Northeastern University Unveils Open-Source Massive MIMO AI-RAN Prototype

In a remarkable advance for wireless communication technology, researchers at Northeastern University have successfully unveiled the first open-source prototype of a massive MIMO AI-RAN system. This groundbreaking achievement comes from the Institute for Intelligent Networked Systems (INSI) in partnership with the Open6G OTIC, marking a significant milestone in creating open, AI-native wireless systems. The prototype assimilates ORAN-compliant massive MIMO hardware from AmpliTech Group, open-source software from OpenAirInterface (OAI), and NVIDIA GPU-accelerated RAN intelligence into a singular, comprehensive platform.

The demonstration utilized an Amplitech mMIMO O-RAN Category B radio unit alongside NVIDIA AI Aerial software for layers 1 and 2 of the RAN, merging them seamlessly with OAI’s L2+ stack to create a fully integrated open-source system. Implementing a two-stage precoding architecture, researchers achieved effective MIMO precoding accelerated by NVIDIA's AI Aerial, which delivers remarkable performance. A standout feature is the integration of layer 2 beam management enabling sustained throughput across multiple user equipments (UEs) despite mobility conditions. This fortifies the promise that high-performance, GPU-accelerated AI-RAN architectures can be constructed and replicated without being tethered to proprietary, closed-stack systems.

Massive MIMO technology, noted for its capability to employ large antenna arrays to serve numerous users simultaneously via spatial multiplexing, has traditionally necessitated tightly integrated, vendor-specific implementations. However, this recent demonstration defies that notion, confirming that a comprehensive system—from the physical layer to the RAN control plane—can be constructed from open, interoperable components. The Northeastern Open6G Open Testing and Integration Center recently granted O-RAN certification to the AmpliTech Group mMIMO CAT-B RU, an essential milestone for successful interoperability.

“This demonstration is landmark,” stated Tommaso Melodia, Director of INSI and Professor at Northeastern University’s Electrical and Computer Engineering Department. Furthermore, Melodia expressed, “It illustrates that performance does not need to compete with openness, showcasing how advancements in massive MIMO technology can thrive under such frameworks.”

The architectural innovation of integrating NVIDIA AI Aerial GPU-accelerated L1 processing with OAI's L2+ layer is pivotal, allowing the rigorous physical layer operations characteristic of massive MIMO systems to be executed efficiently on NVIDIA's AI-RAN platform, while ensuring software openness at the upper layers.

Irfan Ghauri, Director of Operations at the OpenAirInterface Software Alliance, affirmed OAI’s long-standing commitment to developing open, standards-compliant RAN implementations. He remarked, “Witnessing OAI’s CU/DU stack integrated into this full mMIMO AI-RAN prototype demonstrates the boundless possibilities forged through collaborative efforts on open grounds.”

The achievement was heralded by Fawad Maqbool, CEO and CTO of AmpliTech Group, who relayed the success of integrating their 64T64R massive MIMO O-RAN radio unit into an entirely open-source, GPU-accelerated AI-RAN system. Maqbool stated, “This accomplishment elevates our profile within the AI-RAN and Open RAN ecosystems, creating opportunities for customer engagement and strategic collaborations.”

Guided by the ONSI team, the integration of the system, testbed configuration, and measurement validations resulted in a reproducible reference implementation that researchers in both academic and industry spaces can build upon.

This demonstration comes at a time when the Open RAN and AI-native wireless systems momentum is surging, emphasizing the essential qualities of flexibility, interoperability among vendors, and intelligent control, which appear increasingly vital for the rollout of future 5G and 6G networks.

About AmpliTech

AmpliTech Group, Inc. (NASDAQ: AMPG, AMPGW) specializes in the design and manufacture of advanced RF and microwave signal-processing components and systems tailored for satellite, 5G/6G telecom, quantum computing, defense, and space applications. Their integrated divisions work collaboratively to fuel innovations in connectivity and communication systems.

About OpenAirInterface Software Alliance

The OpenAirInterface Software Alliance is a non-profit organization dedicated to developing open-source software that adheres to 3GPP standards for 5G and 6G networks. OAI offers a fully open and interoperable RAN and core network stack utilized globally by researchers, network vendors, and operators.

About the Institute for Intelligent Networked Systems

The INSI at Northeastern University acts as a leading research hub focusing on AI-native wireless systems, Open RAN, and advanced 5G/6G infrastructure. As part of the Open6G initiative, it collaborates with multiple partners to prototype and validate advanced network technologies on extensive experimental setups, propelling the commercialization of scalable and secure wireless solutions.