Are there any lessons to be learned from self-driving cars?
Raja Tadimeti, Co-Founder & Chief Scientist @ WiteSand
Tesla CEO Elon Musk recently unveiled a humanoid robot called the Tesla Bot that runs on the same AI used by Tesla’s fleet of autonomous vehicles. It embodies innovation across multiple technologies. While many may reap the benefits of these innovations, one theme kept prompting me to write this blog.
While the industry has speculated for some time about self-driving networks, the realization has historically been limited by technology and ideas that predate the cloud. The emergence of self-driving cars sparks the imagination and inspires new exploration.
As I look back, I could easily map the progress made with the evolution of cars to the next step in the evolution of the network. Our mission at WiteSand is to make self-driving networks a reality.
Manual to Automated Controls
Shifting of the gears, locking the doors, turning on the headlights while limited alerting such as fuel capacity warnings. Compare these to the network devices being managed through command line interfaces with limited visibility and a lot of manual controls. Every person operating the car or network has the exact same prescribed procedure for these manual controls. Automation of this prescription is the key here.
Independent vs Integrated Gadgets for Feedback into the AI Machine
Deployment of sensors in cars has become prevalent. Infrared sensors, photo sensors, and backup cameras were introduced to alert the driver on hazardous situations while backing up the car or changing lanes. The feedback signals received from these integrated gadgets further allowed us to communicate the “intent” of parking the car, summoning the car, or request to change lanes.
Over the years, various sensors are employed in the network such as configurations, logs, usage, flow records, alerts but use of independent networking appliance tools – many instances of those installed at various customer locations, makes it really hard to correlate and learn.
The power of AI is realized when it is learning every day across multiple locations with differing patterns. Of course, it was not possible without the advent of the cloud.
Access Control and Personalization
Secure keyless entry combined with automatically adjusting seat positions, rear view mirrors, music, and more is the experience cars deliver today. WiteSand’s integrated NAC identifies the user or devices connecting to the network and adjusts the network experience, while making sure zero trust is not sacrificed.
Over the Air Firmware Upgrades
Advanced auto software is now automatically upgraded behind the scenes, with cars becoming smarter every day or week. The network industry is historically wary of firmware upgrades, partly because we haven’t been able to build that trust that it will work every time.
With microservices-based SaaS solutions, we are able to provide continuous upgrades. Not only can it deliver critical fixes immediately, but also minimizes the spread of disparate and mismatched firmware versions.
Data Processing at the Edge Cloud
Cars are now able to provide historical data, including driving history, mileage, charging statistics, and even dash cam recordings, that is used across AI/ML, alerting, and forensics. The vast amount of data, coupled with the limited processing capabilities in cars, gave impetus to the development of the edge cloud for this application.
The same principle, collecting real time signals and processing them in edge clouds to improve operational efficiency, is central to the WiteSand solution.
The auto industry is advancing rapidly towards the eventual goal of fully autonomous vehicles. All the sensor data is intelligently processed in real time using AI/ML algorithms to adjust operations in a tight closed loop, in order to provide a secure and optimal journey. Information across various cars on the road is leveraged to identify traffic congestion and reroute the vehicle.
Just as self-driving cars are becoming a reality on the road, we are on the verge of enabling self-driving networks in the enterprise.