Energy-saving scheme based on downstream packet scheduling in ethernet passive optical networks
详细信息 查看全文
- 作者：Lincong Zhang ; Yejun Liu ; haveball@yeah.net ; Lei Guo ; Xiaoxue Gong
- 关键词：Ethernet passive optical networks ; Packets scheduling ; Hybrid sleep mode ; Energy-saving scheme
- 刊名：Optical Fiber Technology
- 出版年：2013
- 出版时间：March, 2013
- 年：2013
- 卷：19
- 期：2
- 页码：169-178
- 全文大小：999 K
文摘
With increasing network sizes, the energy consumption of Passive Optical Networks (PONs) has grown significantly. Therefore, it is important to design effective energy-saving schemes in PONs. Generally, energy-saving schemes have focused on sleeping the low-loaded Optical Network Units (ONUs), which tends to bring large packet delays. Further, the traditional ONU sleep modes are not capable of sleeping the transmitter and receiver independently, though they are not required to transmit or receive packets. Clearly, this approach contributes to wasted energy. Thus, in this paper, we propose an Energy-Saving scheme that is based on downstream Packet Scheduling (ESPS) in Ethernet PON (EPON). First, we design both an algorithm and a rule for downstream packet scheduling at the inter- and intra-ONU levels, respectively, to reduce the downstream packet delay. After that, we propose a hybrid sleep mode that contains not only ONU deep sleep mode but also independent sleep modes for the transmitter and the receiver. This ensures that the energy consumed by the ONUs is minimal. To realize the hybrid sleep mode, a modified GATE control message is designed that involves 10 time points for sleep processes. In ESPS, the 10 time points are calculated according to the allocated bandwidths in both the upstream and the downstream. The simulation results show that ESPS outperforms traditional Upstream Centric Scheduling (UCS) scheme in terms of energy consumption and the average delay for both real-time and non-real-time packets downstream. The simulation results also show that the average energy consumption of each ONU in larger-sized networks is less than that in smaller-sized networks; hence, our ESPS is better suited for larger-sized networks.