of the networked platooning system and also the bottleneck limiting the network data into the closed loop of the control system. In this paper, the new transmission interference model (HMM) is built, whose parameters can be confirmed by the algorithm and the observation data. Based on this new model, the corresponding unique controller is also designed accordingly. The challenge of this method is to find a method that can suppress large disturbances to the maximum extent and maintain system stability. But if the big interference lasts long enough, it means that the system lasts long enough in the open-loop condition, and the system eventually becomes unstable, which is the limitation of our research in theory.