Research On Energy Saving Technology In Software Defined Networking

Software defined networking (SDN) decouples the control function of the network from data forwarding plane. It controls the data forwarding devices through a logically centralized and programmable control plane. Due to the properties such as programmability, strong controllability and high flexibility,the application field of SDN becomes more and more widely. However, with the rapid development of SDN, it still faces the energy waste problem as do traditional networks. Considering the rapidly rising energy costs and the in-creasingly rigid environmental standard, the energy waste problem can not be ignored.Though some energy saving technologies in tradition networks still can be used in SDN, some new problems arise because of the separation of control and forwarding in SDN architecture. In this dissertation, combining with the energy saving technologies in traditional networks, some strategies for the energy con-sumption problem in SDN are studied from data transmission, controller place-ment and performance optimization of the networks considered energy saving.The main contributions of this dissertation are summarized as follows.(1) Since current energy saving technologies that can be applied in SDN didn't consider the working status of the network before optimizing energy con-sumption, route oscillation problem is easily to be caused. To address this prob-lem, a virtual topology based energy saving scheme is proposed. First, in the proposed scheme, controller maps the current working status of the network into a virtual topology. According to this virtual topology and the real physi-cal topology of the network, routing algorithm calculates paths for new com-munication demands under the premise of keeping the transmit paths of the running traffic flows. The stability of the network therefore improves. Sec-ondly, through analyzing the energy saving problem in the proposed scheme,the problem is modeled as a mixed integer linear programming (MILP) model,and classified into the multi-commodity flow problem, which is well known as a NP-hard problem. Thirdly, Considering the high complexity of the problem,two routing algorithms that can be used in the scheme are proposed. One is based on the constant weights of the links in networks; and the other one ad-justs the weights of the links dynamically according to their working statuses.Through analyzing their complexities, the algorithms are verified to be polyno-mial algorithms. Finally, the performance of the two algorithms is evaluated both in the ability of saving energy and the impact on the other parameters of the network.(2) To reduce the energy consumption of the control network, an energy aware controller placement scheme with considering delay and load balance of the control network at the same time is proposed. First, to ensure the perfor-mance of the control network, the effect of load balance among the controllers on the target delay in the controller placement problem is analyzed from both av-erage propagation delay and the worst propagation delay through modeling and simulation. Then, the energy aware controller placement is modeled as a binary integer programming (BIP) model under the condition of meeting the delay and load balance at the same time. The conclusion that the model only can be used in small scale networks is drawn since the problem is classified into the facility location problem. After that, for large scale networks, a heuristic energy saving controller placement algorithm based on an improved genetic algorithm is pro-posed. The proposed algorithm is verified as a polynomial algorithm through analyzing its time complexity. Finally, the energy saving performance of the two methods is evaluated by simulations; and the trade off between energy con-sumption and delay is also observed in the simulation.(3) To reduce the impact on the network performance caused by energy saving, a performance optimization algorithm based on algebraic connectivity for green networks is proposed. First, on the basis of analyzing the relation between the algebraic connectivity of the network topology and network per-formance, the performance optimization problem of green networks is defined.Then, a heuristic algorithm based on neighborhood search is proposed to solve the problem, which improves the algebraic connectivity of green networks as far as possible under the premise of energy saving. Finally, the improvement in the performance of green networks achieved by the optimization algorithm is verified through simulation, and the effect on the performance of the optimiza-tion algorithm caused by its parameters is also evaluated.