One of the most anticipated themes of Mobile World Congress (MWC) 2018 was artificial intelligence (AI). Industry analysts and media speculated that AI would be the talk of the show, and that AI would be touted as the key to helping communications service providers (CSPs) across the globe to transform themselves into digital service providers. The stage […] 
One of the most anticipated themes of Mobile World Congress (MWC) 2018 was artificial intelligence (AI). Industry analysts and media speculated that AI would be the talk of the show, and that AI would be touted as the key to helping communications service providers (CSPs) across the globe to transform themselves into digital service providers. The stage was set for major announcements around AI initiatives from vendors and service providers alike.
But the truth is, AI fizzled at MWC. There was very little AI news from the show. Tractica monitored 39 potential AI telecom ecosystem players, both large and small, for AI-related announcements at MWC, and of those 39, only four – Sandvine, Ericsson, Huawei, and Amdocs – announced any significant AI initiatives.
Market Pressures on Communication Service Providers
Market drivers for incorporating AI technology swirl around gaining efficiencies, such as handing redundant tasks to machines or unleashing machines to analyze mountains of data. The telecom industry would seem to be fertile ground for gaining these efficiencies, and it is true that CSPs have embraced AI in customer service (chatbots, real-time, and predictive intelligence for live agent assistance) and as a cybersecurity tool, but that is about the extent of it. Given the market pressures that CSPs are under to transform their business, you would think there would be more urgency to find good solutions.
CSPs are in danger of becoming roadkill on the highway that is digital services, as hyper-scale players like Amazon, Google, and Microsoft move into more market domains. These players build and run open-architecture networks with little to no concern about interoperability or access. They can quickly and easily launch new services and close others down. For CSPs, the complexity of digital service offerings require automation. “In chasing the next big thing, application developers, service providers, and other organizations are rolling out new services requiring a back-end that is becoming more and more complex by the day. For instance, micro services that segment parts of a service into smaller clusters in the cloud have proven to increase development velocity and agility, but managing that mess becomes a whole new ballgame,” wrote Sumeet Singh, vice president of engineering for Juniper Networks in a Light Reading article in October.
Automation Networks Are the Ultimate Goal
Adding to the problem for CSPs is margin pressure. CSPs around the world have been dropping their prices, which is threatening revenue. According to CIMI, 2017 was to be the first year that revenue per bit would fall below the cost per bit for most operators. Capital expenditures (CAPEX) and operating expenditures (OPEX) must be addressed. Speaking at an industry gathering in October 2017, Deutsche Telekom deputy chief technology officer Arash Ashouriha stated: “Brutal automation is the only way to succeed.”
Ashouriha was referring to the automation of a broad range of network management-related functionality, including core and radio access network (RAN) traffic optimization, network maintenance, and other disparate back office functionality like billing and accounting, service assurance, and delivery. The ultimate goal of telecom services automation is the autonomous network. In this vision, telecom networks, in the words of Juniper Networks chief technology officer (CTO) Kireeti Kompella, will become autonomous networks in a similar fashion to how cars will be autonomous. Networks will “self-discover constituent parts, self-configure, self-monitor using probes and other techniques, auto-detect and auto enable new customers, automatically monitor and update service delivery, self-analyze, self-report,” said Kompella.
AI’s Slow-Moving Impact on the Telecom Industry
It is these network and back office operations where AI could have the most impact, yet it is in these telecom areas where AI initiatives are moving the slowest. There seem to be a few reasons for this, as described below:
- Slow Market Adoption of Software-Defined Networks (SDN) and Network Functions Virtualization (NFV): CSPs have talked about migrating their hardware-centric networks to more agile and open SDNs for at least 5 to 6 years, but the reality is that the industry has been cautious about the shift, as well as reluctant to move away from reliable and (currently) profitable legacy networks.
- Feuding Standards = Slow Progress on Interoperability: The great strength of telecom communications is interoperability. Each system has to interwork with all the others. It is also the industry’s Achilles Heel, as changes to standards are glacially slow. Complicating things, in this instance, is that there are at least three different industry groups working on standards for interoperability of automating networks: Linux Foundation’s Open Network Automation Platform (ONAP), The European Telecommunications Standards Institute Industry Specification Group on Experiential Network Intelligence (ETSI ISG ENI), and the Facebook-backed Telecom Infra Project (TIP).
- Desire for Centralized AI: Horacio Goldenberg, chief architect for Telefonica Global, told TM Forum in November 2017 what is likely on the minds of a whole lot of CSPs: can we own it? Goldenberg further stated, “But what we are missing really – and I don’t know if anybody knows – is how we would architect the use of AI. Do we have centralized AI, just one brain that controls all customer experiences and addresses internal operational needs? And will it also take care of network needs? Or will we have specialized AI and somehow connect them when required?” Slowing CSPs will be the idea of linking AI assets; if they will be using AI technology for a lot of disparate use cases, should they centralize and/or create a central AI capability and functionality? Or should this be determined use case by case outside AI use case specialists?
The bottom line is, at this point, both the vendor community and CSPs are moving cautiously on leveraging AI widely within the telecom industry.
Read the source article at Tractica.