Technical Challenges for Establishing a Scalable Architecture from Cars to the 5G Cloud

"5G is not a standalone technology. It is about expanding existing LTE capabilities and adding new capabilities. To be more specific, we need to develop features that improve machine communications and computing when vehicles are connected to the 5G network." This statement from Geng Wu during a recent iGR webinar summarizes the challenge facing innovators as the telecommunications and automotive industries collaborate to enable connected vehicles (CV) and autonomous driving (AD). Wu is an Intel Fellow and chief technologist in Intel's wireless, next-generation standards organization.

The stakes are high—the United States Department of Transportation (DOT) claims that just two vehicle-to-vehicle (V2V) applications—Intersection Movement Assist and Left Turn Assist—could result in a 50-percent reduction in crashes, injuries, and fatalities, on average (source: DOT’s V2V Communications Fact Sheet). Another Intel expert, Patricia Robb, who heads up Intel’s Silicon Valley Innovation Center, said in the webinar that studies have shown that advanced driver assistance system (ADAS) technologies are already reducing fatal accidents by 30 percent.

“I'm super excited to work in this space because of the positive and transformational impacts it will have on society,” said Robb.

CV and AD have different technical requirements, because autonomous vehicles are not necessarily connected (although they benefit from connectivity, which can expand their sensing capabilities). Overall, low latency, high bandwidth, and reliability are the three hallmarks for CV, and for AD that uses connectivity.

Much work has already been done to address the technical requirements for both CV and AD. For example, for autonomous vehicles, the ISO 26262 standard assigns an Automotive Safety Integrity Level (ASIL) to all automotive functions. These levels range from ASIL-A (lowest) to ASIL-D (highest). Functions that have zero safety impact, such as the graphical user interface for a car’s radio, have a lower ASIL classification than functions that can have a significant impact on safety, such as automatic emergency braking. The SAE standard, J3016, also lays out a taxonomy for motor vehicle driving automation systems, ranging from level 0 (no automation) to level 5 (full automation). The Intel® GO™ Automated Driving Solution pulls together many of these specifications with powerful and efficient compute performance, functional safety features, multilayered security, and flexible hardware acceleration.

For CV, some standards already exist, including dedicated short range communications (DSRC), Cellular Vehicle-to-Everything (V2X), and the Wireless Access in Vehicular Environments (WAVE) Standards introduced by IEEE. But these do not go far enough, said Wu. The industry needs to develop features and platforms that are more attuned to support machine-to-machine (M2M) communications. In the connected car, human communications, machine communications, and computing merge, multiplying the benefits of all three.

CV bandwidth and latency requirements depend on the application. For example, real-time high-definition map updates may require a higher bandwidth and lower latency than over-the-air (OTA) firmware and software updates. Infotainment, smart parking, and vehicle remote control are other CV use cases to consider. The 3rd Generation Partnership Project (3GPP) standards body is taking a three-phase approach to increasing the data transfer rate for 5G. For the current release 14, they are targeting one to ten megabits per second (Mb/s). Release 15 will increase the target to 10-50 Mb/s, and for 5G NR and future releases, the target is 50 to 500 Mb/s.

“For both the connected vehicle and autonomous driving, we need a new type of interface, network architecture, and cross-industry collaboration that is aware of 5G. A lot of things still need to happen,” Wu said.

These “things” include global and scalable interoperability (the ability of systems and devices to exchange and interpret data), increased security at both the interface and the application layers, a consistent user experience regardless of geography, and integration between data centers, towers, and antennas. A global, interoperable CV platform will enable value-added services and use cases. Part of the solution may be the Open Network Automation Platform (ONAP), which provides a comprehensive platform for real-time, policy-driven orchestration and automation of physical and virtual network functions that will enable software, network, IT and cloud providers, and developers to rapidly automate new services and support complete lifecycle management. Intel is a Platinum-level member of ONAP.

As part of the necessary platform development, added Caroline Chan, vice president of Intel’s 5G infrastructure team, the cloud needs to move much closer to the edge.

"The [5G] cloud is not the traditional cloud you're thinking of. We need to get closer to where the cars and the users are. In this case you can take advantage of short range communication at a higher bandwidth of the higher spectrum," explained Chan.

Using network function virtualization (NFV) and software-defined networking (SDN), the 5G network/cloud will support quick (think minutes, not days or weeks) spin-up of microservices. Through automation and a platform (or network)-as-a-service model, the network will be able to manage itself.

“Connected cars need to be aligned with that vision,” said Chan.

Learn more about Intel’s vision for next-generation transportation.