Yazılım tanımlı heterojen ağlar için servis derecesi tabanlı kontrolör yapısı

Recently, mobile data traffic has rapid proliferation due to increased user demands in wireless networks. The conventional macrocells are stressed because of huge increase in user requests. Therefore, they are not adequate to serve indoor users and keep their service quality, i.e. Grade of Service (GoS), in acceptable levels. In order to meet indoor mobile users with higher service quality, i.e. lower GoS, heterogeneous networks (HetNets) are deployed as smallcells, vary according to coverage area such as macrocell, picocell and femtocell etc. However, conventional HetNets cannot handle increased user flow requests by using same number of physical resources, due to inefficient usage of these resources. The physical resources are clustered and distributed to smallcells statically in conventional HetNets by the operators. Therefore, this static clustering leads an ineffective resource usage. In other words, the number of flows that can be served with acceptable GoS, is restricted into low levels. For example, using two smallcells, each having ten physical resources, can handle more traffic flows than using four smallcells, each having five physical resources, while considering same GoS level. In other words, physical resource efficiency has huge decrease if resources are statically clustered and distributed to small cells. This is also proved by modeling two systems using markov chain and mathematically analyzed according to queuing theory as given in Appendix. Current solutions that try to increase resource efficiency in the literature, are based on switching technologies between radio access technologies (RATs) such as offloading, and cognitive radio technologies etc. Nevertheless, these solutions cannot reach high number of flows that can be served in acceptable level due to limitation of statically assigned physical resources. Therefore, the restriction caused by static clustering of physical resources and distribution to smallcells, should be removed with adaptive clustering approach. In this paper, we solve this ineffective static resource assignment by proposing a novel queuing-theoretic Software Defined HetNet (SDHetN) model which orchestrates topology using adaptive and scalable flow management heuristics. The proposed SDHetN takes its flexible and scalable characteristics thanks to two algorithms; the Topology Control Algorithm (TCA) and the Flow Admission Control Algorithm (FACA). Specifically, the proposed TCA clusters several OpenFlow (OF) switches using the flows' GoS in order to optimize physical resource assignment. The proposed FACA fairly distributes each Flow Authority Virtual Switch (FAVS) that are created in TCA by grouping several switches virtually. We also propose a thread-based parallelization in TCA and FACA increasing the response time and service rate of the SDHetN Controller. %The performance of SDHetN is evaluated by 48 different scenarios and it is shown that SDHetN provides a scalable and fair flow management according to different performance metrics.