Machine Learning as a Powerful Tool to Enjoy Future Intelligent Wireless Networks

Abstract

Wireless communication systems are irreversibly changing our lives. Today, wireless networks are extremely complex systems and they are evolving towards more complex ones because of the increasing diversity and heterogeneity of applications, devices, quality requirements and standards. At the same time, resources used by wireless communications are either naturally limited e.g., time, spectrum, or need to be optimally exploited e.g., energy, computation, infrastructure. Hence, traditional resource allocation approaches that are based on optimization and heuristic techniques start to show their limitations. Those approaches are often centrally-managed, reactive, and not adaptive. They also require a huge amount of control data exchange. Hence, there is a need for new approaches to provide adaptive, proactive and self-organized networking solutions. Thanks to the availability of increasingly powerful computing systems and of huge amount of data that can be efficiently exploited in wireless networks, we envision the employment of machine learning techniques in order to achieve intelligent, adaptive, resource-efficient and data-driven future wireless networks. This talk discusses how wireless network designers and operators can employ and adopt advanced machine learning techniques for adding predictive and adaptive intelligence to the system. The state of the art of using machine learning in wireless networks will be deeply discussed and some interesting issues for new research avenues will be identified.