UBBF 2018: Huawei Starts Ignition on Autonomous Driving Networks
Enabling fully autonomous cars is a huge technical challenge but putting in place autonomous driving networks -- Telco networks that run themselves without the need for human intervention -- cranks up complexity to an entirely new level. The number of network elements involved spanning different domains -- mobile, home broadband, IP transport, enterprise services and data center networks – is enormous.
David Wang, Huawei's Executive Director of the Board and President of Products & Solutions, appeared unfazed by the task. Speaking at the fifth Ultra Broadband Forum (UBBF 2018), he outlined concrete steps towards realising the dream of a fully automated network. They comprised new architectures capable of predictive O&M, backed up by telemetry data and artificial intelligence. By not embarking on such a journey, warned Wang, existing structural problems will only get worse.
“Opex has been increasing faster than revenue for the past ten years,” he said. “We’ve spent too much effort and money on O&M and service provisioning.”
Wang highlighted so-called OTT players – Telco competitors in some cases – as much more operationally efficient. “[Telco] customer experience management is mainly driven by customer complaints,” observed Wang. Reactive O&M, he added, was undermining industry attempts at curtailing churn.
If nothing is done, customer experience will likely deteriorate. Wang quoted one forecast of 75 billion devices connected to networks by 2025, which will stretch human capability and skill to manage well. Wang was nonetheless optimistic that system architecture innovation, in the shape of autonomous driving networks, will come to the rescue.
One step at a time
As a starting point, Huawei has proposed industry definitions on what the various steps might look like towards enabling autonomous driving networks. Level zero is where all dynamic tasks need to be executed manually, and Level Five is the fully automated network.
The intermediary levels, ranging from one to four, describe an ever-increasing reliance on automation and less reliance on the need for highly-skilled personnel to monitor and intervene. “These definitions could be discussed and developed in coming years, but we need to start somewhere,” said Wang.
The Huawei man further advised that initial efforts in automation should be steered towards reducing opex – a painful burden for operators – primarily by increasing O&M efficiency and making better use of network resources. According to Wang, 50% of current opex challenges can be addressed through autonomous driving networks. In keeping with Huawei’s step-by-step approach towards automation, Wang talked of starting work on single domains before moving to multiple domains, and similarly going from single tasks to multitasking. Closed-loop systems can then be formed, said Wang, which essentially means the network looks after itself.
Not just a dream
“Autonomous driving networks are a good dream, but how do we make the dream happen?” asked Wang. “We need to look at scenarios.”
One scenario that holds out much promise for Wang is home broadband service. The lifecycle of the network is clear: design, rollout, provision, assurance and optimisation. “It looks like a closed-loop job,” quipped Wang, “but how can we take it further? Can we go to zero-touch configuration and deployment?”
Wang said Huawei could help operators take home broadband service up to Level Three, which enables ‘intent-based’ closed-loop management. By using the supplier’s intent engine, which forms part of Huawei’s Network Cloud Engine (NCE), operations staff can easily programme the network to run a series of tasks without human intervention, including deployment and predictive O&M. Huawei says the intent engine by working in a closed loop of its own with other NCE engines– automation, intelligence, analysis – makes possible smarter (and more efficient) troubleshooting and error-handling.
Intent-driven networks (IDNs) was a key Huawei theme at UBBF 2018, and Wang outlined different possible use cases for them – each a clear staging post on the road to autonomous driving networks. One example was user-centric predictive maintenance for premium broadband. By using a visualized Optical Distribution Network (ODN) topology, fast troubleshooting and fault demarcation – coupled with high accuracy in root-cause analysis – is possible without any special technical knowledge on the part of the ops team. “Big data and AI can enable detection of poor quality problems within minutes, even though the connection remains,” said Wang. Huawei claims its IDN premium broadband solution can reduce churn by 20%.
Another use case presented by Wang was Huawei’s Optical Service Health Predictive analysis, which includes a real-time fibre-connectivity map. Huawei’s NCE can run ‘sub-health’ predictive analysis on each connection and triage the most important connections to handle first. Wang flagged increased automation in data centers to reduce manual configuration errors as another IDN use case. He anticipated that unmanned date center networks will be the first part of the Telco network to reach Level Five.
Wang conceded that the journey towards autonomous networks will be a long one, and that industry will have to work hard to achieve it. The alternative of not embracing automation technologies, however, is to fall behind other industry sectors – the likes of automotive, aerospace, manufacturing and direct OTT competitors. “Autonomous driving networks are an imperative,” said Wang.
This blog is sponsored by Huawei.
— Ken Wieland, contributing editor, for Huawei.