Energy-Saving Dynamic Wavelength Bandwidth Allocation Algorithms In TWDM-PON

Traditional Time Division Multiplexed-Passive Optical Network(TDM-PON)is increasingly unable to meet the growing bandwidth requirements of users.Time and Wavelength Division Multiplexed-PON(TWDM-PON),which combines the advantages of TDM-PON and Wavelength Division Multiplexed-PON(WDM-PON),has become the most competitive solution for Next-Generation PON(NG-PON).The explosive growth of users'bandwidth demand has led to a sharp increase of energy consumption in the Internet at the rate of 15-20%,while the energy consumption of access network accounts for about 70%.Therefore,energy-saving dynamic wavelength and Bandwidth Allocation(DWBA)for a TWDM-PON has become a hot topic in the industry and academia.First,the paper outlines the development of PON and the main solutions of NG-PON,discusses the structure,technical characteristics and resource management of TWDM-PON,and analyses the energy-saving mechanism and typical energy-saving algorithms of TWDM-PON.Secondly,modular design is introduced at both ends of optical line terminal(OLT)and optical network unit(ONU).Considering the interaction of the two energy-saving mechanisms in ONU and OLT,an Energy-Efficient Dynamic Wavelength and Bandwidth Allocation algorithm with Delay Constraint(EE-DWBA-DC)is proposed based on offline scheduling.In this algorithm,OLT dynamically adjusts the number of active wavelengths and the length of polling period according to the bandwidth requests and packet delay constraints of ONU in each polling cycle,so as to reduce the number of active wavelengths and to increase the length of polling cylcle as much as possible,and dynamically opens/closes the receiving/sending modules of ONU and OLT to save energy;at the same time,it authorizes the wavelength channel and transmission window of each ONU reasonably.In order to reduce the additional energy consumption for node state transition,the algorithm concentrates the working time of OLT and ONU as long as possible.In addition,the time when ONU and OLT are in low power consumption state is prolonged as long as possible,and the energy waste caused by high energy consumption gap is minimized.Therefore,this algorithm can reduce the energy consumption of OLT and ONU terminals and the total energy consumption of the whole network under the premise of packet delay constraints.In addition,based on hybrid scheduling and modular design of nodes,with a joint consideration of OLT and ONU,a Dynamic Energy-Efficient Wavelength Bandwidth Allocation algorithm(EE-DWBA-DS)is proposed to support differentiated services.The algorithm divides each polling cycle into two sub-cycles:on-line and off-line.In the first sub-cycle,Expedited Forwarding(EF)traffic is authorized by online scheduling,while in the second sub-cycle,non-delay-sensitive traffic of Assured Forwarding(AF)/Best Effort(BE)is authorized by offline scheduling.In each sub-cycle,the number of active wavelengths is adjusted reasonably according to the Quality of Service(QoS)requirements of different types of traffic in ONUs.Under the premise of guaranteeing the bandwidth and delay constraints of EF traffic and the minimum bandwidth requirement of AF traffic while satisfying as much as possible the bandwidth requirements of AF/BE traffic,energy-saving can be achieved by minimizing the number of active wavelengths and closing the idle wavelength and its corresponding receiving and sending modules and reasonably allocating the wavelength channel and transmission window of each ONU to extend the low-power duration of nodes,reduce the number of state transitions at nodes and the high energy consumption voids of channels.Therefore,it can reduce the energy consumption of ONU and OLT and the network's total energy consumption while guaranteeing the specific constraints of different types of traffic.Finally,the EE-DWBA-DC and EE-DWBA-DS algorithms are simulated by OPNET network simulation platform.The two algorithms are compared with several typical algorithms in literature.The simulation results verify the effectiveness of the proposed algorithms.