How AI empowers intent-driven service orchestration and assurance

Transform service orchestration with AI and automation

Intent-driven service orchestration and assurance represents the next frontier in service management and orchestration. Services are not only provisioned and activated but continuously monitored and adjusted to meet dynamic business, service and operational objectives. Artificial Intelligence (AI) can play a transformative role in empowering service management and orchestration processes by automating decision-making, predictive analytics, and closed-loop operations to ensure service instances are aligned with the defined intent. 

Figure 1: The role of AI in service management and orchestration

One of the significant challenges communication service providers (CSPs) face in implementing service automation is the complexity involved in end-to-end service blueprint design, decomposition, and assurance of end-to-end customer and resource facing service instances. Our experience indicates that to fully unlock the benefits of service management and orchestration, the system must be configured with a range of software artifacts, including service blueprints, decomposition and integration logic, master and atomic workflows, among others. Delivering an orchestration project requires substantial effort in designing and testing these artifacts.

CSPs are looking for ways to implement service orchestration and assurance operations more effectively. The shift towards managing automation processes rather than OSS system’s configuration and integrations is what can lead to more efficient and streamlined service operations. Evolving service management toward the support of AI-enabled intent-driven and autonomous service operations is the foundation for autonomous networks vision defined by Tele Management Forum (TM Forum) and described in recently published autonomous networks technical architecture.

What makes AI-empowered intent-driven and autonomous service operations possible?

At the core of TM Forum's vision for programmable networks is intent-driven service orchestration allowing service providers to define high-level business and service intents, which are then automatically translated into specific network configurations and actions. Service assurance is equally critical and ensures that the network service instance continues to meet the specified intents throughout its operation. 

To achieve fully automated service lifecycle, from service design and provisioning to service observability and assurance, the implementation of AI-empowered intent-driven and autonomous service operations will be required, and it can be accomplished by leveraging the capabilities of AI-enabled service orchestration systems.

Figure 2: AI enables intent-driven operations | Source: Ericsson

What's exciting here is the potential of AI in enabling autonomous service operations through intent-driven service orchestration for service instance lifecycle management, from service blueprint design to end-2-end service instance operations and assurance.

Enhance lead time for service blueprint design with AI to deliver innovative services faster and more effectively

For intent-driven service orchestration and assurance, AI and ML technologies can significantly simplify the work for service orchestration architects and network designers. For example, AI can enable the generation and validation of initial service orchestration blueprints using custom-trained Large Language Models (LLMs). This process leverages historical data, previous blueprint iterations, network configurations, and specific service requirements. It utilizes ML techniques, including pattern recognition and predictive modeling, to create innovative and functional service blueprints.

Once a service blueprint is generated, AI tools can simulate its deployment using a digital twin - a virtual replica of the real-world service implementation. This simulation can help to predict the performance of the service instance and identify potential issues before actual deployment to the live network. Additionally, AI systems can ensure that the blueprints adhere to existing network standards, operational protocols, and best practices and detect deviations from established models by leveraging unsupervised learning techniques.

Figure 3: AI-enabled service blueprint design | Source: Ericsson

Automating service blueprint design processes can significantly reduce the time needed to design, validate, and onboard new service blueprints, leading to quicker service time-to-market and improved operational efficiency. Ericsson is already harnessing AI capabilities to design and verify service blueprints and the use of AI is not only significantly enhancing service blueprints design and testing times but also facilitates the introduction of new advanced service orchestration and assurance use cases.

Evolve from intent-based end-to-end service orchestration to achieve fully autonomous networks

Intent-based end-to-end service orchestration is the foundation for transitioning towards autonomous network management, where the focus is on what the service needs to achieve rather than how it is to be implemented. AI plays a crucial role in realizing this paradigm as AI algorithms can interpret the defined intents and automatically manage the underlying network resources to meet these goals, enable the network to adapt to new demands in real-time, learn from data to optimize performance continuously, and self-correct in response to network anomalies or failures. Let's delve deeper into the potential of AI to revolutionize service orchestration and provisioning.

The first important aspect is how AI can enhance service orchestration by interpreting intents—high-level business or service goals—and translating them into executable orchestration actions. Communication service providers may incorporate advanced Natural Language Processing (NLP) techniques into service orchestration stacks for the system to interpret the service requirements expressed in natural language, extract meaningful intent information from user commands or service requests, and translate them into actionable standard-based API calls. Through semantic analysis, AI models analyze the context and semantics of these requirements to ensure the intents are accurately interpreted and fulfilled, considering factors such as the current state of the service instance and network.

Figure 4: Facilitate zero touch service orchestration with NLP translation | Source: Ericsson

Manage cognitive intent lifecycle with APIs

Another important aspect is cognitive intent lifecycle management where intents are defined, managed, and monitored through APIs. AI’s role here involves interpreting triggers and contextual information to generate proposals and guide closed loop decisions. AI/ML can contribute to the dynamic service design and deployment by analyzing alternatives on-the-fly and suggesting optimal target service graphs and configurations. 

Figure 5: Cognitive close loop for intent-based autonomous networks | Source: Ericsson

Cognitive intent lifecycle management automates many decisions that would otherwise require human intervention, freeing up resources for other tasks and ensuring more efficient service operations. More importantly, this approach fosters continuous improvement as AI and ML systems monitor outcomes of past decisions and learn from feedback to refine intent interpretation and enhance future service design decisions.

Improve service reliability with AI-enabled service assurance 

Once an end-to-end service instance is properly configured and activated, it requires real-time monitoring to ensure that its state and performance align with the predefined service intent, designed as a set of Service Level Key Performance Indicators (KPIs). To achieve this, service assurance systems ingest network telemetry data and aggregate it into service Level KPIs which are then monitored against configured thresholds to enable both proactive and reactive responses to any deviations from the service intent.

Identify and address issues before they occur!

As an example of proactive detection of service intent violation, service assurance can leverage AI/ML for the analysis of the traffic history and, by applying machine learning models and traffic patterns to historical data and real-time telemetry, AI can predict potential service degradations, “silent” failures, or capacity issues before they occur. This approach enables network operators to take preventive actions to avoid outages or performance issues, or in other words, predictive analytics can help to maintain service performance by identifying risks early. AI's ability to forecast issues or potential anomalies in the future is a key driver for the proactive optimization of network services and resources. 

Figure 6: AI-enabled service assurance | Source: Ericsson

In addition, AI's ability to interpret data on the fly can streamline decisions and remediations when an anomaly has been detected. Instead of relying on static pre-defined policies, AI can not only pinpoint the anomaly but also provide remediation or prescriptive recommendations and/or decisions that can be executed by the service orchestration to close the loop. AI recommendations can range from simple service and/or resource topology changes such as scale up of core network functions, to more complex changes that may affect the service instance topology design.

AI is a pivotal step towards the realization of fully programmable networks

The integration of AI techniques in intent-driven service orchestration and assurance is pivotal for advancing towards programmable networks. AI has the potential to simplify the automation of service blueprint design processes by handling the generation, validation, and deployment of blueprints into the network. It employs ML techniques like pattern recognition and predictive modeling to anticipate network and service demands and optimize service topology graphs and configurations accordingly. ML algorithms further contribute by learning from ongoing network performance data, thereby continuously adapting and optimizing service and network operations which is crucial for programmable networks that must autonomously adapt to changing conditions. 

Moreover, AI-driven solutions enable real-time service monitoring and assurance, maintaining service operations within set parameters of predefined intents and KPIs. This constant oversight and automatic adjustment are integral to maintaining the seamless operation of programmable networks.

Through these capabilities, AI and ML not only reduce the complexity involved in service management but also pave the way for networks that are more adaptive, efficient, and capable of self-management.

Read more about Ericsson’s approach to intent-driven autonomous networks realization:

• Article: Evolving service management toward intent-driven autonomous networks
• Report: Closing the loop: CSPs aim to automate service orchestration and assurance
• Blog post: Model-driven configuration management for multi-domain service orchestration

Further reading

• Solution brief: Ericsson network orchestration and assurance services
• Solution brief: Ericsson Service Orchestration and Assurance
• Solution brief: Accelerating 5G business growth with Ericsson Dynamic Network Slicing