On the benefits of differentiating the filter configurations in flexi-grid optical networks

Abstract

An effective strategy exploiting differentiated filter configurations in flexi-grid optical networks is proposed and evaluated. Results show that 15% throughput improvements can be achieved with respect to traditional flexi-grid approaches Introduction Flexi-grid technology has been recently introduced to increase the overall spectrum efficiency. However, the optical filters enabling flexi-grid present non-ideal shapes. This induces detrimental filtering cascade effects on traversed optical signals, thus limiting the overall transmission performance. To avoid these filtering effects, a larger amount of spectrum resources is then typically reserved. This way, the transmitted signals can avoid the transition region of the filters and they can safely operate on filter flat central regions. However, this typically implies a less efficient spectrum utilization. In , a strategy called super-filter (SF) has been introduced to improve spectrum efficiency by compacting spectrum-contiguous lightpaths. In this study, a further strategy is considered, called differentiated filter (DF). Moreover, the effective combination of these two strategies, named differentiated & super filter strategy (DSF), is here introduced. DSF is then implemented through an effective routing and spectrum assignment heuristic. The benefits of all these strategies are finally evaluated, showing remarkable improvements in the overall spectrum efficiency with respect to traditional flexi-grid configuration approaches. Traditional filter configuration In the traditional flexi-grid approach, the frequency slot is typically configured by assigning unique bandwidth value to all traversed optical nodes (i.e., filters) along the entire connection. For example, the GMPLS parameters n (central frequency) and m (slot width, expressed as number of 12.5 GHz frequency slices) are typically configured with the same value through all traversed nodes. Fig. 1 illustrates an example of connection transmission through a cascade of three nodes. First, it is assumed that m slices are assigned to the connection (Fig. 1a). It is also assumed that after NF=2 nodes, acceptable quality of transmission (QoT) is experienced. However, after the third traversed node, excessive detrimental filtering effects are experienced, preventing the actual setup along the nodes with such tight filtering. Thus, larger bandwidth should be computed and configured in order to avoid the filter transition bands and thus limiting detrimental filtering cascade effects. Fig. 1b shows the frequency slot configured, in all the three nodes, with one additional slice (m+1) reserved to all nodes. In this case, adequate QoT is achieved, at the expenses of more reserved spectrum in all traversed nodes. Differentiated filters and super-filters Fig. 1c shows the differentiated filter (DF) configuration strategy. In DF, different values of filter width (i.e., m) can be configured along the path in the nodes traversed by the same connection. In particular, the first nodes are configured with m slices, while m+1 slices are applied only to the last node. This way, the maximum number NF of acceptable tight filters is not exceeded. Thus, no additional detrimental filtering effects are introduced by the third node, and in particular by its (broader) filter, which is traversed by the lightpath in its flat region. Fig. 2 reports the concept of super filter introduced in. Signals can limit detrimental filtering cascade effects by sharing the central region of the filters with other spectrumcontiguous signals, also originated by different source-destination pairs. This way, the minimum amount of m slices per signal can be successfully allocated to the central lightpaths. Thus, the central lightpaths traverse a cascade of filters in their flat region, without experiencing significant transmission degradation. Joint differentiated and super filter strategy The overall filter strategy, called DSF, combines the benefits of the DF technique (which considers a single lightpath) with the SF technique (which considers multiple lightpaths). The objective is to take advantage of both the