In this letter, we propose a mathematical model for the controller placement problem in Software Defined Networks (SDN). More precisely, given a set of switches that must be managed by the controller(s), the model simultaneously determines the optimal number, location, and type of controller(s) as well as the interconnections between all the network elements. The goal of the model is to minimize the cost of the network while considering different constraints. The simulation results show that the model can be used to plan small scale SDN. When trying to solve larger instances of the problem, the solver is taking too much time and also running out of memory. The proposed model could be used by various enterprises and cloud-based networks to start integrating SDN or plan a new SDN.

No abstract

In this paper, we evaluate the effect of different queueing disciplines and different codecs on a voice over IP (VoIP) network with video (also known as video conference). A real network topology is implemented using Cisco equipment. Experimental results show that low latency queueing (LLQ) and AutoQoS can provide significant improvements over FIFO in terms of delay, jitter and reserved bandwidth for a VoIP call with video. As far as codecs are concerned, they are suitable for different scenarios. It's for certain that G.729 helps when low bandwidth is a must but, without any QoS parameters in place, codecs do not make a good enough difference for acceptable call quality on a congested network.Index Terms-Voice over IP (VoIP), video conference, quality of service (QoS), queueing mechanisms, codecs, case study.

This thesis focuses on deriving exact methods to place controllers on a network for the planning problem of Software Defined Network (SDN). A first mathematical model, referred to as the planning model, is proposed that decides the optimal location to place controllers on a network while minimizing the overall cost. A second mathematical model, referred to as the expansion model, is proposed that determines the changes to optimally place controllers on an existing SDN network. The advantages of the proposed models are to incorporate realistic constrains. A solver optimizes the exact method for both models with many scenarios and the results are compared.In the planning problem, as the input size increased, the time to find the optimal solution also increased in non-polynomial time. The expansion problem results show that changes can be made to existing implementations by altering many of the input variables. The expansion model allows control of how much of the existing network changes.iii To my mother and brothers.May they rest in peace.iv