Ericsson unveils AI-ready radio and RAN portfolio

The rapid global adoption of AI-enabled devices is placing new demands on mobile networks, particularly on uplink performance and latency. Use cases driven by multi-modal artificial intelligence and augmented reality require networks to handle significantly higher data volumes and more complex traffic patterns.

Ahead of Mobile World Congress in Barcelona, Ericsson has announced a new suite of radio, antenna, and AI-powered RAN software solutions to address evolving requirements.

AI-Ready RAN Portfolio Targets Uplink and Latency Demands 

The company said the new portfolio is designed to help communications service providers adapt their networks to support AI devices and new service models. It includes AI-ready radios and Ericsson Silicon equipped with neural network accelerators to enhance AI inference capabilities in Massive MIMO radios.

These accelerators are programmable matrix cores integrated into Ericsson’s Many-Core Architecture and optimised for AI and machine learning workloads.

The portfolio is structured around three areas: radios, RAN software and antennas.

In the radio segment, Ericsson is introducing ten AI-ready radios, including next-generation Massive MIMO and remote radios developed to improve downlink efficiency and uplink capacity.

The portfolio includes high-power FDD Massive MIMO solutions such as AIR 3286 and AIR 3211, which combine Massive MIMO on both TDD and FDD. Additions to the 8-receiver portfolio, including Radio 4891 and Radio 4458, are intended to enhance uplink performance for AI and AR applications.

The range also features high-power triple-band radios, Radio 4488 and Radio 4464, designed to support network consolidation and RAN sharing. Further TDD Massive MIMO developments include AIR 3267, offering 600 MHz instantaneous bandwidth in a 13 kg unit, and AIR 6492, with 480W output power and 256 antenna elements.

On the software side, Ericsson has introduced new AI-enabled RAN functionalities to improve efficiency, reliability, and latency. These include AI-managed beamforming, AI-powered outdoor positioning, and an AI-based coverage prediction model that complements the company’s AI-native link-adaptation software.

To support applications requiring predictable latency and reliable uplink speeds, additional features, including a latency-prioritised scheduler and low-latency mobility tools, have been added. The company states these tools can significantly reduce response times and improve connection stability.

The third element of the portfolio comprises five new antennas, including interleaved and passive models designed to enhance spectrum utilisation and simplify site design.

The energy-efficient passive antennas are based on a trio net design to improve uplink performance, carrier aggregation and spectral efficiency. The interleaved AIR portfolio has also been expanded with three additional configurations to support both TDD and FDD Massive MIMO deployments.

Mårten Lerner, Head of Networks Strategy and Product Management at Ericsson, said that as AI changes traffic patterns and user expectations, networks must deliver performance where and when it is required. He added that the company is integrating AI capabilities across its portfolio, including the introduction of neural network accelerators within its Massive MIMO range.

Ericsson said the expanded portfolio integrates AI more deeply into the radio access network, enabling real-time optimisation of performance and energy consumption. The company added that AI-driven efficiencies across hardware and software are intended to reduce the total cost of ownership and support faster deployment of new services.

Iain Milligan, Network Development and Infrastructure Director at VodafoneThree, said the integration of Vodafone and Three’s networks into a single 5G infrastructure powered by Ericsson’s RAN is intended to support AI-driven applications while improving operational efficiency and energy use.