Ant colony optimization based cooperative energy optimization framework with delay assurance for the future green internet

Abstract

Recently, energy efficiency has become a critical issue of the modern Internet as the traffic load exponentially increased. However, the current Internet is not designed to optimize the energy consumption

rather, it tends to waste considerable amounts of energy. To cope with this situation, many researches on energy efficient technologies have studied during the decade. Although the above numerous previous research can be a good direction to enhance energy efficiency, they fundamentally have some limitations regarding 1) the diversity of energy profiles, 2) the cooperations among several layers, and 3) the end-to-end delay performance.

To overcome these limitations, this dissertation proposes a novel end-to-end delay guaranteed cooperative energy optimization framework by making use of the swarm intelligence of individual artificial ants from ant colony optimization (ACO).

First, this dissertation analyzes and summarizes related works about energy efficient technologies for the green Internet and the well-known ACO technique. For the ACO, this dissertation explains the overview of the ACO technique. After that, the dissertation explains related works about energy efficient technologies. In particular, previous works on a single layer approach are classified into three categories: network-layer, system-layer, and device-layer approaches. According to each layer, the dissertation explains the previous research in detail. Furthermore, the dissertation explains previous researches on a cooperative layer approach in detail.

Second, this dissertation proposes an end-to-end delay guaranteed cooperative energy optimization (ECONet) framework for the future green Internet. To do that, the dissertation newly defines an end-to-end delay guaranteed cooperative energy optimization (ECO) problem. The ECO problem is derived by considering the device-layer as well as network and system-layers in turn. After that, to help you understand the ECONet framework easily, the dissertation briefly describes the general concepts of the proposed ECONet such as its approach to cooperatively optimize energy efficiency, the conceptual operations and data logics for enabling this approach, and the procedures of how to provide the cooperations among the several layers. In particular, the ECONet framework consists of three algorithms in network and system-layers to enhance energy efficiency while guaranteeing delay requirements of flows. Based on this structure, this dissertation proposes three algorithms for ECONet: an enhanced delay guaranteed energy profile aware routing (eDEAR) algorithm in the network-layer and a delay aware sleep control (DASC) and a priority based burst scheduling (PBS) in the system-layer. According to each layer, the dissertation explains each algorithm in detail.

Third, this dissertation proposes a delay guaranteed energy profile aware routing (DEAR) algorithm that can heuristically find the delay guaranteed least energy path in the network-layer. In general, the existing energy efficient routing algorithms just only consider On/Off profile. However, this approach is not well-fitted to energy-aware network environment

it lacks to consider various energy profiles. To cope with this limitation, the DEAR algorithm is proposed.

Fourth, this dissertation proposes two algorithms: a delay aware sleep control (DASC) and a priority based burst scheduling (PBS) in the system-layer. The former reconfigures the power state of a system using end-to-end delay information and buffers packets according to the traffic type. The latter transmits the buffered packets according to their priorities.

Last, this dissertation evaluates the performance of ECONet regarding energy efficiency and delay guarantee as a cooperative energy optimization framework through extensive simulations on real network topologies. To do that, the dissertation first analyzes and discusses on the performance of ECONet and other approaches according to the layer. After that, the dissertation additionally evaluates the performance of ECONet compared with the other comparative framework that is a combination of the existing approaches in the system- and the network- layers. Through the performance evaluation of ECONet, we prove that the ECONet framework is superior to the other existing energy efficient technologies in the energy efficiency and delay guarantee perspectives in energy-aware networks.

Consequently, this dissertation firstly designs the ECONet framework to be capable of handling the dynamic network status more efficiently and cooperations among the several layers. The ECONet framework is the cooperative framework to not only optimize energy efficiency but also guarantee delay requirements of flows by using the swarm intelligence of artificial ant colony. Through the ECONet, network providers can not only adopt various energy efficient technologies to their network equipment efficiently but also keep their revenues by supporting delay-sensitive services successfully.