Table of Contents:
: Spectrum Fragmentation Management Approaches Considering Defragmentation
In Chapter 7, we have already discussed fragmentation management approaches that consider non-defragmentation strategies. This chapter mainly focuses on spectrum fragmentation management approaches that consider defragmentation.
The defragmentation approaches are considered to fill-in the gaps left behind after terminating a lightpath. These approaches are typically classified into two main strands: reactive and proactive. Reactive defragmentation approaches are normally triggered when a new lightpath request arrives in the network. On the other hand, proactive defragmentation approaches are applied without waiting for a new lightpath request. Both proactive and reactive defragmentation approaches are again classified into two types, namely with and without rerouting of existing lightpaths. The rerouting approaches [67,103] reallocate existing lightpaths to the same or different spectrum slots by changing their routes in order to avoid the fragmentation effect. On the other hand, without rerouting, approaches do not allow existing lightpaths to change their routes; spectrum reallocation may be allowed. Based on traffic disruption, both with and without rerouting of existing lightpaths are categorized into the non-hitless and hitless defragmentation approaches, which are discussed in the following.
Non-hitless Defragmentation Approaches
The defragmentation approaches that cause traffic disruption are referred as non- hitless defragmentation approaches [67,103]. These approaches attempt to maximize the size of contiguous blocks of unassigned frequency resources with triggering traffic disturbance. As these approaches always cause traffic disruption, they are not preferred in EONs. To overcome this problem, hitless defragmentation approaches are considered for EONs, as is explained in the following.
Hitless Defragmentation Approaches
The defragmentation approaches that do not cause any traffic disruption are referred as hitless defragmentation approaches. These techniques [182-185] attempt to maximize the size of contiguous blocks of unassigned frequency resources without triggering any traffic disturbance.
Figure 8.1 shows an example of hitless defragmentation and its different conditions. Initially, lightpaths 1 to 4 are active in Fig. 8.1(a). Then lightpath 2 is terminated in Fig. 8.1(b), and we apply hitless defragmentation to retune lightpaths 3 and 4 in Fig. 8.1(c). Finally, lightpath 5 is added to the network in Fig. 8.1(d). In this example, lightpaths are retuned based on proper reconfiguration of allocated spectrum resources. The retuning is executed gradually and the spectrum jumped is not considered. The retuning operations are performed by all involved devices, including filters in intermediate nodes, in a coordinated manner under a distributed control environment or a centralized network controller. Therefore, changing lightpaths from one set of frequency slots to another does not cause any traffic interruption.
To achieve hitless defragmentation, a flexible optical node architecture [ 184] is essential. Figure 8.2 shows the node architecture that offers hitless defragmentation. It uses a pool of universal transceivers, instead of different types of dedicated transponders (see Fig. 8.3), to satisfy the clients’ demand. If dedicated transponders are used, flexibility is insufficient, and hence hitless defragmentation cannot be performed. This is because the synchronization among all involved devices can not be performed in a coordinated manner under distributed control environment or a centralized network controller. In this node architecture for hitless defragmentation, client-side devices no longer include the transceivers; all transceivers are placed in a universal transceiver pool. The client side generates a signal, which is mapped to transport frames, and the modulation format is decided. A bandwidth variable cross-connect switch (BV-WXC), which is placed between the client side and the universal transceivers pool, enables the sharing of transceivers from the universal pool. Using this architecture, the client selects
Figure 8.1: Example of different conditions of hitless defragmentation (a) initial state, (b) lightpath 2 terminated, (c) defragmentation using hitless, and (d) lightpath 5 added in network.
Figure 8.2: Node architecture for hitless defragmentation.
a suitably configured universal transceiver. For example, to support 300 Gb/s client demand, three transceivers are required, each supporting 100 Gb/s band-
Figure 8.3: Node architecture using dedicated transponders.
width demand. 50 Gb/s bandwidth demand is satisfied with one transceiver from the universal pool. 400 Gb/s bandwidth demand can be fulfilled by using two transceivers from the universal pool, each supporting 200 Gb/s. Finally, the optical signals from the universal transceiver pool are multiplexed and switched to the appropriate output fibers by a reconfigurable optical add-drop multiplexer (ROADM) that offers colorless, directionless, contentionless, and grid-less properties.
Figure 8.4 illustrates the BV-WXC control during the retuning process in order to move existing lightpaths from one set of frequency slots to another without causing any traffic disruption. There are two types of control, namely sequential BV-WXC control (see Fig. 8.4(a)) and synchronous BV-WXC control (see Fig. 8.4(b)). They are differentiated based on their approaches in the retuning process. Sequential BV-WXC control uses one large step retuning during which the spectrum between the source and destination the bandwidth are not available. On the other hand, synchronous BV-WXC control proceeds retuning by successive small steps and, after each step, available spectrum can be used. Sequential BV-WXC control is simpler and can be used when the retuning time is small compared to the average inter-arrival time of requests. However, if retuning time is not small enough, synchronous BV-WXC control may be preferable.
Figure 8.4: Illustration of BV-WXC control during the retuning (a) sequential BV-WXC control and (b) synchronous BV-WXC control.
In the literature, there are three main hitless defragmentation or retuning approaches; (i) hop-retuning , (ii) push-pull retuning , and (iii) Make- before-break. These hitless defragmentation approaches are discussed below.