Yazılım tabanlı hücresel ağlar (YThA) için uzaysal-zamansal veri katmanı tasarımı

The tremendous increase on mobile data traffic has stressed conventional cellular networks recently. Network management has also become difficult because of enormous traffic demand. At this point, a flexible and dynamic control mechanism is needed that diminishes the complexity in physical topology. SDcN, is one of the novel next generation approaches that makes it easier to orchestrate physical devices in Data Plane with its centralized control fashion. SDcN paradigm provides a simple management strategy for network equipments in physical topology by moving their control logic to Control Plane. These equipments in physical topology release their complex decision heuristics to the Controller, so they become simple, dummy equipments. However; on one hand, SDcN provides scalability and flexibility on network management with dummy OpenFlow (OF) switches and its nature of centralized authority; on the other hand, these properties cause spatial and temporal complexity in the Data Plane that should not be regarded according to Quality of Service (QoS) of a flow. Spatial complexity, described in this thesis as Memory Usage and/or Hardware Cost in an OF switch, should be minimized by removing redundancy in OF switch flow table pipeline. Temporal complexity, defined in this thesis as Flow Forwarding Delay, should also be reduced by lowering number of comparison in flow table pipeline to enhance QoS of a flow. There are many studies in the literature that try to minimize spatial-temporal complexity in OF switch architecture and also to enhance QoS of a flow. However; most of these works violate dummy characteristic of OF switches because of the fact that they implement some wiser decision mechanism such as hashing, caching and search heuristics within the switch architecture. Any mechanism offered to lower spatial-temporal complexity of OF switch architecture should not violate dummy characteristic of these devices. Otherwise, management of physical devices by a central controller becomes slower, more difficult and complicated. Therefore, we propose a novel OF switch architecture to lower spatial-temporal complexity by using Multi-stage Switch model in flow table pipeline. The proposed model is able to forward a flow by less header matching with its architectural model. The proposed model also protects dummy characteristic with a little alteration in OF protocol header fields. In order to define the next table in the pipeline of proposed Multi-stage model, an additional field called next table identifier is integrated with existing OF protocol header entries. Flow forwarding in each pipeline of Multi-stage model is performed according to this next table identifier. Moreover, the advantages of NoC paradigm are also applied to OF forwarding mechanism with such a Multi-stage switch model by assigning a microcontroller to each flow table in each of pipeline stages. Consequently, thanks to multiple pipeline stages that can serve flows concurrently, the number of flows served per unit time is increased and the QoS of each flow is enhanced. For the performance investigation of proposed model we consider Queuing Theory in the light of a spatial complexity parameter (C) and a temporal complexity parameter (T) defined in this work. Total memory usage in a flow table pipeline is measured with spatial complexity parameter and flow forwarding delay is meausered in terms of temporal complexity parameter. These parameters are considered during the performance comparison between conventional OF switch architecture and proposed MsOF switch architecture with {CON} and {MsOF} indexes respectively. According to performance evaluation results, MsOF switch architecture is less spatial complex than conventional one as the number of input ports in an OF switch (N) increases that corresponds denser topologies. MsOF gains much more memory space by deploying more tables with lower memory sizes, total of (n*k). Moreover, MsOF switch model has less temporal complexity compared to conventional OF switch model, especially in urban areas. With such a Multi-stage deployment for flow tables, proposed MsOF architecture provides approximately 7 times less flow forwarding delay, T_{MsOF} < (7 x T_{CON}), compared with a conventional OF Switch by virtue of processing flows at different stages of pipeline simultaneously.