Mobile World Congress 2019 (MWC 19™, the annual tradeshow put on by the Global Systems for Mobile Communications Association (GSMA), wrapped up last week in Barcelona, Spain. The focus of MWC 19 was 5G wireless communications, and 5G’s effect on technologies and markets that utilize wireless communications. One of these markets was autonomous vehicles (“AV” or “self-driving cars”), and included representation from (and presentations by) auto companies such as BMW, Daimler, Formula 1 Racing, Tesla, Toyota, and Volkswagen.
The impact of AV extends far beyond mobile wireless communications. Tesla recently spoke about their vision for AVs, which Elon Musk has promised to be ready by the end of 2020, and to be so mainstream that by 2037 non-self-driving cars would be as much of a curiosity as riding a horse is today. Apple submitted their voluntary safety report on “Project Titan” to federal regulators, and highlighted machine learning and automation as critical technologies. A recent article by IBM discussed the importance of metadata and data tagging to speed up data analysis, which is obviously critical to safe operation of AVs. Another recent article by Dell in CXO Insights explored the demands that AVs will place on data centers, and how AV-related data centers will have to evolve to keep up with the avalanche of data produced by AVs.
While the Dell article certainly represents one way to look at the AV data problem (at least from the standpoint of cloud data center operators), it misses the obvious point that even with 5G wireless communications it will not be feasible to stream the terabytes of data produced by each AV to a cloud data center. In essence, AVs will have to be “rolling data centers”, and will have to process the data on the vehicle to respond to the safety demands of self-driving vehicles. Since self-driving cars don’t have the power, cooling, or physical space that a cloud data center has, new approaches will have to be utilized to bring AVs to mass production. The technology of computational storage can uniquely help to tame the AV data avalanche. With the ability to rapidly ingest and store terabytes of data, accelerate artificial intelligence/machine learning algorithms, and internally label and sort data, computational storage meets the four requirements that Accenture identifies as critical for AV: data acquisition/ingest, data storage, data management, and data labeling. If you are in the AV business (or related enterprises), feel free to contact me to talk about how computational storage can help you.