Reconfigurable Grooming Of Dynamic Traffic In WDM Optical Networks

With the development of optical network and Wavelength Division Multiplexing (WDM) technology, traffic grooming became one of the most important problems in optical networks with high scientific and commercial values. Efficient grooming of traffic in optical network can not only effectively reduce the network cost, but also optimize its performances. So it has aroused much attention from many famous international research institutes since its appearance in 1998. At present, due to the NP-hard property of this problem, meta-heuristics will be the most powerful tools for tackling them.In order to satisfy the changeful traffic which exists commonly in current network, such as IP traffic, we proposed a new concept of traffic grooming, named as reconfigurable grooming under dynamic traffic, in this paper. We defined this new grooming type and classified it. What's more, we discussed in detail the properties, methods to solve it, and application range of it.Reconfigurable grooming of random traffic was studied profoundly in this paper. Two sets of integer non-linger program equations were proposed to describe Best Fit Case and Full Fit Case respectively. Based on this, two genetic algorithms with splitting traffic method were also developed to solve these two cases respectively.To testify the efficiency of the algorithms proposed, computer simulations were carried out. And then, we made detailed comparison and discussions on them and also analyzed the merit of reconfigurable grooming. The results shows algorithms designed in this paper yielded better results than others previously used in the literature.