AI agents clash in telecom networks raising chip design challenges

When AI agents fight

Implementing multiple AI agents in telecom networks is leading to unintended conflicts as models attempt to balance requirements differently. This issue is influencing the design of next-generation chips with native AI capabilities for 5G and 6G networks.

The Conflict Problem

Dan Warren, director of communications research at Samsung R&D UK, highlighted a real-world example: "We had two AI agents implemented on Samsung products and they caused problems in the network. We had one agent optimizing the air interface and another for load balancing." The first agent pushed devices off a band to switch it off, while the second pushed devices back on to balance the load.

AI agents in the RAN

Rob Curran of Appledore Research noted: "Everyone is aware of the AI conflict issue and it causes people to stumble. Different radio front ends, some optimized for power efficiency, some for performance."

Implications for Network Architecture

This conflict has significant implications for:

Network architectures
Next-generation chip design
Embedded AI implementation

UK chip designer RANsemi is developing specialized chips for the radio access network (RAN) with built-in AI accelerators. Doug Pulley, CTO of RANSemi, explained: "The thing is particularly with embedded AI you need to start to think about whether there's the right amount of compute, interfaces to the companion host processors."

Cloud vs. Embedded AI

Key differences exist between:

High-level cloud AI agents (millisecond latencies)
Embedded AI (microsecond requirements)

Oliver Davies, VP of marketing at RANsemi, stated: "Baseband processing at the RAN edge faces stringent technical constraints, including ultra-low latency, tight synchronization, and high power efficiency."