A blend of in-vehicle and cloud computing, more powerful antennas, automotive Ethernet, and high-performance 5G network communications will be essential ingredients in the evolution of autonomous vehicles.
Guido Dornbusch, Alex Bormuth, Dr. Ayman Duzdar | Sep 12, 2019
Source: Electronic Design
Projections show that by 2020 the power and electrical components, sensors, cameras, radar, GPS and other systems in new model vehicles will generate four terabytes of data in an hour and a half—the average time most people spend behind the wheel each day. Vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) communication with infrastructure and other cars on the road will drive up that number substantially—and compound the projected data deluge.
Connected and autonomous vehicles share many attributes similar to the human body. They are highly complex structures designed to transmit messages over a multitude of pathways. The human brain contains about 100 billion neurons constantly transmitting electrochemical signals to muscles and organs. Concurrently, impulses perceived through sensory receptors are being transmitted rapid-fire back to the brain, enabling the body to communicate, act, and react.
In much the same manner, autonomous driving requires powerful technologies to anticipate and react in real-time to internal and external stimuli. Powerful computers comprise the brain of the self-driving car, while automotive sensors serve as the sensory system vigilantly detecting dynamic conditions on the road. In the not-so-distant future, automotive Ethernet and high-performance 5G communication systems transmitted via multiple powerful antennae will form the nervous system of connected and autonomous vehicles.
Automotive Ethernet Ramps Up In-Vehicle Bandwidth
On the path to commercially available 5G network communications and fully autonomous roadways, auto manufacturers and suppliers must tackle myriad safety and technical challenges. Overcoming data latency is one of the most critical steps to assure rapid vehicular response time to internal and external stimuli.
The human brain and body have evolved to assure survival under an extraordinary range of conditions. Connected automotive design must continue evolving to assure the utmost safety of the driver and others on the road. Intelligent and autonomous vehicles must be able to independently manage safety issues and navigate the roads without relying on data produced by other vehicles or the infrastructure. That relies on data from hundreds of sensors and an agile nervous system communicating with computing units throughout the vehicle.
Automotive Ethernet will help usher in the high speeds needed for autonomous-vehicle data processing.
Ethernet brings a track record of success in multiple industry sectors, making it a solid choice to serve as the nervous system for next-generation cars. Automotive Ethernet is well-positioned to meet industry demand for transmission speeds, fault tolerance and, above all, safety. Furthermore, the Ethernet is widely considered to be future-proof, a feature that’s vital to autonomous driving.
Automotive Ethernet will make it possible to achieve in-vehicle high data speeds. Currently, automobile data networks have speeds of up to 10 Gb/s. Scalable automotive Ethernet can enable higher bandwidth and faster signal processing essential to autonomous driving. In addition, the Ethernet must be fail-safe and reliable. A multi-layered security approach should include hypervisor capabilities so that the platform can run multiple virtual machines and applications simultaneously, with safety enhancements such as multi-zone, fail-functional, and redundancy capabilities.
Equipping vehicles with redundant wiring harnesses can help protect against a partial failure by facilitating the continued operation of the entire system. A ring-shape arrangement of cables is another method to boost Ethernet reliability, allowing individual components to continue communicating with one another even if a failure occurs at one point in the ring.
Another critically important job of automotive Ethernet is to quickly and reliably supply safety-related data collected by vehicle sensors to the computing units. This will help enable the vehicle to autonomously operate, particularly in high-traffic areas. Data is communicated to the vehicle via antennae that meet certain requirements to quickly feed external data to the vehicle’s computer.
Capturing data and communicating with the surrounding environment are essential functions to achieve the full potential of driverless cars. In addition to facilitating in-vehicle communications, antennae are already used to supply signals received from other vehicles or infrastructure. They serve as the voice and ears of the vehicle to the outside world, sending and receiving signals to communicate with its environment.
One such scenario, for example, might occur when the vehicle applies the brakes earlier and more gently, because the vehicle ahead has wirelessly reported a braking maneuver. Another example might be an ambulance reporting its presence wirelessly to vehicles ahead in order to alert drivers to open a corridor.
5G Infrastructure and Communication Key to Adoption
Meeting the rising demand for automotive data bandwidth will require more powerful connected infrastructure and networks to keep pace as autonomous vehicles make the transition to full-scale adoption. Existing 4G networks lack the bandwidth and speed needed to meet data volume requirements of autonomous driving. Radio waves from 4G communications must be relayed to a cell tower before transmitting back to the vehicle, which results in unacceptable data latency.
Due to a current lack of bandwidth, sensor data in today’s vehicles is being transmitted in a heavily preprocessed state. The currently available bandwidth per vehicle generally amounts to about a few hundred kilobits. This suffices for today’s connected vehicle. However, fully autonomous driving will require cars to be able to receive more sensor data, both processed and unprocessed, making higher bandwidth essential.
McKinsey predicts that operators globally will continue investing in infrastructure upgrades to meet the projected 20% to 50% annual increase in data traffic. Designed to supplement 4G networks, 5G technology is projected to accelerate data-transfer speeds from 100 Mb/s to 10 Gb/s or more, and significantly improve bandwidth, capacity, and reliability.
In addition to improving bandwidth (the amount of data that’s transferred), 5G can improve the latency (the delay in data being transferred). A long latency interferes with seamless collaboration between two devices. If you’re designing a transportation system, a long latency means the network can’t help a vehicle deal with a quickly changing environment, which has safety implications. In some situations, 5G is expected to improve latency by a factor of 50.
In the next several years, the expansion of 5G network infrastructure will ultimately enable ubiquitous driverless cars and allow data to be aggregated and shared between vehicles to improve visibility as well as let the network contribute to vehicle safety. Ideally, the autonomous vehicle would be capable of receiving raw sensor data from within the vehicle, as well as from other vehicles and infrastructure, in an amount equal to the bandwidth capacity within the vehicle.
It’s anticipated that each OEM will implement its own brand-shaping algorithms for processing that data, something that can function only if it’s able to access raw sensor data from the environment. This will require significantly higher data streams extending to the gigabit level. Other scenarios involving the transfer of tremendous amounts of data are also important to autonomous driving. This includes downloading of up-to-date high-resolution maps showing construction sites, accidents, and other obstacles potentially requiring the vehicle to react.
Industry-Led Efforts to Develop a 5G V2X Standard
To leverage required levels of bandwidth, the automotive antennae will need to cover a larger frequency range. A number of industry groups are currently developing a new standard for the automotive antennae. The 5G V2X standard is expected to facilitate several hundred megabits/s, with several gigabits/s of bandwidth. At this level, vehicles would be able to receive and send the relevant data quantities for increased safety and comfort. Because standardization groups are currently meeting and defining automotive application cases, experts forecast the first 5G V2X-enabled products will appear on the market soon and fully support highly autonomous driving by 2025.
The frequency range for 5G V2X poses another important technical challenge. Radio spectrum has been a prized commodity since the beginning of the 20th century, and there’s no available, and gratis, frequency range below 60 GHz around the world that could carry the required amount of data.
Standardization groups are hammering out details for a 5G V2X standard, including the antennae to handle the high frequencies.
Higher frequencies, i.e. 60 GHz, are more available because the technical challenges with using them have limited their commercial appeal. The higher the frequency, the greater the losses in transmission and processing. Circuits become smaller and require expensive components. Test equipment becomes more expensive. Even air (specifically oxygen) becomes a problem because O2 oscillates at 60 GHz and absorbs energy.
However, antenna technology offers a way to overcome these problems. Rather than “waste” energy by sending out an omnidirectional (ring-shaped) signal, an antenna array can direct the energy in a specific direction to the other device. This direction can change as the two devices move relative to each other. The directed antennae would need to both receive and send signals, so they’d be connected to each other and to the vehicle’s computers using automotive Ethernet for in-vehicle data transmissions.
Preparations Underway for Driverless Cars and Roadways
Numerous automotive manufacturers have already launched prototype autonomous vehicles with others planning to do so soon. Volvo is already testing self-driving cars on Swedish roadways. Daimler obtained authorization to test autonomous cars on urban streets in Beijing. Daimler and Bosch are planning to pilot highly autonomous ride-hailing vehicles in San Francisco. In 2018, General Motors announced an investment to develop a level-five autonomous vehicle (with no steering wheel, gas, or brake pedal). In 2021, Ford intends to launch a fleet of self-driving cars. These are just a handful of the initiatives in the works.
The connected car of the future has arrived—and will continue to evolve. In 2018, automobiles with connected capabilities represented almost 39% of the US market. An estimated 250 million connected vehicles will be on the road by 2020—and market penetration of connected vehicles will reach over 80% by 2022. ABI projects as many as 8 million driverless cars will be added in 2025.
However, technology leaders and automakers will need to address challenges to bring safe and reliable vehicles to market, and bring drivers on board. According to a recent survey by the American Automobile Association (AAA), over 70% of U.S. drivers expressed fear of riding in a driverless vehicle. It’s absolutely imperative that driverless cars translate the need for a safe, secure, reliable, and connected vehicle foundation into a high-performance computing network on wheels.
The evolution of driverless cars requires integrated high-speed and bandwidth signal integrity, network traffic prioritization, system scalability, and security—all of which demand in-vehicle and cloud computing, an ever-increasing number of powerful antennae, automotive Ethernet, and high-performance 5G network communications.
Guido Dornbusch is Vice President of Product Management, Connected Vehicle Solutions; Alex Bormuth is the Director of Business Development, Connected Mobility Solutions; and Dr. Ayman Dezdar is VP of Technology at Molex.