Optimal Design And Methods For Optical Networks

The 21 st century is the Internet era.Big data,cloud computing,and network video services have brought a huge traffic demand and pressed a great pressure on network operators.Thus,it is important to develop efficient network design approaches for improving network resource utilization.The research in this thesis is dedicated to the optimal design and method for the optical network.We have carried out this optimizaiton at different levels,ranging from nodes,end-to-end optical channels,and the whole network.The main contents and contributions are as follows:Firstly,at the node level,we developed analytical models for the estimation of lightpath blocking probability,and evaluated the impacts of the colorless,directionless,and contentionless(CDC)features of reconfigurable optical add/drop multiplexers(ROADMs)on the wavelength division multiplexing(WDM)optical network.To improve the network resource utilization,we proposed three lightpath routing and add/drop port selection strategies.Simulation results indicated that the analytical models can effectively predict the performance of the optical transport networks under different add/drop capabilities.The simulation results also showed that among the three features,the colorless feature plays the most important role in improving lightpath blocking performance.Under the colorless feature,a saturation phenomenon between the lightpath blocking performance versus the number of add/drop ports was also observed.When the number of add/drop ports per nodal degree reaches a certain threshold level,performance improvement will be marginal with a further increasing number of add/drop ports.Finally,comparing the three adaptive lightpath routing and add/drop allocation strategies,we found that the awareness of the add/drop port state information when routing and assigning wavelengths for lightpaths can significantly improve the lightpath blocking performance.Secondly,at the optical channel level,we proposed a new adaptive Forward Error Correction(FEC)assignment strategy.A typical FEC assignment strategy is to assign the same type of FEC for all the lightpaths in a network,which is however wasteful for many lightpaths with good OSNRs.For better efficiency,they can be assigned with lower levels of FECs for less overhead.Thus,we proposed a new adaptive FEC assignment strategy,which assigns the most efficient FEC type for each optical channel according to their actual OSNRs.In the context of the Routing and Wavelength Assignment(RWA)problem of the WDM network,we evaluated the performance of the proposed adaptive FEC assignment strategy.We developed an Integer Linear Programming(ILP)model and a simple waveplane-based heuristic algorithm for the RWA problem.The simulation results indicated that compared to the non-adaptive strategy,the proposed adaptive FEC assignment strategy can significantly reduce the required FEC overhead.In addition,the proposed heuristic algorithm performs almost as well as the ILP model.In addition to the WDM network,we also applied the strategy to the Elastic Optical Network(EON).Similarly,for the Routing and Spectrum Assignment(RSA)problem,we developed an ILP model and a spectrum window plane(SWP)-based heuristic algorithm.Results show that the adaptive FEC assignment strategy can significantly outperform the non-adaptive strategy for maximizing the overall transmission capacity and minimizing the required FEC overhead per lightpath,and moreover,the proposed heuristic algorithm is efficient to perform close to the ILP model.Lastly,at the network level,we developed a novel Load-Balanced Fixed Routing(LBFR)algorithm.Fixed shortest path routing is one of major lightpath service provisioning strategies due to its operational simplicity.However,it is well known that this approach suffers from the disadvantage of unbalanced network load and severe congestion on some bottleneck links,which can significantly degrade the throughput of the whole network.In this thesis,in order inherit the operational simplicity and meanwhile overcome the disadvantage of unbalanced network load of the fixed shortest path routing approach,we proposed a new LBFR algorithm which can greatly alleviate network congestion and at the same time retain the operational simplicity.Through a training process based on a forecasted network traffic load matrix,the LBFR algorithm finds a(or few)fixed route(s)between each node pair and employs the fixed route(s)for lightpath service provisioning.Also,to evaluate the performance of the LBFR algorithm,we developed a new analytical model considering multiple routes between each node pair.Simulation studies showed that the multi-route analytical model is effective to accurately predict the blocking performance.Also,compared to the traditional fixed shortest path routing algorithm,the LBFR algorithm can achieve much better lightpath blocking performance,and such a performance improvement becomes more significant with the increase of network nodal degree.