Modelling of Virtual Radio Resource Management for Cellular Heterogeneous Access Networks Sina Khatibi, Luis M. Correia INOV-INESC / Instituto Superior Técnico, University of Lisbon Portugal Email: {sina.khatibi, luis.correia}@inov.pt Abstract - This paper proposes a model for the management of virtual radio resources. The virtualisation of radio resources is based on aggregating and managing all the available physical radio resources in a set of infrastructures. In this approach, Virtual Network Operators (VNOs) use wireless connectivity in the form of capacity per service, and do not have to deal with physical radio resources at all. Due to the isolation among instances and network elements abstraction, this virtualisation approach leads to a more efficient and flexible Radio Access Network. VNOs’ services are provided by virtual resources, based on contracts and defined SLAs (Service Level Agreements). The details of the proposed model for the management of virtual radio resources, its relationship with physical resources managers, the estimation of network capacity, and the allocation procedure of data rate to the service of each VNO are described. A practical heterogeneous cellular network is considered as a case study. Results show an increase of resource usage efficiency up to 6%, presenting how the virtual radio resource management allocates capacity to services of different VNOs, when they have different SLAs and priorities. I. INTRODUCTION Recent studies foresee future mobile networks supporting a significant number of users requesting different services and applications, each one with different Quality of Service (QoS) requirements [1]. It is also believed that network operators have to increase significantly the capacity of their wireless networks, in addition to reducing the cost per bit. This increase is a non-trivial task, since the available radio resources are scarce, and the lack of spare spectrum makes the situation even worse. In addition, the traffic load is not always the same during day, changing in time and location. The provisioning of current mobile Radio Access Networks (RANs) is based on peak hour traffic, which imposes an additional cost. It is well accepted by operators that infrastructures should be shared, in order to reduce operational costs. Based on [2], infrastructure sharing can be categorised into passive, geographical and active ones. Passive sharing is referred to the sharing agreement of fundamental establishments, such as tower frame, equipment houses, and power supply. The service area can also be divided into several regions, each operator being in charge of some of them, which is known as geographical sharing or national roaming [3], leading to the achievement of a full coverage in a shorter time. Active sharing refers to the reuse of backhaul, base stations, and antenna systems; based on the analogy presented in [4], for comparing machine and network virtualisations, active RAN sharing can be claimed to be equivalent to the multitasking on machines. Network Function Virtualisation (NFV) is considered to transform the way operators architect their networks [5]. However, NFV does not contain any sharing at all [6], rather meaning the setup of a network of individual virtualised components, such as nodes, links, and routers. The key historical advantages of virtualisation can be taken as an efficient usage of resources, besides isolation. In this case, the novelty is to create complete virtual RANs over the same physical infrastructure [7]. Although the virtualisation of RAN may sound similar to active RAN sharing, they correspond to very different concepts. As illustrated in Fig. 1, the key concept of Virtual RAN (V-RAN) corresponds to the one of Virtual Machines (VMs). In RAN virtualisation, in contrast to RAN sharing, the physical infrastructure is not transparent to the clients. By means of isolation, services with different protocols, algorithms, and requirements can be offered over the same physical infrastructure. Since the two virtual entities of a virtual network are completely isolated from each other, the problems of one of them do not affect the other, thus, the network has a lower downtime compared to shared networks. Fig. 1. Comparison between Virtual Machine and Virtual RAN. The novelty of this paper is on the proposal of the concept of virtual radio resources, and on a new model for the management of these resources. The model has two main parts: estimation and allocation of radio resources. Through this model, the support of different types of Service Level Agreements (SLAs) and contracts is possible. The remaining of the paper is organised as follows. Section II addresses the concept of Radio Resource Management (RRM) in V-RANs, by presenting the model and resource allocation technique. Section III describes case study scenarios, followed by numeric results in Section IV. Conclusions are drawn in the Section V. 2014 IEEE 25th International Symposium on Personal, Indoor and Mobile Radio Communications 978-1-4799-4912-0/14/$31.00 ©2014 IEEE 1152