If you work in Automotive or Manufacturing, Edge computing is naturally a hot topic for you. However, if you work in a global bank, an insurance company or in retail or fashion you might wonder why on earth should I care about Edge/IoT/and Edge processing power?
In the near future you will face a countless amount of integrated use cases where machines, sensors and services need to talk to one another with some type of human interaction required. Your central platform for continuous contextual authentication must be capable of supporting these types of integrated use cases end-to-end, from Edge to cloud. Otherwise you’ll end-up with an architecture that has even more silos than likely already exist.
You don’t think so? I bet my brain against a peanut that you will in the near future (and my brain is rather essential to me!).
At Transmit, we are already receiving requests that confirm this. One of the largest banks in the world has a need to support continuous contextual authentication for sensors. The sensors are attached to containers and are used to monitor the fully automated delivery of goods. In an automated world the payment for goods in international trade is being processed only after the delivery is in a defined target harbor/airport/etc., without the need for any human manipulation or intervention.
(I)IoT requirements are omni-present in automotive and manufacturing. In an automotive context you will have (partly) self-driving cars communicating with dealers to make service appointments or order spare parts just-in-time on your behalf. In the near future, only certified employees within a dealership will be allowed to service electrified or autonomous driving vehicles. Certain updates will also be possible remotely or over-the-air, while others will still require an on-site visit. Who manipulates what at each point in time must be monitored to ensure vendor guarantee requirements are being fulfilled.
Even in fashion, retail and sports equipment, sensors are being built into footballs, basketballs, skis, shoes, shirts and other gear for personalization of equipment, insights into performance, in-stadium payments or a personalized, digitally enriched retail experience.
Another entirely different use case is for going password-less in the enterprise. And not just for employee identities, but also for all service-to-service communication.
So how are Edge devices and the data they produce handled today? Very simply. Today, only data generation happens on the Edge. Over 90% of data produced on the Edge is being sent from Edge Devices to a central enterprise cloud of some sort. Data is being managed, aggregated, analyzed and distributed exclusively in that central cloud data-hub.
In the example of a connected car, the car with its smart and independently communicating head-units (engine, brakes, etc.) produces a huge amount of data every day. Data on road quality, driving behavior, usage of individual components, etc. is being gathered. The car sends all this information to a central data-hub where the data is anonymized/ manipulated/ aggregated and prepared for consumption. Some applications need engine usage data, others road quality data that’s aggregated or partly aggregated, etc. Data consumers must take user consent into account that is often given on the Edge via the infotainment system (GDPR/PSD2 compliance).
While this set-up might be sufficient now, it will neither be sufficient nor desirable going forward given the important changes that are happening in parallel: On the one hand we see an exploding amount of (I)IoT/OT devices in use and on the other hand prices for Edge Hardware are dropping drastically. Formerly very dumb Edge devices or sensors can now be “upgraded” with significant computing power/intelligence for very little money.
Over the next couple of years only 25% of Edge data will be sent directly to the Cloud according to Gartner. 75% will be managed, aggregated and pre-processed on the Edge. Underlying trends observed by Gartner in automotive are the rise of autonomous driving vehicles (Gartner predicts a rise from less than 1% today to over 10% by 2025) and the usage of digital twins in manufacturing (I)IoT/OT projects (Gartner predicts 24% of all enterprise IoT projects to use digital twins by 2025).
Edge empowerment will be a key trend across all industries for the next couple of years. Beside the above-mentioned benefits, Edge intelligence can also be used to address many privacy concerns. Personal data pre-processing on the Edge before distribution to the cloud makes data leaks in the cloud less painful but requires Edge security to improve significantly beyond current methods.
Finally, Edge pre-processing is very resourceful as it decreases the need for bandwidth and cloud infrastructure and hence contributes to a greener IT.
Will the cloud be replaced entirely by Edge? I think not. 100% decentralization will result in (data-) anarchy and chaos with data inconsistency and no possibility for central governance of any sort. Sounds maybe like a problem Blockchain could solve? This is a topic for another post…