Networked-based robust control design using multiple levels of network traffic

This paper proposes a method of designing a network-traffic-dependent controller based on not only the deterministic information from the current timestamp but also the stochastic information from the finite sequence of the past timestamps. For its practical implementation, this paper introduces two algorithms: pattern generation (PG) algorithm and pattern identification (PI) algorithm. The PG algorithm classifies off-line the network traffic into some separate patterns using training sequences of timestamps, based on which a less conservative networked control system is designed. Also, the PI algorithm searches on-line for a pattern representing the recent network traffic, which is applied as a decision reference for operating the designed network-traffic-dependent controller.