6.1 Summary of the Dissertation
It is believed that the WSON technology is moving toward being largely adopted in broadband networks. However, due to the immature all-optical technology, the O/E/O 3R regenerators and wavelength converters are employed in current WSONs in order to compensate for the loss of signal quality and to mitigate the wavelength contention. In many cases, these functions can be technically combined into a single component under O/E/O conversion, namely WC3R. This multi-function device can be more efficient in terms of resource utilization. However, the high cost and electronic bottleneck prohibit the full deployment of WC3Rs in the network. This leads to the existence of translucent WSON which compromises between the capital expenditure and network performance. This dissertation mainly investigates the design and control issues in translucent WSON, and there are two-fold research objectives.
For efficient translucent WSON design, I focus on studying the WC3R placement issue in Chapter 3. The key task is to optimally distribute the WC3Rs in the network, i.e., minimize the opacity of the network without significantly degrading the network performance.
Because WC3R placement is a NP-hard problem, my proposal resorts to the heuristics. For simplicity, it is solved in two steps: node ranking and selection. Firstly, the nodes are ranked based on two types of knowledge about the future network. Without specific knowledge such as traffic pattern, PLI except its topology, centrality-based nodal weight is conceived, i.e., degree centrality, closeness centrality, and betweenness centrality, while if more information is available, the conversion and regeneration prediction algorithm is developed. Then, based
on the node weights, the WC3R capable nodes are selected with a certain non-adjacent degree.
To evaluate the efficiency in lowering the blocking probability under different placement strategies, a novel adaptive routing and joint resource assignment algorithm is proposed to provision the lightpaths in WSON. Moreover, a simulation-based study on even and uneven distribution of WC3Rs with the fixed number of such devices is also conducted in this chapter.
The result reveals that uneven placement can achieve a better performance with the same amount of WC3Rs. Lastly, I go deep into investigating the WSON from the perspective of cost and benefit by employing different types of WC3Rs in order to find the possibility of more efficient WC3R investment.
The second aim of this dissertation is how to optimally use the resource in translucent WSON in order to provide the utmost performance in lightpath provisioning in the distributed manner. RRA, lying in the control plane, contributes the major concern in this research. I address this topic in two cases, i.e., with/without wavelength conversion in Chapter 4 and Chapter 5 respectively. The difficulty to perform the RRA in distributed control plane is that the effectiveness of RRA is dependent on the accuracy of network state information, which is hardly guaranteed in large-scale networks. To tackle this weakness, a multi-solution based approach is proposed in Chapter 4, where the set of wavelengths with the candidate path can form multiple solutions for a lightpath request. Therefore, the key technology of this approach is the selection strategy of probing wavelength set. Here, my proposals of TRWS and TRWS&MC are compared with traditional approaches to show its trade-off performance between the blocking probability and the set-up delay. In Chapter 5, the same topic is studied in a wavelength convertible WSON, where the multi-solution approaches should be carried out within each wavelength segment. Similar simulation comparison is conducted.
The simulation results reveal that by employing the proposed scheme, without great loss on the blocking performance, the inaccuracy of the network state information can be well tolerated, and the set-up delay in lightpath provisioning can be kept at a low level.
6.2 Future Research Direction
The adaptive routing is applied throughout this dissertation, which is acknowledged to have the best performance comparing to the fixed and alternative algorithms, but its effectiveness is sensitive to the accuracy of network state information. To mitigate this inaccuracy, I can also take advantage of the two-way signalling process to collect link/node state information along the lightpath which is being set up or torn down, and then to update this new state information at the source node. And moreover, optimal utilization of WC3Rs along a path can be regarded as another way to get the performance gain in translucent WSONs.
Therefore, these are expected to bring further performance improvement in lightpath provisioning and become my recent study in RRA algorithm design within hybrid distributed
control scheme in wavelength convertible translucent WSON.
In my current research, it is assumed by default that the client traffic has the same granularity and takes up the entire wavelength bandwidth, and each wavelength has the same capacity. But the fact may not be like that, so there are several open issues in the control of mixed line rate system.
In addition, the recovery is a big issue in control of WSON. How to automatically reconfigure the network to detour the failures caused by accident or disaster is also an attractive and hot topic for my future research.