Artificial Intelligence is a hot topic right now, with the potential to help improve efficiency both across the network, and from a cost perspective. Kailem Anderson vice president for software and services at Blue Planet, a division of Ciena, speaks to Fibre Systems about the role of AI in operator networks
There has been a lot of buzz about AI recently, but there is still some confusion around how it should be defined. What is your perspective?
AI is a bit of a nebulous term, and despite having the same underpinning concepts across market verticals, AI in the workplace tends to be a very diverse concept among our customers. At the core of AI, we have intelligent automation, a thinking machine which harnesses the power of data to make decisions and introduce greater efficiency to the workflow.
The current vision of AI for operator networks is largely based in security functions such as Distributed Denial of Service (DDoS) prevention, migration/real time automation and path selection in the network. In the future, we expect to see AI expand into new areas such as defining network topology and eventually automating decision making – using policy and human oversight to ensure consistency and reliability.
Where and how would AI best be implemented?
As an immature technology, AI should be implemented in stages, with rigorous testing to ensure it has both enough data and the right data to form solid decision-making policies. Operators should develop projects individually, and once satisfied with the results in one area, scale deployments to cover more functions.
One thing that is certain – at least for the near future – is that operators will not surrender control of their networks to automation entirely. Rather than automated networks, they are looking to develop adaptive networking practices that harness the power and efficiency of data-driven AI, and combine it with the invaluable experience of their engineers.
How does AI play into data and analytics in the telecoms industry?
There is an opportunity in the telecoms industry to nurture the growth of powerful AI technologies, as the data generated by today’s networks is vast and extremely valuable, in terms of training machines to think for themselves.
By taking the data generated in the daily operation of communications networks, it will be possible to identify patterns and form effective policies to guide the machine’s decision-making skills, as and when new situations arise.
Analytics are also critical to AI and it is important to know how the two are related, to understand the full proposition. Analytics concerns the mining of unstructured data to aggregate information to identify patterns and implement new measures to drive efficiency. Traditionally, the data could be collected by machines, but the analysis and the implementation of new policies would need to be handled by humans. AI has therefore entered the space as an enabler, which can evaluate data without human intervention and then determine the correct action before implementing within the workflow.
For network operators, this offers considerable time and resource savings, as data collection and analysis can be automated, and with intelligent decision making, engineers can be freed up from routine maintenance, to deal with more challenging core issues affecting the network. In addition, AI offers increased security through proactive network monitoring, using historical data to spot anomalies on network services and signs of intruder connections, and thereby identify a threat and conduct self-healing functions to protect the network and preserve functionality.
By leveraging data insights and applying analytics through AI platforms, network providers can more easily evolve their networks to be faster, smarter and governed by data-driven business policies that ensure profitability through providing a superior customer experience.
Will AI change the way networks function, and, if so, how?
AI has the potential to realise significant change in the telecoms industry – enabling intelligent, programmable and adaptive networks that can better meet the demands of the customer and the increasingly dataheavy services to which they subscribe. Take network bandwidth management as a specific example. Today’s dynamic environment contains millions of devices that continue to multiply. Operators are increasingly facing the challenge of ensuring that each of these devices is connected at all times and receiving the services being paid for. AI can address this directly by providing real-time deep network intelligence insights that help operators to properly allocate bandwidth depending on demand; thereby ensuring that the path from data centre to user is established and maintained.
Another area in which AI will thrive in telecoms is in conducting management and maintenance operations without human input. Self-healing networks are envisioned to be the next step in intelligent networking, enabling the network to completely repair (and potentially even reconstruct) itself in a matter of minutes, should a failure occur. Using real-time data analysis, AI will compress decision-making timelines by orders of magnitude, minimising, or even eliminating, disruptions from damaged cables or attempted network intrusions to save service providers and the operators significant revenue losses.
More generally, it will bring greater oversight, and therefore control, to operators, while allowing for usage optimisation without network disruption, and prediction of scaling requirements for hardware and virtualised assets, depending upon demand patterns. In this way, operators can ensure that the user receives a premium experience.
Where can we see examples of AI at work in telecom networks today?
We’re already seeing AI tools incorporated in networks in more progressive markets across the globe, such as Australia (Telstra), South Korea (SKT), Singapore (Singtel), and the US (AT&T).
Looking at AT&T specifically, the company is incorporating AI and machine learning into customer interactions, as well as software defined networking (SDN) functions – where signals from different nodes in the network can notify the operator of an impending failure ahead of time. Engineers can then be dispatched to complete maintenance work proactively, thereby safeguarding services and ensuring that the customer experience is not disrupted.
For other customers, such as Comcast, there’s a significant drive on implementing AI chatbots to more effectively deal with customer queries and improve customer service ratings. We’re also seeing some managed services organisations implementing AI to manage end customer networks. Traditionally these offerings would require tailored policy provisioning, taking account of the specifics of the network, operational norms and extremes and action preferences, which would all fall to the operator to design and implement. With AI, managed services can be automated, with the machine learning the functions of the network and the various failure situations ahead of time, and then deciding upon appropriate responses – all without human intervention.
How do you see AI evolving, and what does this mean in terms of how the company operates?
Though it is still maturing and most ongoing projects are very much in the development stage – managing smaller, less critical aspects and network functions – we expect to see AI branch out into mission critical management and proactive network maintenance in the coming years. As experimental projects go live and combine with other emerging technologies, AI will evolve to uncover new possibilities and use cases, which will continue to achieve greater efficiency and performance in telecommunications networks.
We are helping customers to incorporate AI for proactive network management, to achieve an enhanced experience for the user, with minimal disruptions due to network failures and improved customer support services.
The Blue Planet network health predictor solution uses advanced analytics and machine learning capabilities to allow network operators to identify potential areas of risk in their network, so that they can proactively take action and maintain service delivery.
With help from supplier partners, operators can implement AI in their networks to achieve significant efficiency increases and cost savings, while reacting faster and with better decision making to network operations developments, and offering a greater experience to the user, thereby achieving a competitive edge in their markets.