NSF Award
1745410
Future wireless networks are expected to provide a platform for highly reliable and ultra-low latency information exchange that will revolutionize the way machines communicate and how people interact. A vast number of devices, particularly sensors and actors, is anticipated to be able to connect to a single wireless access point, empowering the deployment of massive Internet of Things devices that generate an enormous amount of information, namely Big Data. The management of both, the wireless connections of the devices and the Big Data they generate, requires substantial changes to legacy wireless networks. The goal of this project is to conduct exploratory research on such changes, which, if successful, will deliver a key corner stone to the future deployment of massive Internet of Things. This deployment will constitute an essential component in various aspects of life, such as improvements in consumer products, e.g. home automation and better consumer oriented entertainment; in health, through smart body sensors and remote, low-cost diagnosis; in more cost-efficient manufacturing; in environmental monitoring, e. g., for enhancing environmental conditions like reduced air pollution; or in the realization of smart cities.<br/><br/>Following the recent great success of artificial intelligence, machine learning, and specifically deep learning in many different applications, the goal of this project is to explore and to perform research on novel deep-learning-assisted autonomous decision-making approaches for the self-management of future wireless communications networks. The problem to be solved is to establish an efficient and effective wireless network self-management architecture that is flexible enough to facilitate the deployment and operation of diverse massive Internet of Things devices. This architecture is expected to be able to exploit hidden information in Big Data generated by the Internet of Things devices through deep learning techniques. The research into this architecture comprises the exploration of novel machine learning algorithms for the intelligent, predictive, and autonomous allocation of radio and network resources at different layers of the wireless network. Ultimately, a novel multi-layer self-organizing network architecture will be introduced that fully exploits the flexibility and capability of future wireless networks for servicing massive Internet of Things. The developed approaches will lay the foundations of substantially enhancing legacy self-organizing wireless networks and will impact the design of future wireless networks, its efficiency, and the realization of various emerging use cases generating Big Data likewise. The results will also provide novel insights into the role of Big Data for the communications industry, potentially fostering the development of new business models and strengthening the industry's national and international competitiveness.
