Watch: What is the potential of applying generative AI to network automation?

Generative artificial intelligence (GenAI) is expected to become a compelling technology trend by 2023, bringing important applications to businesses and individuals, including education, according to a new report from market research firm Omdia. In the telecom sector, GenAI use cases are mainly focused on delivering personalized marketing content or supporting more sophisticated virtual assistants to enhance customer experience

Although the application of generative AI in network operations is not obvious, EnterpriseWeb conducted an interesting proof of concept that demonstrates the potential of generative AI in this field.

Capabilities and limitations of generative AI in network automation

One of the early applications of generative AI in network operations is the use of interactive guidance to replace engineering manuals to help install network elements, speeding up and simplifying installation tasks. In addition, generative AI can provide troubleshooting recommendations based on its engineering manual knowledge base and assist with network planning and design

Applying generative artificial intelligence to service orchestration across network domains is a more challenging task. In order to manage automated systems for telecommunications services, they must be scalable, secure and reliable. Therefore, these systems must operate according to deterministic logic rather than relying on guesswork in statistical paths. Intent-based orchestration systems require the translation of declarative commands into concrete workflows. In probabilistic analysis related to generative artificial intelligence, any random uncertainty cannot be tolerated

Generative AI has limited capabilities in the field of telecommunications network operations, but it can become a valuable assistant, providing advice to network engineers and supporting them in sharing knowledge. In addition, it can provide a revolutionary new user interface for orchestration systems, enabling users to talk to generative AI through natural language to express their queries and commands. This approach will abstract away the underlying complexity, thereby simplifying and accelerating the design, ordering and management of network services. Behind the scenes, the operator's automated systems will interpret user intent and translate it into the actions needed to complete the task

Generative AI is used by EnterpriseWeb to demonstrate intent-based orchestration solutions

In a recent Demo demonstration, New York-based software company EnterpriseWeb demonstrated how to leverage generative AI to support intent-based service orchestration. Their no-code automation platform enables businesses to streamline operations, optimize workflows and improve overall efficiency, integrating and managing complex systems, applications, data and processes. Since entering the telecom space in 2013, EnterpriseWeb has led the European Telecommunications Institute (ETSI) first proof-of-concept for Network Functions Virtualization (NFV) and launched their telecom product CloudNFV

CloudNFV is an intent-driven multi-domain orchestrator based on a no-code platform, designed to simplify tedious and complex tasks in telecom operations and accelerate service delivery. The solution uses a graphical model that encompasses industry standards in a unified format, providing the necessary abstractions for multi-domain and multi-cloud environments. The core of this model is declarative, intent-driven network service orchestration

By combining Microsoft’s natural language programming interface Jarvis and OpenAI’s underlying model, EnterpriseWeb’s generative AI demonstration demonstrates the ability to compose, configure, deploy and manage services in dialogue with Jarvis. Working with analytics software provider KX, the operator's intentions are translated into specific queries and commands, so that the operator only needs to ask the system to start a 5G gateway or reconfigure services, and the system can demonstrate the necessary operations. Once the operator approves these actions, the system will execute them. After the service setup is completed, KX will monitor the service and report events to EnterpriseWeb, realizing an automated cycle of service management

Re-express this sentence:

In EnterpriseWeb's demonstration, the capabilities, limitations and considerations of generative AI in telecommunications network operations were highlighted

In this particular case, the task of calling the orchestrator (EnterpriseWeb) to take action is delegated to the large language model (LLM). However, to achieve this goal, EnterpriseWeb uses the KX analytical database as an intermediary between the orchestrator and the large language model. Omdia analysts noted that this is an important consideration because of the need for a clear separation of concerns. The understanding and control of specific operations by large language models does not exist. Doing so protects the operator's network from the illusion of generative AI and ensures that the operator's IP (domain model and activity) does not flow into the large language model

Carrier-grade service orchestration is generally not compatible with the random responses of generative AI. Instead, a rule-based system is needed to perform tasks reliably. In this case, generative AI can serve as a revolutionary interface that sits at the top of the stack and abstracts away from the more complex lower layers, freeing network engineers from the need to memorize a vast array of device-specific commands or provide automation The system looks for suitable prompts to perform carefully orchestrated actions

C114 communication network is Ace

The above is the detailed content of Watch: What is the potential of applying generative AI to network automation?. For more information, please follow other related articles on the PHP Chinese website!