Georgakopoulos, A; Masouros, C; Mekuria, F; Erfanian, J; Pallois, JP; Tsagkaris, K; Debbah, M; ... Deng, Y; + view all Georgakopoulos, A; Masouros, C; Mekuria, F; Erfanian, J; Pallois, JP; Tsagkaris, K; Debbah, M; Vázquez, M; Dixit, S; Demestichas, P; Kenyon, A; Vinodrai, V; Kumar, V; Deng, Y; - view fewer (2019) Artificial Intelligence in Future Wireless Communication Systems [White Paper]. (Outlook 24). Wireless World Research Forum (WWRF): Zürich, Switzerland.

Text Outlook24 2019.pdf - Published Version Access restricted to UCL open access staff Download (826kB)

Abstract

Artificial Intelligence (AI), Machine Learning (ML) are hot topics nowadays and have achieved huge breakthroughs in the recent years and months. AI is transformative and will be embedded in each link of the overall chain of the communications system. We envision to see AI almost everywhere: in connected devices, machines, objects, automobiles and many connected things but also in all elements making up the network, such as base stations, controllers, core network equipment, operations and management system. However, we need to be aware that this AI proliferation and pervasiveness will generate potential entanglement which will need to be solved. Nevertheless, we can expect AI in each device to process data close to the source to complement the cloud, especially for latency-sensitive and mission-critical applications. This intelligence at the micro level will need to be abstracted to the higher levels for macro control and holistic view of the network. AI in the device will also be useful when the device is disconnected from the network and the cloud. However, AI is not seen as the holy grail of technology by everyone. Some major issues may slow down the adoption of AI by the industry. The first reason may be the difficulty to fully understand the real benefits that AI can bring in daily work despite the current market trend. Another reason is also human, emotional reason: people in the industry may perceive AI as a potential threat to their jobs. Eventually, the AI proponents will have to deal with the above challenges, but even if the AI advocates can overcome these barriers, the need to adapt existing operations processes will be a heavy, cumbersome and tedious task and will need support from the top management to adopt AI. The last obstacle will be the expected transparency of AI. Indeed, delivering AI as a black box may be enough in some cases for basic actions (e.g. like how to optimize parameter settings during a roll-out phase) but would not be acceptable during the optimization or healing phases. Operators who could not be in the position to explain what happened and why it happened could be caught up in explaining to the public, regulators and may suffer legal consequences. Therefore, it would be necessary to be able to explain AI to all stakeholders. In short, for AI in network to be successful, all the above key issues would need to be addressed.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item