Startup error detection and containment to improve the robustness of hybrid FlexRay networks

The research and development on in-vehicle networks (IVNs) is driven by two main requirements: bandwidth and robustness. In this paper we address the robustness requirement. We focus on FlexRay IVNs that are used for safety-critical applications. We analyze and discuss faults that may affect the startup and operation of a FlexRay network. These failures may not only occur during the startup phase of the vehicle, but they may also happen due to a bus problem that requires the bus to be reinitialized during normal operation. Here any startup failure leads to a critical situation like a brake system failure. The fault scenarios we discuss in this paper are the resetting leading coldstart node (RLCN), the deaf coldstart node (DCN), and the babbling idiot (BI). These faults are described in literature, but neither the precise behavior of all involved nodes, nor a clear solution is provided to contain their impact. The idea of a bus guardian (BG) is given in a draft specification of the FlexRay consortium, but no details are given. In this paper, we extend on these ideas by investigating and implementing a detailed (BG) concept, based on our fault analysis. We subsequently evaluate the successful containment of the three fault types in simulation. We also quantify the chip area cost of our solution.