“Horn, OK Please” to “Talk, OK Please”: The Road to Self-Driving Cars

May 9, 2018

The ubiquitous “Horn OK Please’ sign has adorned the Indian highways for many decades now. For ages, Indian cars and drivers have used honking as a safe and secure way to let each other know many things, such as “I am passing”, “there is a havildar up the road”, “to get out of my way”. I sometimes wonder whether it was this rudimentary idea of communication that has led to the recent revolution in autonomous vehicles, where cars can “talk” to each other, albeit silently over radio frequencies as opposed to blaring audio signals.

Humor aside, it is a quite a race to market for the self-driving cars. The automotive industry is in dire need of a scalable and modular test platform to validate the performance and safety of these cars. The innovative testbed presented by Dr. Heena Rathore and her colleagues at the 13th International Conference on Digital Communication at Athens, Greece, might just have the answer. This testbed has been designed with various general purpose wireless research and applications in mind. Wireless communication is bringing a new level of connectivity to cars. As shown in the figure above, with wireless connectivity, cars may “talk” with each other directly in Vehicle-to-Vehicle (V2V) mode, or through the infrastructure (traffic lights, street signs and such) in a Vehicle-to-Infrastructure (V2I) mode. Such a level of connectivity supports safety, transportation efficiency, and internet access. If you think about it, honking was essentially one driver’s way of telling the others what they are seeing ahead and guiding others to take corrective actions. In a similar way, self-driving cars can be made safer by increasing their sensing range, leveraging what can be seen by other vehicles in the front, in the back, or on the sides. Exchanging such information between vehicles can improve driver assist and full automation over time. Unfortunately, conventional technologies which support data rates of megabits per-second and low-latency messaging, will not be sufficient to support the exchange of high data rate sensor or navigation data. It would be analogus to a driver only giving you one data point out of the 100 that he sees. 5G networks hold the promise to support high data rates and low latency for connected vehicles, which is driving tremendous interest in transportation as a key use case. In particular, 5G operating at mmWave frequencies, is especially attractive because of very high data rates, which can be used for the exchange of raw sensor data, enhancing the safety and efficiency of automotive driving. This would allow vehicles to enhance their situational awareness by seeing many car lengths in different directions, and around corners. Additionally, 5G can support lower latency and ultra-reliability to facilitate distributed control for transportation systems. For example, vehicles can travel together with smaller gaps using platooning, or can be coordinated through an intersection at high speeds without a traffic light. These attributes enable safe operation of connected vehicles in a variety of traditional crash hot-spot situations such as overtaking on rural roads, conflicts at urban intersections, and weaving sections on highways.

Needless to say, transportation is one of the many applications that can benefit from 5G capabilities for latencies, massive bandwidth, and connectivity. 5G focuses on various aspects of present-day communication challenges such as area traffic capacity, network energy efficiency, connection density and latency. 5G also is focused on driving spectrum efficiency and mobility, thus making it a prime candidate for massive machine to machine type applications such as in smart cities and smart factories. Dr. Rathore’s work addresses a key gap in the long-term adoption of this standard for these applications by presenting a modular and scalable testbed architecture to test interoperability and scalability of this standard across the various applications. For example, the ability of her proposed testbed to adapt to various mmWave frequencies is crucial for self-driving cars as it has to work across the cellular, high bandwidth wireless (60G-66G) and evolving vehicular radar frequency bands (76G-82G). Likewise, the ability of the testbed to scale to multiple transmit and receive nodes is of big benefit for health applications. Ability to program at different layers is crucial to test machine learning enabled safety and security of 5G enabled wireless devices. Likewise, the ability to try different algorithms is of immense benefit to general purpose research and energy applications. In summary, Dr. Rathore’s testbed will facilitate the development of numerous 5G applications in the long run. Now circling back to where we started, we hope that in the near future, the “Horn, OK Please” sign on the back of the Indian trucks is soon replaced by a more pleasant “Talk, OK Please” and Dr. Rathore’s testbed will help us get there sooner rather than later.