This article has been accepted for publication on IEEE DySPAN 2015, but has not been fully edited. Content may change prior to final publication. A SDN Approach to Spectrum Brokerage in Infrastructure-based Cognitive Radio Networks

Abstract—Cognitive Radio (CR) is a promising approach to overcome the spectrum crunch faced by today’s enterprise and residential WiFi (IEEE 802.11) due to rapid growth of wireless devices and traffic load. However, the expected high-density CR Networks (CRN) will suffer from similar problems as we see today with WiFi, i.e. any uncoordinated spectrum access will inevitably result in interference between Secondary Users and hence in a low spectral efficiency. In this paper we take advantages of the ideas of Software-Defined Networking (SDN) and cloud computing technology to manage interference in CRN deployments in residential areas. Specifically, we propose a SDN-based CR architecture where a cloud-based centralized controller, the Spectrum Broker (SB), takes control over the spectrum assignment to CR Base Stations (CR-BS). Therefore, the CR-BSs under control report aggregated wireless statistics to the SB. Moreover, by configuring proper rules in OpenFlow-enabled CR-BSs, the SB controller can get up-to-date information about the network traffic condition in the CRN. With this information the SB can perform a very fine-grained topology-, traffic- and channel-aware spectrum al-location. Our architecture, as well as the proposed spectrum allocation scheme, were analyzed by means of emulation within Mininet. Results demonstrate a gain of up to 5 ⇥ as compared to a static spectrum allocation scheme