Enabling energy efficiency of the network management function using autonomic network element intelligence

Abstract

Energy efficiency has become an important issue especially in broadband access networks. It is estimated that over 70 % of energy consumption occurs in broadband access networks, which is higher compared to backbone and metro networks.

To reduce energy consumption, this thesis presents a novel Coordinated Power Performance Management (CPPM) system for the access network to the residential home-gateway scenario. This management framework work in accordance with an energy proportionality definition, providing an intelligent, optimised and multi-level agent based solution which allows idle devices to be in sleep or power-off state under various SLA levels. Energy management mechanisms in particular affect the application’s performance and availability. For example, an application can tune its QoS in accordance with the available power levels of network systems. As a first step towards enabling energy efficiency, CPPM identifies the inter-relationship between energy consumption and application performance in broadband access networks to the residential home-gateway. It then employs an agent based approach that works together in accordance with a flexible set of power-performance objectives and determines the existence of a feasible optimal operating point. The key elements of the CPPM approach include (a) power monitoring and control – using a new energy-aware SNMP MIB. The design of a new energy-aware MIB revealed the power-state objects, power management parameters and related performance attributes that are necessary to monitor and control in the access network to HG scenario; (b) a novel power-performance polygon based methodology which automatically selects energy management settings that keep the pathway from the access network to home-gateway operating close to the optimal operatingpoint; and (c) a multi-agent policy based approach which adaptively managesperformance and power consumption based on varying traffic load duringpeak/worst-peak hours.

As a final step, the efficiency of the CPPM model is demonstrated using experiments. The experimental component demonstrates work related to energy-performance in WiMAX and ADSL broadband access networks, energy vs. traffic metrics and a power-state model vs. delay. In order to configure energy efficient high level policies under CPPM, various constraints such as application type, traffic pattern, the power state transition model and SLA elasticity have been considered. With the logic of the proposed coordination agent, the polygon methodology and underlying policies allow an automatic response according to network conditions. The results of high level policies have been compared to baseline conditions which show energy efficiency, feedback-loop based and a coordinated solution in a heterogeneous accessnetwork to home-gateway scenario.

Date of Award	Mar 2014
Original language	English
Sponsors	Vice Chancellor's Research Scholarship (VCRS)
Supervisor	Gerard Parr (Supervisor) & Philip Morrow (Supervisor)