Cognitive Green Radio for Energy-Aware Communications

In 5G networks, the number of connected devices, data rate and data volume per area, as well as the variety of QoS requirements, will attain unprecedented scales. The achievement of these goals will rely on new technologies and disruptive changes in network architecture and node design. Energy efficiency is believed to play a key role in complementing the 5G technologies and optimizing their deployment, dynamic configuration and management [1]. Within the framework of green communications and networks, especially for next-generation green cellular radio access networks, the GREAT (Green Cognitive Radio for Energy-Aware wireless communication Technologies evolution) initiative, a CominLabs Excellence Center (Laboratoire d'Excellence) and Université Européenne de Bretagne (UEB)project, has mainly addressed the fundamental issues of energy efficiency from various perspectives and angles, leveraging on cognitive techniques, at networking level as well as at thephysical layer level. Various cognitive green radio techniques have been utilized and verified to improve energy efficiency in different application scenarios. Particularly, representative learning algorithms and decision-making schemes (transfer learning, reinforcement learning, combined learning, entropy theory, etc.) [2] have been successfully employed to achieve energy saving under spectral/capacity requirements. In order to improve the balance of energy-efficiency and spectrum-efficiency, the fundamental characteristics of networking traffics have been investigated by taking into account the mobile users' behavior (mobility, location/position/density, social connection, living style, etc.), in various outdoor as well as indoor networking environments[3]. The various led theoretical analysis have been validated through implementation in the real world. As a case in point, we present in this work a Wi-Fi-based platform which implements two main functionalities: first, users clustering and second, base station (BS) & access points (AP) switchingon/off [4]. This paper is organized as follows: in the first section the concept of cognitivegreen radio is presented. In the second section we introduce our RSS-based clustering approach. The third section focuses on topology management under energy consumption constraint. The fourth section presents how these two different concepts were implemented in our Wi-Fi-based platform.