Yazılım tanımlı ağlar için yeşil yönlendirmede sinyalleşme maliyetini minimize etme

Currently, with the development of next generation networks, there is an urgent request for reducing energy consumption in telecommunication networks due to its environmental impact and potential economic benefits. Some research studies show that energy consumption in communication networks is mainly related to active network elements such as communication links. Based on this approach, several energy management techniques, generally known as green techniques, have been proposed. Their main goal is to minimize energy consumption by routing network traffic through a set of network resources and powering off the remaining unused resources. The current green techniques impose a signaling overhead on the routing system due to selectively powering off/on the network resources and require a centralized network management paradigm to determine which devices to switch off/on. However, the existing networking paradigms including OSPF or IBGP networks do not provide such a centralized management. Recently, the idea of software defined networking (SDN) has been put forward. SDN is a networking paradigm in which the network control is (logically) centralized in the controller and network devices are simple forwarding devices that can be programmed by the controller. The centralized management of the network makes SDN a proper paradigm for developing green routing techniques. In this thesis, we address the trade-off between energy efficiency and signaling overhead in a software defined network (SDN). We formulate two green routing problems which minimize signaling overhead in an SDN network with a single domain while taking the network energy consumption into account. We formulate these optimization problems using Integer Linear Programming (ILP). Our proposed ILP formulations are Signaling Overhead-Aware Green Routing (SOAGR) and Extended Signaling Overhead-Aware Green Routing (E-SOAGR). The ILP formulation of E-SOAGR, which is an extension of the ILP formulation of SOAGR, presents a more general model by introducing additional features such as multipath routing, taking into account routing table sizes of the forwarding hardware and considering different bandwidth requirements for demands. We implement the two green routing models in the IBM ILOG CPLEX Optimization Studio and find optimal solutions for both SOAGR and E-SOAGR models. However, SOAGR and E-SOAGR are computationally very difficult problems. Therefore, we develop polynomial-time heuristic algorithms to find near-optimal solutions for both models. We compare the performance of our proposed green routing models and heuristics by conducting experiments on both random and realistic network topologies. The obtained results demonstrate the tradeoff between energy efficiency and signaling overhead.