AI is moving beyond centralized data centers and into the physical world, where real-time determinism is emerging as a critical requirement for the next generation of edge computing.

As AI expands into environments such as telecommunications networks, vehicles and industrial systems, machines must sense, reason and respond instantly rather than simply generate answers from static data, according to Paul Miller (pictured), chief technology officer of Wind River Systems Inc. With deployments already spanning tens of thousands of 5G nodes with providers such as Verizon Communications Inc. and Vodafone Group PLC, the shift is driving demand for infrastructure capable of delivering reliable real-time inference directly at the edge, where decisions must happen in milliseconds.

“As you move more toward the edge, you’re moving more toward systems that interact with the physical world,” Miller told theCUBE. “Instead of training large language model-based functions, it’s now inference. It’s [about] the execution of an AI model in a way that interacts with a human being.”

Miller spoke with theCUBE’s Dave Vellante at MWC Barcelona, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the transition from data center AI to edge AI, real-time vehicle-to-everything communication and Wind River’s expansion into the enterprise market. (* Disclosure below.)

Real-time determinism powers edge AI applications

At MWC Barcelona, Wind River and parent company Aptiv Plc demonstrated a cellular vehicle-to-everything system that shares sensor data between vehicles over Verizon’s 5G network. One vehicle detects pedestrians or other objects and sends that information through the network, allowing a second vehicle to receive the data in milliseconds and brake for as-yet unseen hazards.

“We’re sharing sensor data, and that destination car sees that data coming in real time because of the performance of the 5G network, as if it were just new sensors that it had [at its disposal],” Miller explained. “It can see beyond its horizon now. In the demonstration we give, instead of hitting a pedestrian, it’ll actually brake before it even sees the pedestrian, because it’s getting the information from another car. Safety, comfort and convenience — there’s hundreds of applications that leverage this technology and we’re at the forefront of it.”

The real-time determinism required for safety-critical applications, such as vehicle-to-everything, traces back to VxWorks, Wind River’s real-time operating system, which has been in production for 45 years, Miller noted. Unlike general-purpose systems, VxWorks guarantees that computations complete within fixed time windows — a requirement that current cloud AI architectures cannot meet.

“I need to perform some computational problem within a fixed period of time or it’s not going to be sufficient for the application. That requires determination and high performance.” he said. “That determinism and high performance can only be realized if running on a real-time operating system.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of MWC Barcelona:

(* Disclosure: Wind River sponsored this segment of theCUBE. Neither Wind River nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE