- by Markku Räsänen
Cars rolling off the lines today generate up to 25 gigabytes of data per hour. With the adoption of further levels of autonomous driving—up to levels 3 or 4—the amount of data is set to increase to up to 1.4 terabytes per hour. Some of that data will be pushed to the cloud, but much of it will be stored directly inside the car, or “at the edge.” To date, there’s been a lot of discussion on why we need big data to develop cars of the future. But when it comes to the actual storage and preservation of all that data—what are the key driving factors there? One could simply say “more data=more storage,” but that overlooks the core influences behind the adoption of edge storage in vehicles.
Here are the top 4 factors driving requirements for automotive edge storage.
Safety regulations continue to develop in a direction where more data is preferred in order to analyze after crashes. For example, California requires collecting 30 seconds of pre-crash data from autonomous cars during driverless testing. But the type of data that should be stored is currently unclear. Because of this uncertainty, a lot of potentially unnecessary data must be recorded. So that 30 seconds of data can amount to hundreds of gigabytes of data if the car is operating at autonomy levels 3 or 4. Storing all this information will happen in-vehicle no matter what—because in order to meet the requirement of this regulation, no data should be lost or corrupted.
2. Technology challenges
Data should be analyzed in a centralized way. Most people think this should be happening in the cloud. However, 5G is not yet here, and it will not be for a very long time. This makes cloud connectivity a limiting factor. Even when commercially available, 5G will not be a viable solution outside of densely populated areas. In-vehicle data will be more extensive for the foreseeable future compared to the data in the automakers’ servers. As such, edge storage adoption is driven by the need to also perform calculations and data analysis inside the vehicle.
3. New business models
Automakers are not yet on par with traditional technology companies, but they are rapidly transforming. Collecting and analyzing data inside the vehicle will help develop new business methods tied to things such as overall mobility of people, vehicle ownership, and municipal infrastructure. Due to the challenges and limitations discussed above, this is also a strong driver of edge storage. It’s just not feasible to send all the data to the cloud immediately, or even send all types of data. Still, that data should be analyzed and stored for future use.
4. Machine learning on a fleet scale
More data collected and analyzed means better performing autonomous cars, both individually and as fleets of vehicles. In a fleet scenario, each vehicle’s data can contribute to the overall performance of the fleet. Mistakes by the autonomous driving system in such a fleet should only happen once. After which, the whole fleet is updated based on the learnings of that particular situation and that particular vehicle. In order to make these self-correcting fleets, OEMs should collect much more data than the bare minimum required by regulators. This will result in increased data bandwidth and storage requirements.
With these drivers in mind, the role edge storage will play in the foreseeable future will be important. Storing this much data and analyzing it in-vehicle is not just a hardware problem, but also requires software solutions. This is because flash memories wear out over time due to use. The more often the memory is accessed, the more it wears. A proper software solution will help increase the memory lifetime. Companies like ours help automotive companies solve these issues on the software side, alongside all the hardware vendors providing the physical medium to store all of this data. In this way, car makers can focus on the benefits of all this data that’s stored and collected—making our rides safer and more comfortable, efficient, and enjoyable.