Aira Technologies, a pioneer in the application of Artificial Intelligence (AI) to radically improve wireless telecommunications, announced that it has successfully integrated its AI based solution into Intel FlexRAN reference software and CapGemini based DU stack running on a Dell PowerEdge XR5610 server and demonstrated >2x improvement in spectral efficiency. Intel FlexRAN reference software, with a software based L1, provided the flexibility to tap into the right part of the L1 to overlay Aira’s AI-based solution to achieve the spectral efficiency improvement without the need for additional AI hardware like GPUs. AI-enhanced channel prediction helps minimize inter-user interference to maximize multi-user MIMO performance.
Anand Chandrasekher, Founder and CEO, Aira Technologies said, “The application of ML to wireless baseband processing in this fundamental manner is an industry first. Together with Dell and Intel, we are showcasing three very important things – the power of the Open RAN movement to accelerate innovation; the extraordinary potential for ML in cutting edge wireless applications; and, the enormous potential for AI to improve the RAN.”
Ravikiran Gopalan, Founder and CTO of Aira Technologies said, “Radio Access Networks (RAN) operate in varied network conditions and radio environments. ML presents a natural framework to classify these conditions accurately and process the RAN signals optimally for each of these conditions. We are seeing tremendous MU-MIMO throughput gains from our ML-based channel prediction model and we are working on applying our ML framework to other RAN functionalities.”
By directly integrating the Aira AI-based software into Intel FlexRAN reference software running on the Dell PowerEdge XR5610 server, Aira was able to achieve its spectral efficiency improvements without compromising the critical timing requirements of baseband processing. In addition to the full RAN stack, the entire AI workload was processed within the CPU. The CPU supports the AI-powered Aira Insight Engine, which provides more accurate estimation of the SNR, user speed and RF environment. It also supports the AI-powered channel prediction model that uses these estimates to compute better beam weights. The benefit is a greater than 2X improvement in spectral efficiency with no additional hardware.
Also Read: ADLINK and Ataya reveal 5G private network solutions at MWC Las Vegas
“Aira is innovating on Dell infrastructure to improve RAN performance,” said Manish Singh, Chief Technology Officer, Telecom Systems Business, Dell Technologies. “Harnessing the potential of Generative AI to build new AI and machine learning models will simplify RAN operations and provide new efficiencies for network operators.”
Intel’s latest vRAN processors and Intel FlexRAN reference software, along with CapGemini L2 and L3 software platforms, provided crucial building blocks and facilitated easy incorporation of Aira’s cutting edge ML technology on Dell’s Intel-based servers.
“AI and machine learning are poised to enhance the operation of virtualized radio access networks, providing new energy and performance efficiencies,” said Cristina Rodriguez, Vice President and General Manager, Wireless Access Networking Division at Intel. “Aira is innovating on top of Intel’s vRAN platform by applying their sophisticated AI models to the 5G physical layer to improve RAN performance.”
The Aira demo in the Dell booth at Mobile World Congress 2024 also features Aira’s RANGPT, the world’s first LLM-based utility to observe, analyze, control, and automate the RAN. RANGPT, which is hosted in public or private clouds, lets wireless experts analyze data, get insights, iteratively experiment, and deploy code as rApps in hours instead of months. RANGPT will be used to query a network emulator to highlight the performance benefits of Aira AI-based channel estimation and prediction technology.
SOURCE: PRWeb
