Correlation between traffic granularity and defragmentation periodicity in elastic optical networks

While elastic optical network technologies have emerged as promising solutions for future ultra-high speed optical transmission, the unavoidable spectral fragmentation problem that appears in such networks significantly degrades their performance. In light of this, spectral defragmentation technologies have been introduced in elastic optical networks, aiming to increase the spectrum utilization. During the defragmentation operation, the available fragmented spectrum bands are consolidated by re-allocating existing connections, either re-routing them along alternative routes and/or retuning them onto different spectrum portions. Obviously, spectral defragmentation increases network complexity and cost. Therefore, it is highly desirable to limit its application as much as possible, while keeping network performance within acceptable margins. In this paper, we focus on analyzing the correlation between the optimal (i.e., minimum) spectrum defragmentation periodicity in the network with the granularity of the supported traffic. For this purpose, we initially introduce a novel algorithm for efficient spectrum defragmentation. The proposed algorithm aims to consolidate the available fiber spectrum as much as possible, while minimizing the number of disrupted active connections. Then, supported on extensive simulation results, we show how spectral defragmentation periodicity can be effectively configured by having knowledge of the offered traffic granularity. Copyright c © 2013 John Wiley & Sons, Ltd.