About This Training
Historically, automotive network segments used a single data rate speed based on the network technology being used. If more bandwidth was needed for an application, a different network technology was used. The typical speeds used today are 20 kb/sec for LIN, 500 kb/sec for CAN and 4 to 8 Mb/sec for CAN-FD. Connecting one network technology to another (for example, LIN to CAN) requires a Gateway device, which use a CPU to facilitate the cross protocol conversion.
The introduction of automotive Ethernet changes all that, and with it comes new challenges. Automotive Ethernet networks today support full-duplex link speeds of 10 Gb/sec down to 100 Mb/sec and multi-drop 10 Mb/sec link segments, and Ethernet can interconnect all these different link speeds with lower cost Bridges (quite often called switches) that rate covert from one speed to another usually in hardware, since there is no need for any protocol conversion (and most CPUs can’t keep up with the higher packet rates anyway).
New challenges appear when higher speed links are connected to slower speed links. Without any mitigation, these network congestion points typically drop some packets when large bursts of high-speed packets are mapped to slower speed links. Think of a central compute ECU bursting out data from a super high-speed link to end nodes at the edge of the network connected with 10 Mb/sec shared media. For some applications, this is not a big problem, but for others it is a critical problem.
Thankfully, these congestion point problems have been around for decades in non-automotive Ethernet applications and many solutions have been tried and work for their intended use cases. Unfortunately, some of these same working solutions have been applied to other use cases with less than satisfying to disastrous results. There are options, if properly applied, and now is the time to impart this knowledge and experience for automotive network designers.
This presentation will cover many of the congestion mitigation techniques and their intended use cases, along with some disastrous uses, so that educated decisions can be made.