By FOUR DOT INFINITY (project partner)

The transition towards the 6G Radio Access Network (RAN) introduces significant challenges alongside great opportunities for mobile operators and developers of intelligent communication services.

To meet the requirement of offering higher speeds and reliable connectivity while simultaneously reducing energy consumption and operating costs, providers must integrate RAN control applications. In this context, FDI, as a pioneering SME, specializing in xApp development in the context of SNS 6G-LEADER, provides tailor-made solutions, bringing intelligent decision-making directly to the RAN.

At FDI, we are currently working on xApps that utilize Deep Reinforcement Learning (DRL) along with other optimization techniques for more efficient allocation of network resources, such as dynamic allocation of Physical Resource Blocks (PRBs) and power control. This allows the network to adjust the distribution of available resources in near real-time to achieve throughput maximization for its users, without excessive spectral and energy resource consumption. In greater detail, the xApps we are developing consider practical constraints, such as the service requirement of users and devices connected to the RAN, and interference levels in the network, ensuring optimal resource allocation service demands and allowing the RAN to achieve spectral and energy efficiency in multi-cell scenarios.

To this end, we extract datasets from practical scenarios using open-source tools to train our AI/ML models that drive the throughput maximization xApps. Meanwhile, we strive to guarantee that AI/ML integration in the RAN will not lead to excessive computing and power costs, thus incorporating techniques, such as pruning to further cut the complexity of our DRL models.

The value for providers is clear: with intelligent resource management, the RAN can operate more efficiently, without wasting spectral and energy resources, going above current static or rule-based approaches. This not only increases network efficiency, but can also help reduce the energy footprint, supporting the EU Green Deal and SDGs for more sustainable and “green” wireless networks.

The work we are currently doing in FDI with the xApps is the first critical step towards the next generation of wireless networks, where AI/ML will significantly contribute to improving user experience and network reliability, without ignoring their environmental impact. Our ultimate vision consists of a RAN that will utilize xApps to optimize its operations in near real- time, being highly adaptive to dynamic variations of traffic, mobility and interference patterns that may arise.