Bench Talk

(Source: Rittis/Shutterstock.com)

Driverless, autonomous transportation is a grand dream of this decade. What’s needed to make this dream a reality is a unique combination of technologies that overcomes the overarching challenges of autonomous driving.

Autonomous vehicles (AVs) need to handle an unprecedented amount of information in real time. For example, to avoid collisions, detect obstacles and pedestrians, and notice a free parking spot, a high volume of metadata must be exchanged among the vehicles, the city infrastructure (e.g., traffic lights, public safety systems), fog/cloud service providers, and even automakers. According to Intel^®, 3.9TB (terabytes) of data must be processed for one such vehicle per day. That’s the equivalent of the daily data usage of 2,666 internet users. Car components such as ultrasonics, radar, global positioning system, cameras, and infotainment systems all add up to this data surge.

The Institute of Electrical and Electronics Engineers and the US Department of Transportation are among the leading contributors to developing new standards for the secure and scalable exchange of data among vehicles (vehicle to vehicle [V2V]), vehicle to infrastructure (V2I), vehicle to network (V2N), vehicle to pedestrian (V2P)—collectively known as vehicle to everything (V2X).

But, V2X standards alone won’t be enough. For vehicles to make complex, autonomous decisions in real time without sabotaging road safety, V2X systems must be orchestrated with a mobile communication ecosystem whose speed and data processing clout can match human reflexes. The reported accidents involving self-driving road tests in recent years emphasize this requirement. Among the 3rd Generation Partnership Project (3GPP) wireless standards, 5G is geared to meet these requirements.

The common premise of AVs is that they continuously sense the observed environment to select their driving trajectories in real time. In bandwidth-constrained radio environments, they have to deal with a high level of noise and interference and a highly dynamic configuration of external entities. To support V2X use cases, high data throughput is not enough. The network infrastructure must also support low-latency, high-reliability network and data security across diverse communication ranges.

Common V2X use cases that can eventually lead to fully connected automated vehicles include:

• Cooperative awareness (e.g., emergency vehicle warning);
• Cooperative sensing (exchange of raw sensor data);
• Cooperative maneuver (coordination of the trajectories among vehicles for a lane change, platooning, intersection control, etc.);
• Vulnerable road user—that is, notification of pedestrians, cyclists, etc.; and
• Road traffic efficiency (dynamic update of routes, maps, etc.).

To support these use cases, more than one wireless technology may be relevant. For example, for short-range, direct communication among devices (V2V, V2I, and V2P) there’s no dependency on the network for scheduling. 3GPP Release 12/13 LTE Proximity Services device-to-device interface can be used to reliably transfer large amounts of data among neighboring vehicles with ultra-low latency. In the case of device-to-network (V2N) communication, traditional cellular cloud services are necessary where 5G New Radio technology is an appropriate choice for what it promises.

Connectivity and Performance

Scaled data and devices are the top challenges of the self-driving era. Compared with 4G/LTE, 5G offers 1,000 times more bandwidth per unit area, 10,000 times more traffic, and 100 times more connected devices per unit area. 5G small-cell technology using millimeter wave bands is more spectrum efficient, which is a big plus for bandwidth-constrained V2X environments.

Reliability

5G promises ultra-low latency (>1 millisecond) that meets the reliability expectations of mission-critical V2X use cases. Relatively speaking, wireless data are easier to intercept and more vulnerable to man-in-the-middle attacks. 5G is expected to improve security by including mutual authentication, local secure elements, Transport Layer Security, 99.999 percent availability, and over-the-air firmware updates for 5G electronics.

Network Slicing

Much like virtual machines on virtualized hardware platforms, the network slicing feature of 5G can support diverse classes of automotive services from different providers over the same infrastructure. It would allow, for example, the telecom operator, the road operator, and the automaker to offer different services to the vehicle and its passengers over a common 5G infrastructure.

To connect vehicles and the back-end infrastructure, 5G V2X applications require new design requirements for the communication systems. 5G V2X specifications are part of 3GPP Release 16.

Industry forums such as 3GPP and the 5G Automotive Association have identified use case–specific performance requirements for 5G V2X systems in terms of latency, reliability, and data rate. 3GPP identified the following five Categories of Requirements for 5G V2X.

• General aspects: Interworking, communication-related requirements valid for all V2X scenarios
• Vehicles platooning: Vehicles traveling together with an extremely small distance between them
• Advanced driving: Semi-automated or fully automated driving
• Extended sensors: Information exchange of all V2X-enabled devices and network elements
• Remote driving: Operators control driving remotely (e.g., in dangerous environments)

It is important to note that the requirements for 5G V2X systems depend on use case scenarios (Figure 1). A typical lane change needs much less stringent latency/delay and reliability requirement than a cooperative maneuver in an emergency situation.

Figure 1: The figure presents the latency and data rate V2X requirements for 5G, as defined in the 3GPP guidelines. These requirements are undergoing fine-tuning based on inputs from automobile original equipment manufacturers. (Source: Use Cases, Requirements, and Design Considerations for 5G/ V2X/arxiv.org)

AVs are essentially data centers on wheels; they rely heavily on computing power at the edge. To transfer a week’s worth of self-driving car data over an advanced Wi-Fi connection, today it would take 230 days. As such, disruptive innovations in products and application-specific integrated circuit processing technology for new radio and antenna architectures are essential for supporting 5G V2X.