There is a common misconception that AI at the edge replaces AI in the cloud and vice-versa.  When you look at the challenges that each of these two approaches has independently, it becomes clear that they need each other. Let’s talk about the cloud first.

There are now dozens of cloud providers and thousands of software companies creating innovative ways to turn data into “actionable insights”. These solutions usually live in the cloud which allows them to easily scale the solution across multiple clients.  These cloud applications also require large amounts of data usually generated from multiple sources. Getting this data out of an IT system has its challenges but it is manageable. But what about the OT systems like the fire alarm, energy management, lighting, security, or elevators just to name a few? Some use a gateway and start collecting a stream of data from a system, but how much of that data is really needed, and how much does your cloud provider charge you to ingest it? The next step comes when it’s time to put those “actionable insights” to work. Some cloud applications are not equipped with bi-directional communication between building systems, and some data sources are not designed to be written back to. This can make generating ROI difficult when you can’t implement all the strategies discovered by the software. This leaves us with two challenges:

#1: Getting OT data out of a building is difficult and can be expensive

#2: Real-time control from the cloud is not always practical

 Next, let’s talk about the edge and how edge servers can address the challenges. When we use the term edge, we are commonly referring to a location close to the source of the data or activity we are interested in. Cloud applications have eliminated much of the need for on-prem application servers, but building system manufacturers like to keep their applications local. This opens the door for low-cost, easily deployable edge servers designed to collect and manage this data at its source. Edge servers can interface directly with existing systems at the hardware level to collect and monitor data in real-time, but there are some limitations. Edge servers lack the AI and ML stacks found in the cloud. Although they can run applications, they lack the sheer computing power and scalability found in the cloud.

It’s easy to see how combining these two approaches creates a mutually beneficial outcome. Edge servers acting as a “data broker” or “independent data layer” can backup and normalize data to make it available to any cloud application or resource. They can also intelligently filter out unwanted data while managing polling intervals. This reduces traffic on the local network and eliminates “waste data” from being sent to the cloud. Edge servers can also run local control strategies that are supported by the AI/ML algorithms in the cloud. The edge server can make changes to the local systems based on continuous feedback from the cloud, all while monitoring results locally, and in real-time.

When you combine the power of edge computing, with the power of AI and ML in the cloud, it becomes an accelerator for innovation that brings us closer to achieving long-term, sustainable outcomes that can change the way we operate buildings.

Contact me to learn more about edge server technology for smart buildings, or click below:

https://www.linkedin.com/smart-links/AQGZexv8JGR3bg