Research On Key Technologies In Flexible Optical Networks

In recent years,Internet Protocol Television,Video On Demand,cloud computing,big data as well as other new services take on an exponential growth.The traditional Wavelength Division Multiplexing(WDM)network adopts the one size fits all mode in performance management and wavelength allocation,which leads to the low flexibility and inefficient resource utilization,and it is not suitable for future high speed,large capacity and scalable optical transport networks.Based on the Orthogonal Frequency Division Multiplexing(OFDM)technology,the Elastic Optical Network(EON)applies to the spectrum grid with a smaller granularity,and according to the requirement of traffic for the demand of users and considering the length and signal quality of the physical transmission path,it makes a dynamic allocation of the spectrum,sets the corresponding modulation formats,thus greatly improving the spectrum utilization.The main work and innovations of the dissertation are as follows:First of all,in terms of the static Routing and Spectrum Assignment(RSA)problem in the EON,the model of Integer Linear Programming(ILP)is presented,and the Co-Evolution with Grooming algorithms based on Distance-Adaptation(DACE-Gx)are proposed,DACE-Gx algorithms combining the heuristic algorithms with the co-evolution algorithm,and is applied to search the approximately optimal solution through iteration.DACE-Gx algorithms use the grooming strategy and distance adaptation strategy,can improve the bandwidth utilization and reduce the consumption of spectrum resources.The co-evolution in GACE-Gx can provide global search capability,and the solutions for heuristic are used as the better initial solutions for co-evolution,which ensures the directionality for search.The simulation results show that compared with the corresponding heuristic algorithms,the proposed GACE-Gx algorithms have reduced the maximum sequence number of the frequency slots and enhanced the frequency slots utilization ratio.The traditional Poisson model to generate the voice service only has a short-term correlation,so it is unable to describe the real network traffic accurately.The research shows that,whatever changes in the number of network users,topological structure and service types,the actual network traffic is always reflected as self-similarity.Based on the study on the burst of self-similar traffic,the RSA algorithms Based On Threshold(BOT) are proposed.BOT algorithms first set a threshold,the network is not considered in a state of burst when the threshold is not triggered,and the fixed routing algorithm should be selected,making the least cost of spectrum resources and save the computation time;Once the threshold is triggered,the network is considered to be in a state of burst,the adaptive routing algorithm is adopted for routing and spectrum allocation,it is able to make full use of idle spectrum resources.The experimental results show that the BOT algorithm can reduce the blocking probability compared with the existing heuristic algorithms.Along with the arrival and departure of dynamic traffic,the spectrum is divided into many fragments,and a mass of fragments will increase the blocking probability and decline the spectrum utilization.Firstly we give the concepts of channel density and minimum gap,which can reflect the tightness of the connection requests in the spectrum.The larger the channel density and the smaller the minimum gap,the tighter the arrangement of connection requests will be,and conversely,the looser the arrangement of connection requests will be.Based on these two concepts,the algorithms of Periodic Aggregation Defragmentation(PAD)and Triggered Aggregation Defragmentation(TAD)are proposed,and reroute and/or reallocate the spectrum for the connection requests with the minimum channel density and the maximum minimum gaps,rearrange the loosely distributed connection requests into a closed state and squeeze out some space for the following connection requests.The experimental results show that PAD and TAD can reduce the blocking probability and improve the spectrum utilization.