A Performance Model of Controller Area Networks for Vehicles under Unsaturated Traffic

Abstract

This paper proposes a performance model of Controller Area Networks (CAN) to measure the effect of bit-error, especially on an identifier field of CAN messages over erroneous channels. CAN were introduced to allow various devices embedded inside cars to communicate in real-time over a shared channel. For resolving collisions in a real-time way, CAN adopt a contention-free protocol, namely a bit-map algorithm where messages are bitwise arbitrated based on their priority, precisely the identifier in CAN's messages when multiple nodes contend for the shared channel. Contamination on the identifier, therefore, can severely degrade the performance of CAN since all senders recognizing errors immediately terminate their transmission and wait for a certain amount of time to resume competition. Furthermore, when the number of errors experienced exceeds the predetermined threshold, CAN stays at the busoff mode where no messages are sent, requiring outside intervention to be back to the normal operational mode. To predict the performance deterioration of CAN over noisy channels, this paper builds a Markov chain model of CAN.

Autonomous Driving Path Planning based on Sarsa-Dyna Algorithm

“…At present, the problem of path planning has attracted the attention of many scholars, and there are some successful research results, which can be divided into global planning and partial planning. The algorithms for global planning include the A* algorithm [1] [2] and its various improved versions, etc. These algorithms have performed excellent results in offline static environments.…”