Close
Close

Published

Car2MEC pilot winds down, cars as data centers look a bit slow

Bavaria’s Ministry for Economic Affairs funded a twelve-month project to test how Multi-Access Edge Compute (MEC) technologies could be used in automotive applications, specifically, in vehicle-to-vehicle communications. However, the project missed its 20ms latency target, which doesn’t bode particular well for the ‘cars as mini data centers’ crowd.

The trial involved Continental, a major (but rather troubled) automotive OEM, Deutsche Telekom, Nokia, and research and testing institute Fraunhofer. It was looking to test both the technology and the business case, to see if this was something you could actually make money from. In terms of architecture optimization, the project managed to get latency down to 30ms, which is impressive for mobile networks, but still short of the 20ms target.

It was carried out on the A9 autobahn, a 529km road that runs north-south and links Berlin to Munich. The project was initially established back in 2016, but the past year was the main focus of testing. It saw Deutsche Telekom and Nokia deploy two MEC instances, which are essentially cloud applications running at the network edge, with Continental and Fraunhofer handling the in-car side of things. Porsche-subsidiary MHP, a consulting firm, was also involved in assessing the business case. Moving the applications closer to the cars was necessary to bring the latency down, as a round trip from the car to a cloud application and then back to the car would take far too much time for a lot of the desired functionality.

These functions included emergency warning systems, traffic alerts, speed limit assistants, and high-definition mapping. These were served to the test cars over the existing LTE network, from the two edge-compute locations. However, what is not clear is the potential latency reduction that new 5G technologies might provide, over their older 4G counterparts. It might be enough to get to the 20ms target, but if the bulk of the current 30ms can be attributed to the MEC hardware’s decision making, then it doesn’t really matter what the radio-access protocol is – it will still take too long.

 

Now, Fraunhofer sounds pretty confident that 5G will solve that problem, but we will withhold judgement until the next text can confirm this. Sure, 5G should be faster, but the main issue is not the speed of 5G data transfer, but rather the speed of 5G operator deployments. There’s no point having a blazing fast MEC application if the network doesn’t cover the whole of the road.

“While the edge cloud improved end-to-end latency by more than 20ms, the 4G radio access network is still a limiting factor for more demanding future services,” said Karsten Roscher, senior scientist at Fraunhofer ESK. “However, we believe the combination of edge cloud and 5G will shape a completely new landscape of connected and distributed applications.”

So, if 5G radios solve the latency problem, then the industry is still going to be waiting on the MNOs and infrastructure vendors to deploy the MEC installations near the roads that you want to canvas. This means that these V2X deployments are going to be beholden to the commercial interests of this ecosystem, and the MNOs, automakers, and public bodies are going to have to thrash those details out.

V2V might be the better choice here, but then we’ve strayed past the MEC scope. Sure, the self-driving cars of the future are going to have heaps of processing power, but not enough to really call them ‘data centers on wheels,’ as the refrain goes. Direct V2V communication would facilitate cooperative driving features, but until you move the MEC functions completely into the cars themselves, slinging the refrain around is somewhat misleading.

In Europe, there’s still consternation over the use of LTE in V2X communications. The European Union favors ETSI ITS-G5, which is in the same came as the IEEE 802.11p that is used in the US (in the FCC’s DSRC band. However, the GSMA has recently come out against the EU’s position.

“This piece of legislation relies on a biased view of technology and impedes innovation,” said Afke Schaart, VP & Head of Europe of the GSMA. “If the EU stays on this road, it will isolate itself further in the global 5G race and severely harm 5G investment in Europe.”

The counter view was that “this decision gives vehicle manufacturers, road operators and others the long-awaited legal certainty needed to start large-scale deployment of C-ITS services across Europe, while remaining open to new technology and market developments,” said the European Commissioner for Mobility and Transport, Violeta Bulc. “It will significantly contribute to us achieving our ambitions on road safety and is an important stepping stone towards connected and automated mobility.”

Fraunhofer seems to have the view that both technologies can be used alongside each other. It notes that ITS-G5 doesn’t require new infrastructure, as it is a V2V technology, but that the cellular approach (C-V2X) would be the link to wider cloud and edge applications. To this end, a pure ITS-G5 approach would be too insular, while a pure C-V2X one could prove too slow for the desired safety-based features.

Olaf Kleindienst, partner at MHP, said “everybody is talking about the connected car and, in some ways, the connected car is already there. But it still has some way to go and many questions remain, particularly around the fully connected car. The fully connected car represents a new environment for the automotive industry and the established key players must reinvent themselves, with new business models. We see projects like Car2MEC as an excellent opportunity to think ahead and discuss questions around this new environment at an early stage, with different perspectives.”

Close