Master Thesis - 2014

High speed and large capacity backbone network will be required for future high-quality multimedia services. For future networks, all-optical networks with Wavelength Division Multiplexing have been expected as one of promising solutions because of their rapid and wide band characteristics. The elastic optical network, which is a new network architecture developed from the WDM network, are intensively researched because it can flexibly exploit spectrum resources by selecting modulation format based on both demand size and optical reach.

It is difficult for the typical and traditional optical node architecture to selectively supply diverse functions such as time, frequency and spatial multiplexing and optical regenerations. It is important for optical nodes to dynamically reconfigure their signal processing modules in order to accommodate various traffic demands. Architecture on Demand (AoD) nodes are new optical switching nodes which interconnect between input/output ports and signal processing modules by large port-count optical switches. Their interconnections can be dynamically and flexibly changed in order to supply just enough signal processing according to the traffic requests. On the other hand, large scale AoD nodes require large port-count optical switches beyond a limit of a single optical switch. Furthermore,immature resource allocation waste switch configurability and capability from network perspective, Therefore, there is a requirement for an appropriate and practical AoD node architecture which use multiple optical switches considering the increase of power consumption.

In this thesis, firstly, the actual AoD node architecture which reduces its power consumption is proposed. Secondly, a spectrum resource allocation method which effectively tilizes the proposed AoD node architecture is also proposed. The proposed resource allocation method can improve fragmentation in spectrum domains classifying connection requests based on the required bandwidth. Moreover, in the proposed AoD node architecture, these classified resource allocation can replace flexible and expensive signal processing modules with fixed and cheap ones.

Finally, effectiveness of the proposed method is confirmed through computer simulations. In particular, the proposed method can reduce the power consumption of optical nodes more than 80%. Furthermore, the proposed method can partly improve call-blocking probability.