RICHARDSON, Texas, Feb 28 (Bernama-GLOBE NEWSWIRE) — Mavenir, the cloud-native network infrastructure provider building the future of networks, today announced a groundbreaking achievement with its full-stack AI framework, Network Intelligence as a Service (NIaaS). For the first time globally, deep reinforcement learning (RL)-based live AI Agents – an integral component of Mavenir’s NIaaS – are autonomously controlling and performing full-closed-loop optimization for operational 5G non-standalone (NSA) networks, delivering over 40% gains in key performance indicators (KPIs) related to mobility in radio access networks (RAN). The significant performance gains were also cost-effective, since NIaaS runs only on CPUs in the field rather than GPUs. Hosted in Mavenir’s RAN Intelligent Controller (RIC), NIaaS has proven these milestones in live Open RAN deployments with two Tier 1 5G NSA operators in South Asia, with each deployment corresponding to 25+ sites and 100+ LTE and New Radio (NR) cells.
Presenting the first-of-its-kind telco AI capability for large-scale Open RAN operations, Mavenir’s NIaaS has significantly reduced the rate of unstable handovers (including mobility ping-pongs, too-late handovers, too-early handovers, radio link failures and wrong-cell handovers) by 40% to 60% for the aggregate KPI and individual KPI components, thereby improving the mobility robustness of 4G/5G mobile subscribers in the live networks.