Bringing the Automated Intelligence Management framework into the CloudCO environment

Bringing the Automated Intelligence Management framework into the CloudCO environment

By Mauro Tilocca, TIM – Telecom Italia, and Broadband Forum Service Provider Action Council Chair

Over the past decade, there has been some revenue flattening, and an increase of OPEX due to network scale growth. Suddenly, there is a need for network monitoring and maintenance to take advantage of innovative technologies and solutions.

So, there are certainly advantages for network operators. In particular to the home and access networks, where it is fundamental to minimize the customer complaints and reduce on-site maintenance, as well as anticipate faults or degradation via analysis of the status of the network and traffic conditions.

Over the past few years, Broadband Forum has continued to develop specifications related to the Automated Intelligence Management (AIM) framework. These solutions rely on Artificial Intelligence (AI), Machine Learning (ML) and automation to improve operations and maintenance efficiency, as well as reduce overall OPEX.

Broadband Forum’s specification TR-436 defines an overall framework that enhances the capabilities of the Broadband Forum’s CloudCO architecture. TR-436 brings improved data collection and processing, and delivers AI-based inference of the network conditions by providing recommendations to fix a fault or recognize a degradation trend before the user perceives any service issue. Broadband Forum’s work also continues with WT-486, which specifies the interfaces for the AIM framework in TR-436.

This series of specifications, within the SDN/NFV Work Area, focuses on enabling network automation and low-maintenance operations for simplifying network validation and engineering, streamlining network deployment and upgrades, and improving operations with less error-prone and automated Oracle Access Manager (OAM) in the CloudCO environment. The work will help automate some management functions and realize rapid troubleshooting and action pre-emptive maintenance.

Improving the end-to-end customer experience with AIM

When an AIM orchestrator receives requests from the customer management layer, such as a CRM or an Order Management system, it implements closed‑loop logic, collects data from the network and performs data processing and analysis. This in turn, generates AI-based recommendations that allow the operator to fix poor network conditions.

The AIM framework was showcased during this year’s CloudCO Demo in the monitoring and troubleshooting of the home Wi-Fi network. Service KPIs are monitored against certain requested service levels. If they are not fulfilled, the AIM solution is able to recognize that, generate recommendations, and issue reconfiguration commands to re-establish the expected service levels.

These new capabilities help improve use cases such as zero-touch provisioning and end-to-end customer provisioning. But even more importantly, in the service assurance domain, they can provide single domain and end-to-end root cause analysis, path degradation predictions, and in the future, end-to-end customer experience assessments.

These AIM components are implemented as virtualized functions in a cloud-native environment and can be developed and deployed seamlessly in a timely and agile fashion. Overall, these solutions allow CSPs to reduce their OPEX while offering a high level of flexibility and integration within deployed networks thanks to their cloud-native nature.

This year’s CloudCO Demo

This service assurance with performance monitoring and automated traffic steering is attractive for Service Providers as it enables an automated and intelligent way to operate the network with minimal human intervention.

Showcased in October at Network X in Amsterdam, the demo includes work from the Broadband Forum’s Quality Experience Delivered (QED) project that probes the network to recognize network performance (packet loss, latency) between the subscriber premises and any location in a provider’s network. The probes cover the network itself and the Wi-Fi home network, to implement closed‑loop service assurance enabled by the AIM framework, along with QED, to guarantee the best service path and user experience in case of Wi-Fi or network congestion.

A home Wi-Fi network with a demanding application was showcased, with the quality of a 4K video being suboptimal when streamed. In the demo, network performance was measured by the probes embedded in the network and Residential Gateway and reported to the AIM pipeline, where the quality issue was diagnosed. The Wi-Fi application embedded in the CPE then performs traffic steering on the Wi-Fi network, resulting in improved quality for the video stream. The Service Provider has full visibility on the network conditions but no human interventions or actions are required because the AIM closed‑loop logic autonomously reconfigures the network.

The AIM pipeline also analyzed reports which included performance measurements from various probes located at different customer premises locations and network locations. Degraded network performances, represented in the demo by an overloaded BNG User Plane, are recognized by the AIM pipeline and trigger the User Plane Steering Function (UPSF) to make available a different BNG User Plane for all impacted subscribers. While the network delay was initially poor because of the congestion inside the network, it was brought to an acceptable level as a result of traffic steering by the dBNG.

Access the full recording of this year’s CloudCO Demo here.