Intelligent applications with network-focused machine learning (ML) models and inherent telecommunications expertise can significantly improve mobile network operators’ (MNOs) network performance. As 5G networks continue to evolve, these applications will help complex 5G+ networks with physical and multi-cloud components function at full capacity and capability, and provide expanded functionalities enabled by artificial intelligence (AI) technologies like neural networks, Large Language Models (LLM) and generative AI. The benefits include drastically simplified operational tasks that deliver a more capable, resilient, and trusted network that can be managed with a light operations touch.
Consequently, AI-powered network applications have a growing role in MNOs’ multivendor network stack. A practical example of intelligent applications that can help operations teams realize immediate value are tools that find and solve complicated network problems in minutes, rather than days or weeks.
Redefine network operations with intelligent applications
Operators can realize immediate value with AI/ML by giving intelligent applications as much historical network data as possible, including hardware IDs, severity and timestamps of alarms, and resolution tickets. These types of tools use this information to build a simulated network for predictive modeling, adaptive control, and to train other applications.
Once tools ingest the information from hundreds of thousands of nodes and end points, they use neural networks to recognize new combinations of known behaviors and identify new behaviors. This intelligent environment has the power to help operations teams deliver accurate diagnosis, rapid responses, and efficient resolutions – all while minimizing costs.
At Mobile World Congress 2024, Fujitsu demonstrated how operators can ignite the power of AI in their network operations with intelligent applications. Virtuora intelligent applications are multi-vendor, multi-domain network management tools that use AI/ML to investigate and interrogate hundreds of thousands of network nodes, swiftly cutting through noise to pinpoint adverse performance behaviors and their causes. With intelligent applications, operations staff won’t waste precious time chasing dead-ends and dispositioning overlapping data points.
Fujitsu also demonstrated how intelligent applications and Fujitsu generative AI go beyond sorting and understanding the converged network’s multi-variant big data landscape to deliver actionable insights with specific countermeasures tailored to individual network scenarios and conditions.
Fujitsu generative AI
Intelligent applications combined with Fujitsu generative AI create customer solutions that investigate specific network conditions, and then use natural language interactions and LLM technology to generate a response to the user’s query. Industry examples that are not network specific include ChatGPT, Google Bard, and Azure Cognitive Services. And with Fujitsu automated data wrangling, LLMs learn from a variety of network data, including resolution tickets, log files, past failure knowledge, and packet capture data — regardless of type and vendor.
For operators to use LLM to perform network investigations in multilayer, multivendor mobile networks effectively, network knowledge is required. However, that information can be difficult to obtain and verify because different vendors use different data models and formats. Fujitsu has developed automated data wrangling to automate the previously tedious and time-consuming task of preparing data for AI analysis. This process involves cleaning, structuring, and transforming raw data into an LLM-friendly format that intelligent applications can parse correctly and efficiently. The resulting generative AI delivers the result of network investigations, as well as recommendations for specific issues and areas of optimization.
As recommendations are implemented with intelligent applications, that information is fed back into the network model, continuously improving the accuracy of investigations and remediation, as well as informing adjacent network automation. This advanced function makes it possible for operations staff to constantly focus on enhancing network productivity and service delivery, getting the most out of what they have right now without costly upgrades and unnecessary capital expenses.
Virtuora intelligent applications and Fujitsu generative AI help operations teams navigate complex vendor ecosystems using AI/ML tools without being data scientists or knowing how to train data models. These applications embrace multivendor, multilayer network data, seamlessly integrating with existing infrastructure and eliminating expensive and disruptive infrastructure re-architectures to accommodate AI. This capability allows Operations teams to leverage the power of LLMs without compromising the current environment or customer experience.
The Virtuora cloud provides service orchestration and delivery for the end-to-end network. Now, Virtuora intelligent applications and Fujitsu generative AI help operators realize value with what they have already installed, becoming a powerful operations partner with network specific knowledge that empowers operators with trusted network solutions.
Unlock the transformative power of AI with Fujitsu.
Special thanks to the following individuals for their technical expertise and collaboration on this blog: Joe Krystofik, Kazumi Doi, Mutsumi Saito, Yuta Teranishi and Shinji Uranaka