Presentation Title

Energy Efficient EPONs

Location

Nessmith-Lane Atrium

Session Format

Poster Presentation

Research Area Topic:

Engineering and Material Sciences - Electrical

Abstract

Ethernet Passive Optical Networks (EPONs) offer a path toward low-cost energy efficient networks. The solution EPONs offer to the data bottleneck has led to their use in markets today. Information and communication technology (ICT) is estimated to contribute for up to 10% of world’ss carbon footprint. If we can increase the energy efficiency then we can impact the effect ICT’s have on the environment. Energy-efficient EPONs offer a solution to the ongoing problem with energy consumption, but it is possible to make the efficiency better. The solution we are purposing is to use an adaptive, or dynamic, time cycling approach. The current approach uses a fixed time cycle with the dynamic bandwidth allocation (DBA) fluctuating with data traffic. If we apply the same logic that as there is more data traffic the optical line terminal (OLT) polls faster and with less data traffic the OLT polls slower. Optical network units (ONU) account for most of power consumption because there are multiple ONU’s per OLT. While the OLT may always be actively and receiving and sending data, the ONU’s are not constantly sending and receiving usable data. If we increase the downtime of the ONU during low traffic periods then we can decrease the power consumed during those periods. 70% of power consumed by ICT’s is from telecommunication and 85% of that power is consumed by idle devices. In EPONs, the ONU is always powered on even if the ONU is idle which consumes a significant amount of power. Therefore, the idea of cyclically turning the ONUs off (sleep-cycle) was proposed to reduce the unnecessary power consumption making the EPON energy efficient. However, this cyclical sleep-cycle is fixed which is not optimized for the different amount of incoming traffic. The fixed sleep-cycle can result in long queueing delays under high traffic rates, while waste more energy under very low traffic rates. In this paper, we are proposing to use an adaptive sleep-cycle in EPONs to optimize queueing delays while maintaining an optimal energy-efficient network. Sleep-aware ONUs are only 40% efficient while our proposal would allow for greater efficiency at the ONU.

Presentation Type and Release Option

Presentation (Open Access)

Start Date

4-16-2016 2:45 PM

End Date

4-16-2016 4:00 PM

This document is currently not available here.

Share

COinS
 
Apr 16th, 2:45 PM Apr 16th, 4:00 PM

Energy Efficient EPONs

Nessmith-Lane Atrium

Ethernet Passive Optical Networks (EPONs) offer a path toward low-cost energy efficient networks. The solution EPONs offer to the data bottleneck has led to their use in markets today. Information and communication technology (ICT) is estimated to contribute for up to 10% of world’ss carbon footprint. If we can increase the energy efficiency then we can impact the effect ICT’s have on the environment. Energy-efficient EPONs offer a solution to the ongoing problem with energy consumption, but it is possible to make the efficiency better. The solution we are purposing is to use an adaptive, or dynamic, time cycling approach. The current approach uses a fixed time cycle with the dynamic bandwidth allocation (DBA) fluctuating with data traffic. If we apply the same logic that as there is more data traffic the optical line terminal (OLT) polls faster and with less data traffic the OLT polls slower. Optical network units (ONU) account for most of power consumption because there are multiple ONU’s per OLT. While the OLT may always be actively and receiving and sending data, the ONU’s are not constantly sending and receiving usable data. If we increase the downtime of the ONU during low traffic periods then we can decrease the power consumed during those periods. 70% of power consumed by ICT’s is from telecommunication and 85% of that power is consumed by idle devices. In EPONs, the ONU is always powered on even if the ONU is idle which consumes a significant amount of power. Therefore, the idea of cyclically turning the ONUs off (sleep-cycle) was proposed to reduce the unnecessary power consumption making the EPON energy efficient. However, this cyclical sleep-cycle is fixed which is not optimized for the different amount of incoming traffic. The fixed sleep-cycle can result in long queueing delays under high traffic rates, while waste more energy under very low traffic rates. In this paper, we are proposing to use an adaptive sleep-cycle in EPONs to optimize queueing delays while maintaining an optimal energy-efficient network. Sleep-aware ONUs are only 40% efficient while our proposal would allow for greater efficiency at the ONU.