Timer-Based Distributed Channel Access in Networked Control Systems over Known and Unknown Gilbert-Elliott Channels

In this paper, we consider a system consisting of multiple (possibly heterogeneous) decoupled control subsystems which aim at communicating with their corresponding controllers via shared (possibly) time-varying wireless channels. To address the resource allocation problem in a distributed fashion, we propose a timer-based channel access mechanism in which the subsystem with the smallest timer value, in a channel, claims the slot for transmission in that specific channel. The value of the timer is inversely proportional to a cost which is a function of the temporal correlation in the channel variation and the subsystem state. This cost can be calculated individually and does not require explicit communication between the subsystems, since it is based on locally available information only. The temporal correlation in the channel variation may be unknown and, in such cases, each subsystem tries to deduce it via machine learning techniques. The performance of our proposed mechanism is demonstrated via simulations