Multicast traffic grooming in optical WDM networks
Date of Issue2013
School of Electrical and Electronic Engineering
In optical wavelength-division multiplexing (WDM) networks, multiple wavelengths with gigabit transmission rates may simultaneously transmit in a terabit/second capable fiber. WDM optical networks appear to be the most promising transport technology to facilitate and sustain the bandwidth intensive multicast applications, such as IPTV, video conferencing and streamed video broadcasts, which are expected to be the major drivers for Internet traffic growth. As the bandwidth requirements of the most multicast connections are largely less than the bandwidth of a wavelength, multicast traffic grooming which grooms (or multiplexes) multicast connections with low bandwidth requirements into high-capacity wavelength channels makes resources, like wavelengths, electronic switches, routers and multiplexers, highly utilized. Special considerations are required in multicast traffic grooming as a tree topology would be an optimal basic structure to follow for their implementation. This thesis investigates the algorithm design, optimal network design and performance analysis of optical WDM networks with multicast traffic under both static and dynamic traffic scenarios. Light-trees instead of lightpaths are used to support multicast connections because a light-tree offers one-to-many connection in one (electronic) logical hop. The sharing of light-trees is crucial to improve the utilization, but this may be considerably limited by the size of light-trees, because the probability of a large light-tree with many destinations to be shared by multiple multicast connections that have the same destinations of that light-tree is very low. To improve the light-tree utilization, we propose two multicast traffic grooming algorithms whereby light-trees are divided into small ones so that they can be efficiently utilized. When add/drop port resource is inadequate, the light-tree sharing becomes crucial for the performance of networks. To further improve the light-tree utilization, we propose multicast traffic grooming algorithms with leaking strategy, which allow a light-tree to deliver the traffic of a multicast connection to nodes which are not in the destination set of the connection. The results show that significant improvement can be achieved with the leaking strategy. The theoretical analysis of the multicast traffic grooming problem could guide the design of efficient and practical algorithms. In this study, we investigate the optimal design of network with multicast traffic grooming. A light-tree based Integer Linear Programming (ILP) formulation is proposed to minimize the network cost, and a heuristic algorithm is also proposed to achieve scalability and its performance is compared with the optimal solution. Due to the large numbers of variables and constraints caused by combinations of light-tree destinations, the light-tree based ILP model can only be used in small size networks. To improve the scalability of the light-tree based ILP analytical model, the hop constrained light-trees in multicast traffic grooming is introduced. An ILP formulation is proposed for optimal assignments of hop constrained light-trees for multicast connections, so that the network throughput can be maximized. A heuristic algorithm with a polynomial complexity is also proposed and its performance is compared with the ILP solution. Since using light-trees may lead to some serious negative side-effects because of light splitting, like non-uniform gain over the operating waveband, gain saturation and additional noise arising from intensive use of optical amplifiers, multicast traffic grooming in Tap-and-Continue (TaC) WDM networks is investigated. A network node with TaC devices can tap a small amount of incoming optical power for the local station while forwarding the remainder to an output port. We propose a simple and efficient node architecture with TaC mechanism. An ILP formulation with the objective of minimizing the network cost is developed and a heuristic algorithm of polynomial complexity is proposed for use in large networks.
DRNTU::Engineering::Electrical and electronic engineering