The evolution of IVN architectures
The earliest incarnation of an IVN was the Local Interconnect Network or LIN, a simple single-wire protocol where a master controller controls a group of slaves, each identified by an ID. Invented later, in the late 1980s by Bosch, the connected area network (CAN) bus IVN is a point-to-point system that wires all ECUs (nodes) to each other. Communication is packet based via a single twisted pair cable. The data rate capabilities are comparatively low; the system is asynchronous, hence unsuitable for time-critical operation. However, it does feature a unique non-destructive arbitration method for handling message collisions. Both systems feature today for non-time critical systems such as windows, lights, rain sensors, and door locks. Several alternative protocols, including “FlexRay”, MOST, and CAN FD (FD = flexible data rate), came to market to increase data rate capabilities.
Figure 1 shows the various IVN types and data rates.
However, the benefits of scalability and adaptability that Ethernet offered made it an enticing choice for designers. To successfully integrate Ethernet into the automotive environment, modifications are needed to reduce susceptibility to electromagnetic interference (EMI), add time sensitivity (TSN), and improve robustness and reliability. By exploiting the versatility of the original Ethernet design, a modified communications interface (Physical Layer), more resilient to the hostile automotive environment, is implemented.
The modified physical layer employs single unscreened twisted pair (UTP) cable, transformer coupling, and full duplex capability, with a range of 15m.
Ethernet is exceptionally versatile; ECU nodes can be added at will without reprogramming. Collisions are eliminated because the system is full duplex (data can be transmitted and received simultaneously). Traffic is optimised, and virtual sub-networks are created by adding intelligent switches.
ECUs connected in a ring topology, as shown in Figure 2, offer a level of parallel redundancy suitable for time-critical safety systems. As communication is full duplex, communication can occur in the other direction if there is a break in the ring at any point.
Recent developments
A new automotive Ethernet system 10BASE-T1S is entering the market, providing up to 10Mbit/s data rate over a single twisted pair cable. This system, developed as a direct competitor to CAN, is a bus architecture that offers advantages in terms of the amount of cable used and cable routing through the vehicle. The system avoids message collision by using a pilot signal with a beacon containing an ID. Each node recognises its ID and only transmits information if its ID is in the beacon.
CAN XL is the next step in the CAN bus evolution and boasts data rates of up to 20Mbit/s while retaining the non-destructive message collision arbitration enjoyed by its predecessors, CAN and CAN FD. The system is interoperable with CAN FD, and its physical layer is operable with all transceivers. Both of these systems aim to bridge the data rate gap between CAN FD and 100BASE-T1.
Classic IVN architecture
Initially, the ECUs (nodes) were wired together in a point-to-point scheme; ECUs were mounted on each major sub-assembly (the engine, gearbox, braking system, for example). This distributed approach made sense in a mechanical context, as the sensors that acquired data were situated in the same proximity as the ECUs that processed it. The downside was the considerable cabling cost, connecting everything, and the speed limitations posed by this approach.
Domain IVN architecture
Designs that feature advanced driver assistance systems (ADAS), which process data from multiple sensors (sensor fusion), require many additional electronic devices and control units. Also, high data rate capability is fundamental to the integrity of these systems, as they are time critical. Centralising and consolidating these ECUs into functional groups (domains) helps reduce complexity and cost. The added advantage of this approach is that the individual domains can use different bus systems, Ethernet and CAN, which can coexist. Network gateways perform the vital function of translating the data between the various buses, ensuring reliable communication. Additionally, a connected vehicle, with V2X communication, can receive over-the-air software updates, for a particular function (domain), without the need for expensive factory recalls.
Zonal IVN architecture
An extension of this approach is to use a zonal architecture. Powerful zone controllers, effectively centralised ECUs, are placed in key vehicle locations (zones) with sensors and actuators connected directly. Cabling runs are shorter, which lowers both weight and cost. The zone controllers are connected to a central controller by a high-speed backbone. The central controller handles sensor fusion (ADAS) and high-level decision-making. Because of the time-critical nature of ADAS systems, a large amount of data is generated by the various elements and has to be communicated and processed. The high-speed backbone in the zonal architecture ensures that time-critical data is shared instantly and provides adequate parallel redundancy for safe autonomous features.
The importance of data rate
High data rates are pivotal for the integrity of sensor-fused systems, such as ADAS. However, not all systems in the vehicle require them. For instance, the simple LIN bus is an adequate, cost-effective solution for windows, door locks, lights, and seat adjustment. Powertrain and actuator functions are often controlled by a CAN bus, and FlexRay is often employed for anti-lock braking, electronic power steering, and vehicle stability functions.
It is not uncommon to encounter several different systems within an IVN. CAN FD and 10BASE-T1S are helping to bridge the data rate gap.
Conclusion
In-vehicle networks have evolved dramatically in the last decade. Today’s connected vehicles include infotainment systems, ADAS, and V2X communication - producing and transmitting vast amounts of data to and from the cloud every hour. IVNs are pivotal to the performance of the connected vehicle; their importance will only increase as we head further toward vehicle autonomy.
