The Linux Foundation’s networking arm released an open-source framework centered around autonomous networks.
OpenAN (Open Autonomous Works) was formalized by the LF Networking (LFN) group following initial development by Huawei and China Mobile. Other telecom players represented in the venture include Orange, ZTE, and Thailand’s AIS.
Software vendors Infosys and AsiaInfo are also OpenAN members.
According to the OpenAN website, the framework promises a shared roadmap toward Autonomous Networks Level 5 (L5), as defined by TM Forum. The architecture sees a multi-agent collaboration framework around the operations support system (OSS) and element management system (EMS) functions. OpenAN sits between the service layer and the network layer, driven by the newly developed A2A-T (Agent-to-Agent for Telecom) protocol. The software development kit (SDK) extends Google's much-adopted A2A protocol to enable cross-layer and cross-domain integration and deployment of telecom-specific AI agents.
Is Huawei framing itself as the Google of telco AI agents?
Talking autonomous networks with the Chinese giant's ADN chief
A2A-T was launched by Huawei at this year’s Mobile World Congress (MWC) in Barcelona. In an interview with SDxCentral following MWC, Sam Wang, general manager for Huawei and a founding member of the vendor’s autonomous driving network (ADN) unit, touted the release as providing unified interaction architecture for agentic collaboration, addressing telecom operator challenges in collaboration, reliability, and security within autonomous networks.
“We’re building in features like security, a controllable negotiation process, as well as features to support collaboration, for example, fault optimization,” Wang said.
According to OpenAN literature, A2A-T adds extensions including a structured task schema (Task-T); network event subscription and publication (Notification-T); inter-agent arbitration (Negotiation-T); and security (Authorization-T).
Agents and autonomy
In the Huawei and OpenAN vision, agents become an integral part of autonomous networks. The LFN venture highlights a “Home Broadband Complaint Handling Agent” which recalls a workflow from the orchestration center; retrieves the matching AI agents from the registry center, and coordinates them over A2A-T.
Different complaints trigger different workflows; for example, a billing issue invokes the customer-care and billing agents, while a content delivery network (CDN) issue invokes customer-care and CDN-centered agents.
The group aims to launch scenario solution packages with more membership announcements in December’s TM Forum IHPP (Innovation Hub Pioneer Project) event. Commercial verification is expected in 2027.
As recently explored by SDxCentral, most telecom agent usage currently revolves around back-office examples such as IT service desks and energy bill analysis. According to ABI Research, that picture is likely to change, with analysts forecasting the 5G radio access network (RAN) automation market will reach $6.6 billion by 2029, while open RAN automation will account for around 42% of that revenue by 2030.
Comments