2019 Global AI Challenge

In June 2019 the GSMA announced the four mobile network operators, STC, Telenor, TELUS and Turkcell, who were selected as challenge leaders in the first GSMA Global AI Challenge. The Challenge investigated three specific areas: connectivity in rural areas, mobile energy efficiency and enhanced services in urban areas.

The Challenge comprised an intensive five-day hackathon and is part of Turing’s Data Study Group (DSG). It is focused on raising the profile of AI as a key enabler in the mobile industry and the transformational opportunities it provides, while also exploring how science, society and the economy might benefit.

The DSGs took place in September 2019 with initial results available at MWC Los Angeles 2019.

1. STC: Bandwidth allocation for mobile users: a solution for urban and rural areas

Different mobile users have different bandwidth needs. Bandwidth availability usually depends on the user density in the targeted area. Reserving bandwidth is usually not desired in mobile networks as it is considered a waste of network resources. This DSG explored how mobile network operators can make available the necessary bandwidth to specific urban or rural areas, which lag behind either because of high population density or lack of necessary infrastructure.

At the peak time of the AI hype when operators are seeking to understand “How should telco’s onboard AI and its capabilities?”. The GSMA Global AI Challenge initiative was the right method at the right time, creating a collaboration between telecom operators and top AI institutes/researchers to accelerate innovation and development of AI solutions within the telecommunication world.  STC’s participation in the GSMA AI challenge was very successful kick-start on STC’s journey of applying AI in its businesses.


2. TELUS: Understanding the influences of network measures on customer perception of network reliability

TELUS regularly conducts surveys with its customers to gather feedback and identify service improvement opportunities to prioritize investments in a way that reflects their customers’ needs. This DSG focused on understanding how the network and the customer experience (whilst using the network) influence the results of this survey. More specifically,  TELUS wanted to understand: (i) how accurately the customer’s experience of reliability on their network can be predicted, and (ii) what are the main drivers of network performance to influence their customer’s rating of their experience.

“Attending the Global AI Challenge provided us with new perspectives on how to approach our data, and a new appreciation of how rich our data is in terms of providing insights to our business as we continue to explore different permutations of the data set that we assembled for the original 2019 challenge. Participating in this event allowed us to work with the best and the brightest in AI, and is a great avenue for recruitment of top AI talent. This event was also invaluable in jumpstarting collaborations with other operators globally so that we can share best practices about AI and work together to solve common challenges across our industry.  Additionally, we feel that the Global AI Challenge enables the GSMA and the telecommunications industry to build a reputation of driving innovation through AI leadership, to which TELUS is very happy to contribute.”


3. Telenor: Green radio – dynamic power saving configuration for mobile networks

Mobile networks waste energy by keeping too many radio-cells turned on when demand is low during off-peak. This challenge was about automating next-day power saving schemes for each individual cell tower in a country, based on current load and expected demand profile in the area.  The solution should optimise power saved while avoiding negative impact on the user’s network experience. Access the post-Challenge report here.

“The AI Challenge gave good visibility for our project both internally and externally. It was also an efficient way to learn techniques and methods that will be useful for solving other related future problems.  The AI challenge was also a very good opportunity for meeting experts from other operators and discuss common problems”


4. Turkcell: Improving security and customer experience – detecting and locating network jammers

In urban areas jammers cause interruption on mobile 3G and 4G communication networks which leads to severe service quality deterioration. This situation causes customer complaints and time and labour costs during detection studies. Based on network data including some network service quality indicators and jammer geographical location, this DSG investigated methods for real-time detection of jammer presence, identification of jammer type and its location.

“We were very pleased to have been part of the Global AI Challenge. The hackathon that took place in London offered a unique experience, for us where Turkcell R&D personnel and network planning experts joined distinguished researchers from Alan Turing institute to find creative solutions to our jammer detection and localization problem. Having different academic and professional backgrounds and thus diverse perspectives, the team developed valuable insights and came up with alternative solutions. Inspired by the proposed algorithms and approaches from the hackathon, we continued to work on the problem and have been creating a tool to identify and localize active jammers in our networks to reduce the burden on network planning and optimization teams. In addition, this challenge also provided a good networking opportunities with  telecom operators from the different parts of the world. We have since been collaborating with one of the participating telecom operator on a research grant proposal on a topic related to AI for telecommunication networks. Once again, we would like to thank GSMA and Alan Turing Institute for the opportunity and their gracious hospitality.”


Further information can be found on the Turing website here.