The Co-design Of Schedule And Control Based On FlexRay Vehicle Network

With the rapid development of computer networks and ‘X-by-wire’ technologyin automotive electronics, the network control system (NCS) of vehicle is formed bythe shared network. And with the increasing electronic control unit (ECU) added tothe in-vehicle network, it is more strict to the vehicle network with a highertransmission rate, fault tolerance, extensibility and real-time, etc.. Hence, the NCSwith FlexRay bus can meet the real-time communication of large amounts of dataamong the nodes. And FlexRay bus is mainly used in the communication in vehiclesafety systems, such as brake system and steering system.Due to the limited network resources, network scheduling is to rationallyallocate network resources to solve the conflicts during the communication. However,higher sampling period can improve the control performance of NCS, whichoverburdens the network in turn. So, keeping the network control and scheduling arecontradictory. The co-design of control and scheduling is to consider the restrain ofscheduling and control to guarantee the performance and utilization of networkresource. In this paper, the co-design of control and scheduling based onFlexRay-NCS is analyzed, and the scheduling of Static Segment of FlexRay bus isanalyzed as well. The main work is as follow:Firstly, the scheduling of network control systems and communicationmechanisms of FlexRay bus are introduced. A few of scheduling methods areintroduced, and the co-design model based on RM scheduling is presented, which isproved by the TrueTime toolbox on Matlab. Secondly, the setting way of StaticSegment is proposed in detail. Then, considering the influence of data delay to theperformance of network, the strategy of switch network based on activepacket-dropping is proposed to reduce the delay influence. At the same time, in orderto improve the utilization of network resource and reduce the conflict of the data, thescheduling method based on various sampling time point is employed to allocate theresource more rationally. At last, considering the practical application with a mass ofnodes in the network, the manual allocation of sampling time point cannot satisfy thecomplicate network. Therefore, the active packet-dropping strategy based GA is proposed to allocate the sampling time point of each control-circuit. Building theindividual model of scheduling constraints, it is feasible to find an optimisticschedule by evolution to satisfy the overall performance of network, including datadelay, utilization and stability. And through modeling and simulation on Matlab, it isproved that this searching method of GA can improve the computing speed andnetwork performance.