HOW AI IS REVOLUTIONIZING TELECOMS
HOW AI IS REVOLUTIONIZING TELECOMS
What’s the first thing that comes to mind when you hear the term AI? Is it Stephen Spielberg’s 2001 film A.I. Artificial Intelligence or perhaps it’s machines taking over the world, marking the end of human civilization?
Just recently it seems everybody is talking about Artificial Intelligence (AI), especially in the realms of telecoms. AI is making a huge resurgence as the technology advances to a point where machines really can take on some of the tasks only humans could do until a few years ago. In fact many of the advanced tasks that AI is currently being developed for are well beyond the scope of human capability due to their speed and complexity – thus demonstrating the true value of AI.
For telecoms service providers it is just the beginning. AI is opening new doors and finding applications in customer experience, self-optimizing networks, and network virtualization, to name just a few. AI lends itself to highly complex, repetitive tasks where machine learning can enable computers to make rapid decisions based on algorithms, heuristics and learnt experience, and no longer just on pre-programmed procedures.
So what is AI and why now?
According to the Oxford English dictionary “Artificial Intelligence is the theory and development of computer systems able to perform tasks normally requiring human intelligence”. This can include tasks such as visual perception and speech recognition, but in telecoms it can also include finding the best configuration for a virtual network or optimizing the use of cloud resources. Until now, lack of computational power and data storage have limited AI initiatives but new cloud computing and storage capabilities have opened the door for AI, enabling faster calculations and access to huge amounts of data.
AI and machine learning
The terms AI and machine learning are often used interchangeably but machine learning (ML) is actually a sub-category of AI. Machine learning provides AI systems with the ability to automatically learn and improve from their own experience without having to be explicitly programmed. This behavior is very similar to the way humans learn. When AI is applied to very large amounts of input data such as in ‘big data’ applications, this is called deep learning where information is processed in layers. The result of one layer becoming the input for the next. Deep learning provides capabilities such as automatic feature detection and allows machines to recognize specific patterns and behaviors which may then be classified and used in problem solving.
AI in telecoms
This ability that AI has to identify certain behavioral patterns can help in automatically resolving complex problems associated with particular network events. For example, being able to distinguish between normal behavior and unusual/exceptional behavior in the network can be used to identify particular fault conditions which may otherwise be very difficult to find.
According to a recent report from TM Forum, a significant number of service providers are already conducting AI proofs of concept, working with AI suppliers or have developed some internal expertise in AI. Telecoms applications include customer experience, network automation and service management.
Customer experience is probably the most advanced application where chatbots are being developed and trialed with service providers to aid or even replace customer care agents. Chatbots can be very effective in resolving customer problems and facilitating cross-selling of compatible products. They also allow for very slow customer responses and don’t suffer from frustration or fatigue, even with difficult customers.
In network automation, AI can be applied to network provisioning and optimization of virtual network resources. With the internet of things (IoT), automation will be key due to the huge amounts of continually changing traffic demands and network configurations. Intent-based management and data analytics will be used to orchestrate end to end services across these complex, multiple technology domains in near real-time.
Regarding service management, AI can be applied to resource optimization and regression testing, fault and root cause analysis, and traffic congestion avoidance. Networks will become self-configuring and self-healing, removing the burden from human beings of having to manage increasingly complex and rapidly changing networks manually.
Pushing the boundaries in AI
When applied to geospatial data obtained from geographic information systems (GIS), AI can be used to predict and classify specific types of user behavior according to location and demographic, leading to a better understanding of service and product usage. Companies like ESRI are specialists in geospatial information mapping and data analytics. Their products enable event prediction and just-in-time supply chain management which can be applied not only to commercial product promotion and delivery but also in supporting major disaster relief.
In Europe, the European Telecommunications Standards Institute (ETSI) have formed a special cognitive architecture group who are specializing in the use of AI to improve telecoms networks. Their work is based on an ‘observe-orient-decide-act’ model which uses AI techniques and context aware policies to adjust offered services based on user needs.
AI and the future
Despite alarmist reports of AI running out of control, outwitting humans and becoming autonomous ‘super-intelligent’ sentient beings, it is more likely AI will be confined to very specific applications under tight control, at least for the foreseeable future. Whatever we think, AI is here to stay and enabling communications service providers to improve their networks and offer better, more reliable services – something we can all benefit from!