Bridging the Gap Between Telecom Realities and IT Aspirations.
Contributed by Ned Taleb, CEO and Founder, Reailize & B-Yond.
In the dynamic landscape of the telecommunications industry, the buzz around “Network Automation and AI” is ubiquitous. Almost every vendor in the telecom market incorporates these terms when presenting their solutions or products. However, amidst the hype, the actual progress in the network side of telecom providers often falls short, especially when compared to the rapid advancements seen in the broader IT sector and the TechCo segment.
This disparity is not unexpected. A first-hand account from a European Tier 1 operator reveals the challenges faced when cloud engineers attempted to implement a 5G Standalone (SA) virtualized core. The stark realization hit them when their personal and family devices were provisioned to the new virtualized core. The stringent requirements of network availability and reliability, distinct from other cloud services, became glaringly apparent. Telecom networks operate under a different set of demanding standards, requiring attention to details that might be overlooked in other realms. Hence, hiring cloud and ML engineers is not going to quickly resolve skill gaps on the telecom operator side. Such new resources will require some time to learn and adjust to the telecom’s reality. An alternative is to partner with companies that have already gained the blended experience with Telco-applied Cloud and ML Engineering solutions.
In the intricate dance of nature, cross-pollination occurs as a bee gracefully moves from one flower to another, facilitating the exchange of vital elements. Remarkably, a similar phenomenon unfolds in the vast landscape of technology, both at an individual and company level. Those individuals and organizations that traverse diverse projects globally, offering cross-domain and cross-technology solutions, stand poised to propel the AI and Automation journey for Communication Service Providers (CSPs).
A compelling illustration emerges from our own endeavours, a testament to the power of diverse exposure and collaboration. Enter the realm of Packet Capture (PCAP) analysis—a technical and engineering-intensive domain demanding considerable expertise and domain knowledge for meticulous issue identification. However, innovation often thrives on breaking traditional moulds. Through a symbiotic partnership between our Data Scientists and Subject Matter Experts (SMEs), a groundbreaking approach emerged. PCAP files were transformed into a unique “language,” and a sophisticated Large Language Model (LLM) algorithm was deployed to automatically unearth words out of context, pinpointing with precision the faults within the system.
Allow me to illustrate it with another tangible example from our experiences. Our team undertook a significant engagement with a Tier 1 operator in North America, embarking on the establishment of a cloud infrastructure across various cloud vendors for diverse projects, including the deployment of 5G Standalone (SA). As we worked diligently on the ground delivering on the agreed scope, the meticulous documentation by our Subject Matter Experts (SMEs) and the close collaboration between our cloud and Machine Learning (ML) engineers yielded remarkable results over time. The outcome? A robust platform that genuinely transformed the deployment process from a matter of months to mere hours, and this is no exaggeration. This achievement not only underscores the power of a tailored approach but also highlights the impact of sustained collaboration and attention to detail.
In reflecting on our extensive journey, a compelling sense of responsibility drives us to deepen our contribution to AI and automation in the Telco sector. A pivotal step in this direction is our recent venture, the “LLM Discovery Workshops.” These workshops foster collaboration as we empower clients to shape their unique use cases using Generative AI and LLM technology. Leveraging open-source resources like LangChain and open LLMs, we ensure a secure integration with information systems, underpinned by robust authentication and access controls. A notable illustration from this initiative is the creation of a “Ticket Recommender System.” Here, the LLM model identifies parallels with past tickets by determination of a similarity score that is then used to suggest a resolution codes and a resolution team for a newly created ticket.
In essence, the path to effective network automation and AI integration lies in recognizing the unique challenges posed by the telecom sector. It involves a blend of experience, expertise, and a pragmatic approach that addresses immediate concerns while keeping an eye on the larger strategic goals.