Basic Knowledge Basics of autonomous driving - Part 4
In the fourth part of the essential series on autonomous driving, V2V, as well as V2I communication, is examined, and the most promising communication technologies are presented.
If self-driving cars are going to be on public roads in the future and realize the numerous potential benefits they promise, it is essential to take a comprehensive look at the communication aspects. While the integrated sensors presented in the last article are crucial, they can only provide part of the information about a car's environment. A detailed map of the environment is required so that the vehicle can independently decide on the right path to take. This can be generated using information from other road users and nearby infrastructure.
Both Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication is based on either wireless or radio systems. Which of the different protocols is best suitable will be shown in the future. An ideal solution must offer high-speed data transmission (both receive and transmit) and low latency to increase responsiveness and provide optimal safety for all road users. High levels of security are essential to prevent vehicles from falling victim to hackers.
Car manufacturers, tier-one suppliers, semiconductor companies, and telecommunications companies are currently developing various types of V2V and V2I technology that will be integrated into the next generation of vehicles. The most promising candidates are presented below.
Dedicated Short Range Communication (DRSC)
Dedicated Short Range Communication (DSRC) is currently the best-known method for communication between vehicles. It has been used in the automotive sector since 1999 and is based on the 802.11p protocol for wireless connection. Dependent on the country, it runs on the 5.8 GHz or 5.9 GHz band and has a range of up to 300 meters. Thanks to its relatively low latency of about 5 ms, it offers significant safety benefits compared to other communication technologies.
Companies such as Toyota, General Motors, and Ford have looked at DSRC as a cost-effective solution for their communication needs. Although it is easy to understand and inexpensive, the technology has some disadvantages that limit its potential use. The frequency bands for DSRC differ in different countries, so integration is a more significant challenge for the global market, as almost the same cars are sold in Europe, Asia, or the USA.
Security is also an issue, as data is encrypted. Still, DSRC is vulnerable to jamming, false alarms, and man-in-the-middle attacks (where hackers modify the communication between two parties). It also supports low data rates (measured in Mbps). At the same time, prototypes of autonomous vehicles are already operating in the Gbps range for the information captured by their sensors alone (like LiDAR image processing). DSRC is, therefore, only of limited appeal for the future.
5G provides a solution
Numerous car manufacturers have completely disregarded LTE 4G technology because, despite the infrastructure already in place, it does not provide the power needed to support autonomous vehicles. With peak data rates of only 300 Mbps and an average latency of about 50 ms, LTE networks impose severe operational constraints. Instead, most companies are focusing on the upcoming 5G networks as the fundamental technology for V2V and V2X systems.
Special interest groups, such as the 5G Automotive Association, bring together automotive manufacturers, suppliers, chip manufacturers, and telecommunications network operators to jointly develop technology for automotive applications. The advantages of 5G are distinct: incredibly high data rates, some of which are well over 10 Gbit/s, exceptionally low latency of only 1 ms, and the ability to use existing infrastructure.
Speed is the key. If you download 4.5 GB of data (for example, a movie) over your DSL line at home at an average rate of 50 Mbps, it takes 13 minutes. Over a 5G connection, the same download is possible in 4 seconds. Even the amount of data required for self-propelled vehicles can flow freely with 5G, without the need to buffer or prioritize individual data packets. Probably the most significant advantage of 5G, however, is its extremely low latency - an essential aspect for any autonomous operation.
The time a human being needs to detect and react to an incident on the road is about 1 second. If he's in a car traveling at around 100 km/h, he covers a distance of approximately 28 meters before stepping on the brakes and slowing down. A networked vehicle with a 5G high-speed connection could react 1,000 times faster and start braking after just a few centimeters.
There is still a long way to go, particularly in terms of safety. As an example, the British government has defined four areas that still require special attention.
- Cross-layer safety: A uniform framework is needed to coordinate different safety methods for each safety layer.
- Cross-domain security: 5G networks generate a large number of different use cases. Therefore, a collaboration between groups is required to integrate security solutions across domains.
- End-to-end security: A secure connection for communication between the user and the core network is essential.
- Integrated security: As the network evolves, security mechanisms must be integrated into the design from the outset.
Requirements for network coverage
Perhaps even more importantly, as networked vehicles become mass market, better network coverage is essential. In rural areas, mobile phone coverage is often insufficient. This may be frustrating, but it does not pose an immediate danger. On the other hand, if a networked autonomous vehicle that relies on external information fails to receive a signal on a busy or complicated stretch of road, the consequences can be fatal.
Communication networks of the future must, therefore, not only be extremely reliable and offer a high degree of redundancy. They must also be expandable to cover larger geographical areas - including communities where good 4G coverage is currently not even available. Until that happens, vehicle autonomy will probably only be relevant in urban areas.
This article was first published in German by next-mobility.news