Spectrum Defragmentation And Assignment Algorithms In Flexible Bandwidth Optical Networks

With the rapid development of Internet, the demand of bandwidth rises sharply. It takes many opportunities and challenges to the optical network. Traditional wavelength-division multiplexing (WDM) network has a significant drawback due to rigid grid and coarse bandwidth granularity. In this case, a spectrum-efficient and scalable optical network architecture called flexible bandwidth optical networks, which allocating spectrums with speeds ranging from Gb/s up to Tb/s based on the demands and reach requirements, was proposed as a promising technology for future transmission networks.In the flexible grid networks, with new request arriving, we should establish a connection and assign proper spectral resources for it. When request ends, the connection is released, the spectral resources could be assigned for new requests. In a dynamic case, this channel setup and tear down processes lead to fragmentation of spectrum resources. Note that the probability of using these pieces of fragmentations is low since they are not consecutive, spectral defragmentation has been proposed to solve this problem.The interruption of the high-priority requests weakens the network transmission performance. Currently, First Fit algorithm (F-F) is the most common method to assign spectrum; it pack all of the in-use slots toward the lower end of the spectrum space and it is easy to implement. However, because of the inevitable large capacity trend of network, channel setup and tear down processes of connections lead to more spectrum fragmentation necessarily, the significant effect of using F-F to assign spectrum is uncertain.In order to solve the problems above, this article carried on the thorough research, and put forward the effective solution. The main contributions of this article are listed as follows:Firstly, a SDN network architecture with the characteristic of control logic centralization, open network capability, separation of hardware and software, general hardware architecture, automation network customization and so on is proposed in the article. A extensible ASON simulation experiment platform is also introduced.Secondly, the article presents a novel priority-oriented defragmentation algorithm for flexible grid networks to reduce interruption of high priority services. According to this theme, the transmission of high priority services remain unchanged when operating defragmentation, only low priority services are interrupted, which is good for maintaining the network performance. Then we proposed a scheme that for the services of low priority, firstly sorting the services in accordance with the order of the size of the remaining time of the requests, and then operating defragmentation. we can conclude that both Blocked Services Number and Defragmentation Operation Number decrease as the proportion of low priority services increases and Priority-oriented defragmentation outperforms other defragmentation scheme in terms of blocking rate with different average holding time of the services. Numerical experiments results show that our algorithm can further reduce blocking rate and defragmentation operation number as well as maintain the network performance.Thirdly, the article proposes two novel hierarchical spectrum assignment (HSA) algorithms, including HSA with large section segment (HSA-L) and HSA without large section segment (HSA-NL), taking into consideration both overall and part of spectral resources to achieve the most concentrated spectral resources after one service resources allocated. Numerical simulations show that HSA-L algorithm proved more effective compared with F-F especially with the large capacity trend of network. The values of reduction ratio of blocking probability of HAS-NL and HSA-L compared with F-F are both in upward trend with average duration and traffic loads increasing. we can figure out that HSA-L (hierarchical spectrum assignment algorithm with large section segment) algorithm has a better performance trend in the term of reduction ratio of blocking probability than F-F at the same parameters with the increase of max demand bandwidth (ranges from1slots to8and the threshold M set to be7) of requests and spectrum slots of each link. Numerical simulations show that HSA-L algorithm proved more effective compared with F-F especially with the large capacity tread of network.