SDN Based Inter-Technology Load Balancing Leveraged by Flow Admission Control Suneth Namal ∗ , Ijaz Ahmad † , Andrei Gurtov ‡ and Mika Ylianttila § ∗†§ Department of Communications, University of Oulu, Finland ‡ Department of Computer Science and Engineering, Aalto University, Finland Email: [ ∗ namal, † iahmad]@ee.oulu.fi, ‡ gurtov@hiit.fi, § mika.ylianttila@oulu.fi Abstract—In this paper, we have followed the idea of exploiting inter-technology load balancing leveraged by flow admission control with software defined networking (SDN). By means of the highly flexible interfaces, SDN is simply effective in the 5G network architecture where load balancing and dynamic flow admission control are considered as essence of networking. In one hand, load balancing presented in this paper reveals a drastic reduction of unsatisfied-user percentage almost by five times while, on the other hand, it enhances per-flow resource allocation more than 200%. In addition, this proposal substantially off-loads the core-network, avoids over-utilizing the cellular-network and reserves resources based on cut-off priority. Explicitly, it allows a certain degree of freedom to choose the network options based on user-preference and their priorities. Finally, we compare our load balancing algorithm leveraged by flow admission control with an analytical model to ensure the correctness and efficiency. I. I NTRODUCTION Deployment of virtualization and SDN technologies are foreseen mainly in the transport level, and user/data plane procedures will presumably remain intact. However, there are ongoing researches focusing on the possibilities of using SDN technologies for user-level management. Load balancing is such a classical networking function supported by routers, switches and load balancers [1]. Due to existence of various wireless technologies that are evolved in parallel with the emergence of heterogeneous 5G networks, the adaptive cou- pling based on load status and delay constraints has become a must in 5G network architecture which is inspired by the dynamically diversified traffic conditions [2], [3]. Load balancing based on IP addresses or other primitive methods are unlikely to overload the system heavily, though such basic functionalities are decades old and being proven inefficient for scaling and virtualizing network functions and applications. SDN empowers isolation and redirection of traffic by means of the remote assistance of the controller. It could be real- ized by configuring the flow tables on switches and routers. Adding, modifying, and removing flow rules on networking infrastructure are the major function of remote controller. In- telligent flow configuration with the understanding of physical topology opens-up opportunities to freely develop network applications [2]. Thus, admitting new flows happens to repeat this process. Load balancing in SDN, leveraged by admission control is a technique of manipulating these flows. This could be realized at the centralized controller who can configure flow-paths, such that they optimally utilize network resources and enhance the user-experience. In large networks, it is understood that defining optimal path while admitting to a network is always efficient compared to that of regulating the existing flows. This is the driving force behind load balancing empowered by flow admission. It is empowered by three main components: flow tables, controller and secure channel. Existing load-balancing algorithms assume that the requests are entering to the network through a single point where the load balancers are placed though, there could be several such other choke points in a large network [4]. Because of this, the need for isolating such network functions from infrastructure is growing. Successful load-balancing leveraged by flow admission control optimizes resources, minimizes response time, maximizes throughput, and avoids overloading. In this paper, we introduce a novel SDN based load balanc- ing solution where flow capacity is defined by the number of requested physical resource blocks (PRBs). The results reveal a drastic reduction of the number of unsatisfied and angry users in the network and a substantial improvement of resources allocated per user. This paper is organized as follows. In Section II, we presents our SDN based load bal- ancing architecture. Then, our simulation model is presented in Section III. Section IV describes simulation results and finally, in Section V, we conclude this paper. II. SDN BASED LOAD BALANCING ARCHITECTURE SDN enablers, such as OpenFlow allow to retrieve packet count at switches and routers. In this architecture, the cen- tralized controller or the internetworked set of distributed controllers retrieve the load status through the Load Balancing (LB) application part which is an application programming interface (API) on switches. Then, this information is directly communicated to the controller who will decide how to admit new flows. The ability to control flow tables of legacy networking elements by means of logically separated control plane with an efficient forwarding approach is the benefit of this programmability that allows to overcome load imbalance problem. The SDN based common interfaces that enable inter-system communication is an essence that offers necessary platform for our solution. Managed handover is a must in applications like these. Typically, WLAN handover is initiated by the mobile stations by scanning the surrounding access points/base stations and by selecting the best one with the highest signaling