Research On Energy-Saving Mechanisms And Algorithms In Ethernet Passive Optical Networks

With the intensifying global warming, it is imperative to reduce energy and carbon emissions. As one of the most effective "last-mile" solutions, broadband access network contributes to about 75% energy consumption of the whole network. Passive Optical Network (PON) with high bandwidth and high stability has become the most feasible and most energy-efficient broadband access solution due to the passive feature of the remote nodes. As a ripe PON technology, Ethernet PON (EPON) has become one of the most widely deployed networks in recent years. With the increase of the users and traffic demands, the energy consumption of PON also increases rapidly. Therefore, how to effectively reduce the energy consumption of PON, especially of EPON, has become one of the hot topics with the significant prospects of academics and industrialization.Currently, most energy-saving studies on EPON adopt the method that sleeps the low-loaded Optical Network Unit (ONU), but always sleeps the whole ONU, i.e., the transmitter and receiver of ONU should be wakened up or turned off at the same time. This method is not flexible enough, which not only saves less energy but also decreases the Quality of Service (QoS). Furthermore, the existing energy-saving algorithms almost focus on EPON that only supports unicast traffic, so the energy-saving problem of EPON with multicast traffic can not be solved effectively. Therefore, energy-saving problem of EPON is still with large research space to study, and is one of the most significant challenges of building green network.Based on the conclusion and analysis about the domestic and overseas researches on energy conservation of PON, this thesis proposes novel ONU sleep mode, and combines the network features with the novel ONU sleep mode under different network environments to effectively reduce the energy consumption of ONU, thus promoting energy conservation of EPON. The main contributions and innovations in this thesis are summarized as follows:(1) Aimed at energy-saving problem of 1G/10G-EPON, Chapter 2 first proposes an energy-saving mechanism based on sleep and dynamic double-threshold downstream rate selection. In this mechanism, ONU can go to deep sleep state after upstream transmission and its sleep time can change with upstream traffic load, so that the energy consumption of ONU can be decreased from the perspective of sleeping ONU. Then, this mechanism uses low-energy link under low load according to dynamic double-threshold downstream rate selection algorithm, thus decreasing energy consumption from the perspective of downstream rate selection. Based on this mechanism, the thesis proposes an energy-saving mechanism based on double-sleep-state and dynamic double-threshold downstream rate selection by allowing the idle receiver to sleep during the upstream transmission slot so that the energy consumption of ONU can reduce further.(2) Aimed at the unequivalence of upstream and downstream traffic, Chapter 3 proposes an energy-saving mechanism based on downstream scheduling and hybrid sleep. This mechanism allows the independent sleep mode of the transmitter and receiver of ONU, to effectively reduce the energy waste when the upstream traffic and downstream traffic do not exist simultaneously. This mechanism also combines the independent sleep mode with ONU deep sleep mode to form the hybrid sleep mode. By reasonable downstream scheduling, the proposed mechanism not only can promote energy conservation but also can reduce the downstream packet delay on the other hand.(3) Chapter 4 first proposes a hybrid sleep based energy-saving mechanism for multicast traffic. The characteristic of multicast traffic is that, ONUs with the same multicast traffic should receive them simultaneously but can not send upstream traffic simultaneously. According to the characteristic, the proposed mechanism introduces hybrid sleep mode, allowing the transmitter and receiver to sleep independently and to combine with ONU deep sleep mode, so that the unnecessary energy waste can be reduced. By ONU deep sleep and wakeup algorithm, the proposed mechanism can decrease the wakeup times of ONU to reduce the energy consumption of ONU. Based on this mechanism, hybrid sleep based improved energy-saving mechanism for multicast traffic is proposed. The second mechanism adds an additional constraint for ONU wakeup by ONU deep sleep and wakeup optimization algorithm. Therefore, it can reduce the energy consumption of ONU further without sacrificing QoS.To determine and evaluate the performance of proposed mechanisms, this thesis builds a simulation platform by using OPNET simulation software. Simulation results demonstrate that the proposed mechanisms can save more energy than the traditional energy saving mechanism while keeping the performances, such as traffic delay and throughput, at an acceptable level. The research achievements in this thesis are able to provide the theoretical reference and technical support to the development and construction of future passive optical network.