A Genetic Algorithms based optimization for TTCAN

Abstract

Controller Area Network (CAN), a typical event- triggered protocol, has been widely used in constructing distributed real time network. A common opinion is that the arbitration mechanism of CAN might introduce jitters, which is highly undesirable in time and safety critical applications. Time-triggered protocols address exactly the problem. Time- triggered protocols, such as Flexray, Time-Triggered CAN (TTCAN) and Time-Triggered Protocol (TTP/C), have been applied in time and safety critical applications because they provide deterministic response times for time and safety critical tasks. Recently, much attention has been paid to them. TTCAN, a time-triggered extension of CAN, can easily apply to the numerous of existing CAN-implementations in a wide range of real time requirement fields. A key problem of TTCAN is how to design an effective scheduling table off-line. In this paper, we analyze message set and design scheduling table from a new angle of view, and the key point of the view is to make event-triggered messages gain the maximum transmission times by minimizing the space occupied by time-triggered messages. Then a Genetic Algorithms (GAs) based method is proposed to optimize the scheduling table consequently.