Ericsson presents Explainable AI, new capabilities in its Cognitive Software offering that significantly improve the ability to quickly create value through the adoption of artificial intelligence solutions in network design and optimization.
Thanks to Explanable AI (XAI), operators will have full understanding (“explainability”) of the actions that their AI-based solutions recommend, to identify the root cause of events that negatively impact network performance and user experience.
In this way, those involved in optimization will have visibility into the factors that contribute most to the problem, the impact that each of them has on performance and the recommended actions.
These solutions are based on AI models trained on global datasets that are the largest and most diverse available on the market; models can also be retrained locally, to implement new use cases quickly and flexibly.
A further innovation is the introduction of an intuitive and adaptable user interface that allows operators to increase productivity.
Jean-Christophe Laneri, Vice President & Head of Cognitive Network Solutions di Ericsson said: “AI-based systems are rapidly becoming more sophisticated and complex, so facilitating operators to adopt these technologies is key to minimizing the time between introducing AI and starts producing value. Explanable AI provides methods and techniques for understanding AI models, making otherwise opaque decision-making processes transparent. The availability of these features in our Cognitive Software is a fundamental step in the network automation journey.”
Some customers who have already experimented with the use of AI in network optimization commented on this announcement, highlighting the concrete results obtained in network performance and user experience.
Jean-Paul Arzel, Executive Vice President of the CTIO of Bouygues Telecom commented: “At Bouygues Telecom we are constantly striving to improve the quality of service offered to our consumer and business customers and are carefully evaluating how best to leverage AI in our business. With Ericsson we have achieved very promising results from the application of AI to network optimization: we have reduced traffic congestion, improved capacity and also spectral efficiency, to make the best use of the available bandwidth. Ericsson has been a long-time partner of ours and we really appreciate their commitment to helping operators adopt AI faster and offering capabilities that help improve the network to meet our customers’ expectations”.
Perihane Elhamy Ahmed Metaweh, Chief Technology Officer of Robi Axiata he said: “We have over 56 million subscribers to our services in Bangladesh and therefore we continuously seek to improve their experience and increase our efficiency by leveraging automation. In our recent work with Ericsson we improved data volume and throughput in urban areas by more than 11% using AI-based network optimization solutions. We believe this evolution in the Cognitive Software offering will help us achieve even greater results”.
Added to these first impressions is also the evaluation of Adaora Okeleke, Principal Analyst di Analysys Mason: “Operators are starting to understand the value of AI but still face several obstacles, including the lack of transparency in their AI solutions, poor access to high-quality data, and the difficulty of adopting these technologies on a broader scale. Ericsson’s focus on addressing what concerns operators today by offering solutions that foster flexibility and trust is well aligned with current market needs and will help accelerate the adoption of AI.”
Cognitive Software, a cloud native architecture for networks
Ericsson’s Cognitive Software architecture is cloud-native. It provides automated container-based deployments, ensuring a secure environment in which software is promptly updated, thus increasing efficiency. The modular structure allows the adoption of the different components of the Cognitive Software to be scaled separately, depending on the needs of the moment. This also facilitates integration with CI/CD pipelines for continuous integration and deployment and encourages the adoption of DevOps methodologies.