Energy Efficient Dynamic Bandwidth Allocation with Two-Stage Control Mechanism to Satisfy Classified Delay Constraints in TDM-PONs

• Main Authors: Yun-CheChung, 鍾允哲
• Other Authors: Chuan-Ching Sue
• Format: Others
• Language: en_US
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/77921944830268710703

碩士 === 國立成功大學 === 資訊工程學系 === 102 === With the rapid growth of the Internet, web services like Voice over IP (VoIP), Video on Demand (VoD), Internet Protocol Television (IPTV) which requires low-latency and broadband network are arising. Therefore, energy consumption of network infrastructures like Access Network is dramatically increasing. The issue about saving energy on access network like Ethernet PON (EPON) and Gigabit-capable PON is considered a well-known problem. Many research proposed sleep mechanisms that concerned about single class’s delay constraint. While there are multiple classes of service exist on the network transmission, saving energy with high-priority class considered is less efficient, and high-priority class’s delay constraint won’t be satisfied while saving energy with low-priority class considered. The proposed scheme switch between Deep Sleep mechanism and Cyclic Sleep mechanism according to the number of packets which are going to violate their class’s delay constraint in a certain time. If an ONU has packets which are going to violate their delay constraint, the OLT commands ONU to enter sleep mode by using Cyclic Sleep mechanism while the ONU is not transmitting packets. If an ONU has no packet which are going to violate delay constraint in a certain time, the OLT commands the ONU to switch to Deep Sleep mechanism to sleep as long as it can. The probability of entering sleep mode under Deep Sleep increases while the proposed scheme let ONU enter sleep mode while there are packet in ONU’s upstream or downstream queue. The probability of exiting sleep mode under Deep Sleep decreases while the ONU decides to start transition from sleep to active mode if there are packets going to violate it’s delay constraint in a certain time. The simulation result shows the ONU of proposed scheme is able to enter sleep mode under higher network load and ensure the delay performances under the delay constraints while the previous schemes that only considered high-priority delay constraints could not. The proposed scheme is saving more energy while the ONUs stay in sleep mode for longer time.