Connected Cars Facing a Data Log-Jam

Nathan Eddy

---

17th June 2020

Vehicles are becoming increasingly connected and different stakeholders are looking to benefit from the information that this data provides.

One of the key challenges is the ensuring that the data is being delivered to its intended location and that delivery of information, be it to the vehicle or from the vehicle out to the network, happens without delay. Krishna Jayaramen, head of connected vehicles, mobility for Frost & Sullivan, explained that current data demands for connected vehicles is not a major issue, however the evolution of autonomous vehicle capabilities will present a major challenge.

He said: “The scenario is completely different—we could be talking about terabytes of data. Most traditional carmakers are still collecting massive amounts of data over the cloud and that’s why the data sets that could be anywhere from between 1-10 terabytes. That’s data from cameras, LiDAR, and radar and the data jam issue pops up more with a real-time data feed we need with AVs.”

It will be necessary to consider concepts like ring buffers or data compression technologies, whereby electronic control units (ECUs) would be updated to receive OTA updates, which could include airbags or the vehicle’s suspension. Jayaramen explained: “You can’t really send that data in bulk, because there is a cost association. The biggest cost is movement of data, it’s not really processing of data. What happens is there is a technique to collect bits of data and push it out of the car and is expanded once it reaches the server.”

He pointed out the data jam could happen within the next three to four years as automakers unify connectivity and autonomy to a more holistic degree. “There’s a lot of legacy systems out there,” he said. “We are trying to bridge the gap, and they’re still finding ways to bridge those gaps.”

Other issues that will have to be addressed are the potential data log-jams that could arise based on location-specific circumstances, such as rush hour traffic jams or load balancing if there is an event where potentially tens of thousands of connected vehicles will be providing data. “Stakeholders are not interested in the raw data itself, but the insights that can be derived from it,” Sam Barker lead analyst at Juniper Research. “By placing the processing of this data closer to the edge of the network, and closer to the road network itself, there is no need to send the vast quantities of data generated by connected vehicles over networks.”

He noted this enables only the insights to be sent, which will in turn reduce strain on the network, and also pointed out buffering issues shouldn’t be an problem, as the data that is needed most of the time is not live. “Network providers will have to account for the peak demand for network bandwidth, by ensuring that its network capacity can handle local circumstances like rush hours and special events that could lead to increased traffic or traffic jams,” Barker said. “We expect areas of road networks that are more heavily used to command greater investment of edge networking capabilities.”

The design of 5G networks will also be a key factor, with caching technologies implemented in 5G data centers to allow a super-low latency. “This means that connected cars will be able perform requests for information, based on certain criteria, with an almost instant response,” said Lars Larsson, CEO of Varnish Software. “This will be critical to their wider adoption, not only for the data requests that inform their driving but also the in-car entertainment now all passengers are in the ‘back seat’.”

Larsson explained in shifting to 5G, rush hour and traffic jams should not have any adverse impact on the performance at all and there should be more than enough capacity to serve even the most traffic intensive areas if the network and the caching layer is designed correctly.

In the countryside and other places outside of built-up urban areas, the circumstances may be slightly different. There, low band frequencies will most likely be used to be able to have longer reach, up to 18 miles from the antenna. “However, this brings with it a slightly higher latency and lower bandwidth that could impose limitations for some connected car applications,” he said. “Buffering problems could still be caused by network congestion.”

To address the first issue, Larrson said he is seeing more and more providers of streaming services, for example, adding their own content distribution servers, points-of-presence where a public content delivery network, who provides a shared service, is unable to provide the quality required for the viewers. “It will take some time before this network is built out properly, though,” he said. “However, I am sure car manufacturers will have no problem inventing applications that will use these features to make our car journeys safer and more enjoyable.”