Driven by the requirements from traffic volume versatile services and spectrum scarcity, the concept of cloud network and cognitive radio could become new features in the next generation mobile and wireless access networks. For example, to deploy LTE femtocells and Wi-Fi networks in the TV white spaces provides a new way for traffic offloading and enables spectrum sharing as a cognitive radio network. In this paper, we propose and prototype a software defined network architecture with the OpenFlow protocols for heterogeneous network spectrum sharing in the TV white space. Then, we analyze the controller architecture for cognitive radio and the OpenFlow enabled infrastructure architecture. We implement a prototype of software defined cognitive radio network with this SDN controller and LTE/Wi-Fi network simulator. Some essential control functions of cognitive radio are implemented and new potential scenarios based cognitive radio networks are given.
With the trend of network design moving from the network centric to the client centric, the concepts of software defined network (SDN) and network function virtualization (NFV) are candidate techniques for the next generation Internet as well as the 5G mobile communication. The service-oriented network envisions people access the information just like the electricity, water and gas in the daily life, so we argue that radio spectrum usage is also a service. Unfortunately, the current radio access networks(RAN) environment is fundamentally heterogeneous and they are isolated each other, such as LTE, Wi-Fi and WCDMA. With a view of cognitive radio, a new spectrum usage paradigm is required in order to allow the mobile user access any reachable heterogeneous spectrum around it. Furthermore, quality of user experience(QoE) can be guaranteed . Therefore, we believe SDN/NFV architecture is potential to enable novel spectrum usage with the cognitive radio techniques in the next generation mobile internet or 5G.
The traffic volume for the future mobile Internet increases in form of orders of magnitude in the coming year and leads the eve of big data coming. The physical transmission and spectral efficiency is difficult to get promoted further, because the spectrum efficiency of LTE has reached within the twenty percent, as is very close to the shannon capacity limit. As a result, cognitive radio is an important way to offload traffic for the macro-cell networks to the dense deployed femtocells and Wi-Fi networks in the 5G. The dense small-cell deployment with LTE femtos and Wi-Fi without cell planning may appear to provide more spectrum occupancy per user by reducing the number of users per cell. However, this leads to a complex network management under the network architecture already defined. Therefore, the new concept of SDN/NFV provides us a promising way to manage interferences and enable dynamic spectrum sharing.
Based on the two observation, we are dedicated to define a software defined wireless network (SDWN) architecture for the heterogeneous and broadband RANs to simplify network management with a novel spectrum usage paradigm oriented to 5G. As an example, the coexistence of LTE femtocells and Wi-Fi has been investigated in TV white space. The main contributions in the paper can be summarized as two. One is a multi-tiered cloud architecture proposed for the SDWN. A prototype for cognitive radio scenarios defined with this SDN architecture has been implemented for cognitive radios. The other is a developed spectrum monitor and spectrum manager are proposed under the SDWN architecture. New spectrum access/handover scenarios are supported in a SDWN based cognitive radio networks.
The organization of this paper is defined as as following: In the section II, we review the background of SDWN and the recent developed concept of 5G oriented cognitive radio. Then, we define a multi-tiered SDWN architecture for heterogeneous RANs in the section III, which enables the spectrum sharing in TV white space and decouples the data/control planes with an Openflow interface in the infrastructure. Section IV introduces an initial prototype design and implementation with the new features for the SDWN enabled spectrum sharing environment. In the end, we make conclusions for the paper in the section V.
II. BACKGROUND OF COGNITIVE RADIO AND SDWN
This paper is motivated to develop cognitive radio with a practical usage in the software defined network architecture, because our argument is spectrum usage is also a service in 5G mobile networks and spectrum sharing policy could be defined and reprogrammable by software as a SDWN controller. The extensive measurements carried out in the major urban areas in the world have shown that the spectral efficiency is poor in a range of 300MHz to 3GHz with a high spatio-temporal variation.The previous research on the cognitive radio was focusing on improving the radio spectrum utilization resources mainly within the primary-secondary user models in the UHF TV band. For example, the scenario of LTE femtocell and WiFi deployed in TV white space allows the mobile operators to improve coverage and capacity of their network and reduce their CapEx and OpEx.
However, in the 5G wireless communications, to meet the challenge requirements of huge capacity, massive connectivity, high reliability and low latency, cognitive radio is expected to play an important role in two aspects. First, since the spectrum band for 5G will be extended to even 60GHz millimeter-wave range, the usage of cognitive radio can be extended to improve the spectrum utilization within new types of spectrum sharing models, such as dynamic licensed spectrum leasing. Second, 5G will take aggressive spatial reuse of spectrum as a enabler with new techniques such as massive MIMO and Ultra-dense Deployment. In this context, cognitive radio can be used to control the interference issues from space, frequency and time domains with a very smart manner. On the other hand, the controller-based SDWN architecture provides the coordination opportunities for heterogeneous wireless network management on spectrum and interferences, which is defined a new service on controller and implemented as an essential component of the network operation system, e.g. Floodlight. In this paper, we design an layered SDWN based architecture for the coexistence scenario in a heterogeneous network of the LTE femtocell and Wi-Fi network in the TV white space with Openflow. Openflow is a protocol, initiated at Stanford, to enable switches on the wired network to be intelligent and programmable via a standardized interface. The Open-Flow protocol is standardized by the ONF to mitigate operation cost while simply network management and speed network innovation. Furthermore, Openflow has been extended from wired network to wireless infrastructures as OpenRoad. The OpenRoad is dedicated to solve mobility problem with a topology with 5 switches, 30 Wi-Fi APs and a WiMax AP. The SDN controller deals with the seamless handover problem between Wi-Fi and WiMax successfully. Till now, Openflow has been used in wireless mesh network, sensor networks, and cellular networks. The flowvisior is an initial NFV techniques for SDWN.
III. ARCHITECTURE ISSUES IN SDN BASED COGNITIVE RADIO NETWORKS
A. SDN Architecture for Cognitive Radio Networks
With the vision of cognitive radio for the next generation mobile communication, we assume whatever eNodeB or STA can be reconfigurable on the part of baseband and radio in a large range with the software defined radio techniques, which is essential support to implement the cognitive radio defined in this paper. We attempt to analyze design requirement with the vision and assumptions. The new proposed control plane is expected to be responsible for interference management and control in the coexistence of heterogeneous wireless network. The mechanism of interference appraisal and event detection should be provided with the network state monitoring function. Once the controller detects the event happened, controller will decide to send clients the spectrum handover command. This kind of method for spectrum mobility should be apparent to clients, and the QoE should be considered within it.