Download this article in PDF format.
Cars and trucks are quickly becoming Formula One race cars—at least in terms of data. With more types of onboard devices requiring near-instantaneous data transmission than ever before, engineers and designers are challenged to accommodate the accelerating need for bandwidth and speed.
At the core of just about every innovation today, from advanced driver-assistance systems (ADAS) to collision detection sensors and infotainment systems, is data. Traditional automotive data networking technologies such as controller area networks (CANs), local interconnect networks (LINs), and Media Oriented Systems Transport (MOST) were not designed to support the bandwidth these systems demand. In fact, the need to implement time-sensitive networking (TSN) standards has forced engineers to look outside the automotive arena for alternative transit solutions.
Ethernet is the obvious choice. This staple of the IT world, while not exactly new to automobiles, is being applied with increasing frequency, and for a number of reasons. Ethernet technology allows for fewer cables of lighter weight—not an insignificant advantage. Also, automotive engineers know that Ethernet is proven technology, supported by many device manufacturers, and has a strong hardware/software support ecosystem.
Yet Ethernet can’t satisfy all requirements for data networking performance—which is why there’s also a need for TSN technology. TSN guarantees that high-quality data packets are delivered with low latency, something Ethernet doesn’t natively support. In addition, TSN provides a network-wide clock for packet synchronization across systems, and prioritizes time-sensitive data streams over those of lower priority. Finally, it guarantees a minimum level of availability for emergency transmission.
Drive Testing Limitations
Validating high-speed Ethernet devices for automotive use is a complex undertaking.
Automakers, along with in-vehicle device and system OEMs, have stringent requirements for latency, synchronization, conformance, availability, and QoS. These requirements must be met, as consumers must be able to rely on their cars and trucks for safe, reliable performance. The potential cost of failure—not to mention recalls, liability, and damaged reputations—is simply too high.
Until now, the most commonly used form of validation for automotive systems is the drive test. But drive testing has its limitations. Foremost among these is the narrow range of operating conditions—when testing digital systems, real-world scenarios simply aren’t diverse enough.
Simulation testing, using a purpose-built testing device, is the better choice for Ethernet-based automotive networks. Simulations offer a greater array of connections, domains, and traffic profiles, giving engineers the ability to create—and repeat, on-demand—virtually any kind of condition they wish. Furthermore, because the operating environment is completely controllable, the user can choose which types and amounts of data to be transmitted, the amount of bandwidth to be allocated to each type, numbers of nodes, and so on. It also allows engineers to reproduce conditions exactly, to determine whether a failure was random or situation-specific.
Test simulators provide ready-to-use test cases for AUTOSAR, Open Alliance, and Avnu industry compliance, along with European (ETSI ITS-G5) and American (USDoT WAVE-DSRC V2X) government standards. Finally, they ensure that tested components or modules will be interoperable with any other verified hardware or system.
Many types of tests can be conducted with a network simulator, generally falling under two types of methodologies. First is mathematical simulation, a process in which algorithms are applied to calculate and simulate best/worst-case scenarios for packet delays. While mathematical simulation establishes theoretical performance limits, experimental simulation—the second method—delivers practical validation. Experimental simulation uses a combination of hardware and software to simulate real traffic within real networks to determine delays under specific conditions.
Both classes of simulation have their place. Mathematical simulation and analysis can be done virtually, with no network present—while experimental requires an actual network with real devices. Used in concert, they offer the best overall results.
Multiple Test Options
Within the experimental environment, users have their choice of test conditions. Network simulators support emulation of multiple talkers or listeners on each port, using any combination of TSN and/or non-TSN traffic. In a one-armed setup, the simulator acts as either the talker or listener (used when testing an end point such as an Ethernet transceiver). A two-armed setup, by contrast, is one in which the tester acts as both talker and listener (i.e., testing a switch or bridge). Any number of test scenarios can be staged as well. The following are just a few examples:
Emulating listeners and talkers in ADAS: Without a doubt, ADAS devices have the toughest performance requirements, since emergency-braking and collision-avoidance systems require tightly bounded latency. Using a network simulator, engineers can reliably verify these latency mandates. They also can simulate adverse network conditions such as data overload or multiple simultaneous talkers. An added benefit is the ability to precisely record transmission and reception times.
Lip-synched multimedia playback: Media systems often require surround-view images collated from different cameras, or DVD playback on multiple displays, speakers, and headphone ports. Using a simulator, testers can verify a network’s bandwidth reservation, timing, and synchronization for media applications. A simulator can also measure timing variance between one output device and another, in both video and/or audio modes.
Connected car (car-to-X) testing: GPS and other forms of wireless external vehicle information are evolving at light speed. Soon real-time traffic data will be an integral part of mapping functions, enabling vehicles to self-select preferred travel routes. Not all data is equal in importance, though—and Ethernet networks must be able to discern and prioritize various streams. When both TSN and non-TSN data are present, networks must be able to support high-priority and best-effort instances. Similarly, when there’s bandwidth competition between a real-time function (such as a mapping app) and an audio/video display with a QoS requirement, networks must be able to respond properly. A simulator allows testers to determine how the network will behave under these differing conditions.
Simulators as Long-Term Solution
Today, IT and network equipment manufacturers and providers from the aerospace, industrial, and automotive industries are all gearing up to validate higher throughput and worst-case latency for the end-to-end performance of their networks. Whether for plane, train, automotive, or industrial robot applications, test simulators can emulate every kind of network protocol.
TSN looks to put in place standards to help increase determinism of Ethernet-based systems so that engineers can reliably validate systems and devices for any Ethernet OSI Layer, from Layers 2-3 (conformance/performance testing) to Layers 4-7 (vulnerability performance). These standards will be critical as vehicles depend on more data to achieve a reliable situational awareness, and to ensure that any autonomous driving mode is a safe and smooth driving experience.
Having the right testing equipment—and picking a testing partner with expertise in network validation—will be essential in meeting these challenges. As data becomes as important to modern vehicles as oil and gasoline were in the past, simulators will provide the assurance that every vehicle operates at peak performance.
Jeff Warra is Senior Technical Marketing Engineer at Spirent Communications.
