KUALA LUMPUR, March 3 (Bernama) -- Mavenir, a cloud-native network infrastructure provider, has announced a major milestone with its full-stack artificial intelligence (AI) framework, Network Intelligence as a Service (NIaaS), marking a global first in autonomous optimisation for operational 5G non-standalone (NSA) networks.
The company’s deep reinforcement learning (RL)-based live AI agents, a core component of NIaaS, are now autonomously managing full closed-loop optimisation for NSA networks, achieving over 40 per cent improvement in key performance indicators (KPIs) within radio access networks (RAN).
Mavenir President of Access Networks, Platforms, and Digital Enablement, BG Kumar in a statement said the company is thrilled to lead the industry with its AI-native NIaaS, which delivers tangible improvements in network performance.
He highlighted that unlike traditional self-organising network (SON) systems, which focus on functionalities rather than KPI enhancements, its AI-driven NIaaS is KPI-centric, using in-depth intelligence to optimise factors affecting KPIs across cells, sites and markets, ensuring measurable improvements in network performance.
This innovation, which only requires central processing units (CPUs) for operation instead of graphics processing units (GPUs), has proven to be both cost-effective and highly efficient. The success was demonstrated in live Open RAN deployments with two Tier 1 5G NSA operators in South Asia, covering over 25 sites and 100 LTE and New Radio (NR) cells.
Mavenir’s NIaaS is the first of its kind to enable large-scale AI-driven optimisation for Open RAN operations and has successfully reduced the rate of unstable handovers, including mobility ping-pongs, too-late and too-early handovers, radio link failures, and wrong-cell handovers, by 40 per cent to 60 per cent, improving the mobility robustness of 4G/5G subscribers in the live networks.
Looking ahead, Mavenir’s NIaaS will continue enhancing other telco KPIs, such as connectivity, coverage, traffic management, accessibility, and efficiency.
While it is currently hosted in Mavenir’s RAN Intelligent Controller (RIC) for Open RAN deployments, it is also available as a stand-alone telco intelligence tool for other domains, such as Packet Core and IMS, and supports third-party network functions.
This breakthrough marks a significant step in the evolution of AI-powered telecom operations, offering operators AI-native tools to improve network efficiency, reduce operational costs, and elevate service quality. Mavenir’s continued innovation in AI and machine learning sets a new benchmark for the telecom industry.
-- BERNAMA
