JOHN WILLIAMSON NETWORK AUTOMATION
IN THE WORKS What might future optical network automation solutions consist of? There’s some consensus that digital twins will be part of the deal. Hollander speaks of the increased adoption of digital twins to simulate and optimise network performance and roll-out strategies. “The concept of digital twins is gaining traction in the context of optical network automation,” agrees Fiala. “It involves creating an environment testing and validating AI model outputs, confirming their accuracy and effectiveness.” “Widespread adoption and growth of AI will be constrained by the electrical grid’s ability to power new datacentres,” calculates Pesic. “As a result, operators will need to prioritise energy efficiency and optimise network operations to minimise the impacts of insufficient electrical infrastructure. This may involve automating the powering down of unused network elements and powering them up as needed.” MEASURING UP Some of the individual benefits of optical network automation have been quantified. As remarked by Pesic, a 2024 investigation conducted by Nokia and Analysys Mason, concluded that: • Automating network lifecycle management processes led to an overall savings of up to 56% in operational costs through simplifying complex network operations tasks and shortening the time to provision, configure, deploy and manage optical networks • Operational cost savings of up to 81% for network operators offering service virtualisation and slicing, because of a reduction in the time that it takes to complete service order orchestration (90%), service fulfilment (83%) and service assurance (54%) processes • Automating network planning for both the planned and deployed network optimises network resources and enables legacy network equipment to be retired, which contributes to CapEx avoidance of 30%
for permitting processes, and optimise business planning with ROI evaluations. “Technologies like digital twins, AI agents for investment decisions, and Gen-AI platforms for telco-specific tasks further refine automation,” he argues.” Gen-AI can also be used to automate network testing, helping to pinpoint the most critical areas to test, leading to higher quality and faster time to market.” According to Pesic, although the integration of AI/ML is still in its nascent stages, relatively speaking, Nokia believes AI/ML will simplify network lifecycle management by reducing the time required to detect, respond to, and resolve issues. She reckons that several areas within the optical transport domain, which are currently under development, demonstrate considerable promise in leveraging AI/ML algorithms to: predictive traffic management; fault detection and diagnosis; predictive maintenance; automated troubleshooting; and natural language Interaction. AI SPEED BUMPS? But for the further use of AI/ML in optical network automation, a few issues may need to be addressed. Fiala notes that network data is key to the accuracy of AI-generated responses. “To develop sophisticated and precise AI models for specific networking use cases, vendors and service providers have to collaborate and share extensive network datasets,” she reasons. Ciena adds that AI-driven capabilities heavily rely on AI models, which currently reside in the cloud due to computational and memory requirements. “Service providers need secure access to the public cloud or, alternatively, can utilise their own private cloud infrastructure,” Fiala contends. Finally, she mentions that AI models must be “smaller” so that they can be deployed in private clouds or on- premises. “The technology is evolving at hyper speed to address this challenge, striving to optimise the resource requirements for AI consumption as well as AI model training.”
Do cognitive and intent-based networking have key parts to play here? “There is certainly a role for this technology in the management of fibre customers,” says Tongish. “This is not an area that IQGeo get directly involved in, but mitigating customer churn is vitally important to the success of all broadband businesses.” Related to the greater utilisation of AI/ML, Tongish emphasises that improved network data quality is a must. “Automation will fuel better network data quality,” he predicts. “Network data quality is going to become more important as we try and add more sophisticated network technologies.” Pesic ends with a somewhat mixed message, pointing out that some network operators may lack the in-house expertise to fully leverage the benefits of increased automation in network operations. “This will create a growing demand for vendor-provided professional services that can support multi-vendor, multi-layer networks.” • Operators expect to benefit from a 10% uplift in revenue from faster service turn up, combined with an accelerated time-to-market for services that led to higher win rates and the ability to offer differentiated services made possible by optical network slicing and network-as-a-service. Adding to these sorts of positive stats, Hollander reports that Amdocs has found that automation can lead to as much as 50% improvement in time to market. It also leads to increased quality (over 95% initial quality) and consistency across the network, leading to lower operations costs. There may be caveats, though, and not all types of optical network operators may exhibit the same enthusiasm and requirement to embrace automation. Tongish suggests that, for smaller players, it could be less of an issue since they don’t face the same kind of economic pressures as their larger counterparts that are moving from a race to build fibre to a race to retain customers.
Marie Fiala Director, Portfolio Marketing at optical & routing systems, Ciena
Steve Tongish CMO, IQGeo
Nir Hollander General Manager, Amdocs Mobile Networks
Jelena Pesic Director, Optical Strategy, CTO Office, Nokia
www.opticalconnectionsnews.com
9
ISSUE 40 | Q1 2025
Made with FlippingBook Online newsletter maker