Abstract

Admission control in wired networks has been traditionally used as a way to control traffic congestion and guarantee quality of service. Here, we propose an admission control mechanism which aims to keep the power consumption at the lowest possible level by restricting the more energy-demanding users. This work relies on the fact that power consumption of networking devices, and of the network as a whole, is not proportional to the carried traffic, as would be the ideal case [1]. As a result some operating regions may be more efficient than others and “jumps” may arise in power consumption when new traffic is added in the network. The proposed mechanism aims to keep power consumption in the lowest possible power consumption level, hopping to the next level only when necessary. Document type: Part of book or chapter of book

Abstract

In order to meet a tremendous amount of data storage requirement in next-generation wireless networks, an increasing number of cloud data centers has been deployed around the world. The underlying core networks are expected to provide the ability to store data in a dynamic and scalable computing environment. The traditional Internet Protocol (IP) has shown to be restricted due to its static architecture, which accordingly motivates the development of Software-Defined Networks (SDNs). In the SDNs, Traffic Engineering (TE) is simpler and programmable with a controller without the requirement of reconfiguration for all network devices. However, the existing TE algorithm of the SDNs rejects a number of requested flows caused by their undetermined routing paths where only flow bandwidth is considered in path determination. This paper proposes a Quality-of-Service (QoS) based Flow Assignment algorithm which enables the computation of end-to-end path for traffic flows guaranteeing the QoS requirements including bandwidth, end-to-end delay and packet loss probability. Through the Open Source Hybrid IP/SDNs platform, the proposed algorithm is validated and shown to significantly reduce flow rejection rate of up to 50% compared to the conventional approach, and therefore can be used to implement an effective DiffServ mechanism for flow allocation in the SDNs. Document type: Part of book or chapter of book