Friday, February 23, 2024

Aira Technologies demonstrates RANGPT™, the world’s first LLM-based utility for RAN query and control

Related stories

spot_imgspot_img

Aira Technologies showcases a new LLM-based capability for securely querying and controlling the Radio Access Network using the power of Generative AI

Aira Technologies, a pioneer in the application of Machine Learning (ML) to radically improve wireless telecommunications, announced that it has developed a new capability, RANGPT, that allows Mobile Network Operators (MNOs) to query and control the RAN using conversational language. Aira is demonstrating RANGPT at the first Aira Technology Day, co-hosted with its partners in Redwood City, CA. The partners are leading AI and telecommunications practitioners who have gathered for a full day of panel discussions on how AI could revolutionize wireless networks.

With 5G, the increased complexity of mobile networks poses many challenges as MNOs try to balance performance, sustainability and economic efficiency. Telco networks generate enormous amounts of data and operators recognize that this is a gold mine of information. This data can be used to better design the networks, improve user experience, lower the cost of running the network, and even drive new revenue. However, doing useful experiments with this data has been tedious and time consuming since it requires expertise and coordination between multiple teams such as wireless and data science groups.

“Quick experimentation is impossible and ideas get killed in the process. Aira addressed these challenges by creating RANGPT to provide a natural language interface to network data,” said Ravikiran Gopalan, co-founder and CTO of Aira Technologies. “With RANGPT, a wireless expert can analyze data, gain insights, iteratively experiment, and ultimately deploy code as rApps in a matter of hours—a process which previously would have taken months.”

RANGPT allows MNOs to interact with the network through conversational queries, to gain insights on the state of the RAN by analyzing network data. These insights help address performance issues and optimize energy consumption. A sequence of RANGPT queries and control instructions can be stitched together easily, to form the basis of RAN automation applications required to meet network opex reduction targets.

Also Read: Yardi Announces Virtuoso, A Human-centered Approach to AI

To build RANGPT, Aira developed core technology models that work with publicly available Large Language Models (LLM) like GPT4 from OpenAI, Claude from Anthropic, or Llama from Meta. In response to a query, the Aira RANGPT modules work together to analyze network data resident in a data lake. The result of the analysis is rendered back to the user in any convenient form, including text, graphics and charts. RANGPT retains the context of queries to allow for ever more sophistication in successive probing.

The utility also can be used for control of the RAN, allowing for manipulation of the actual underlying hardware to troubleshoot or isolate faults. Query and control features, coupled with the ability to generate software code as required, means that RANGPT can serve as an automatic code generator for new xApps and rApps.

RANGPT is built with security and privacy of MNO Key Performance Indicator (KPI) data in mind. The LLM used can be a proprietary capability, residing in a private cloud, so the MNO has full control of all the RANGPT modules.

“Aira is built on the fundamental premise that the application of ML to all layers of the RAN infrastructure stack can deliver breakthrough performance benefits to the operator community,” said Anand Chandrasekher, co-founder and CEO of Aira Technologies. “What we have achieved here showcases the extraordinary potential for ML in cutting-edge wireless applications, offering accessibility without compromising control or security. We look forward to ushering in the era of AI defined networking to make cellular networks significantly more efficient.”

Attendees at the Aira Technology Day event saw a demonstration of RANGPT running on actual network data for an unidentified MNO. The system was able to answer questions related to spectral efficiency of different cells and take certain actions. Successful demonstrations such as at Aira Technology Day underscore the progress and enthusiasm that Generative AI is starting to create in the operator community.

SOURCE: PRWeb

Subscribe

- Never miss a story with notifications


    Latest stories

    spot_img