Research On Dynamic Restoration Methods In WDM Networks With Traffic Grooming Capacity

With the development of social economy, many kinds of data business and multimedia real-time business are constantly emerging, and people are increasingly dependent on network. The result is that greater bandwidth and higher communication capacity are requested for the network. Wavelength division multiplexing (WDM), which divides the bandwidth of a fiber into many disjoint wavelengths, greatly improves the transmission bandwidth of the network. To improve resource utilization and reduce the cost of optical network, an efficient approach is to pack low-speed traffic connection stream onto high-speed wavelength channels, which is referred to as traffic grooming. As the capacity of WDM network with trafficgrooming is huge, network failures such as link fiber fractures will cause catastrophic consequences, a severe data loss and business interruption. Therefore, the survivable traffic grooming problem has been in a prominent position.This study addresses the survivability problem in WDM networks with traffic grooming capacity employing dynamic restoration methods. Two restoration schemes are proposed, namely connection dynamic restoration scheme (CDRS) and lightpath dynamic restoration scheme (LDRS). CDRS is a fine-granularity restoration scheme and operates at the low-rate connection level while LDRS is a coarse-granularity restoration scheme and operates at the high-rate lightpath level. A complete dynamic restoration scheme is composed of traffic grooming algorithm and restoration protocol. Firstly, a heuristic rerouting algorithm based on link bound auxiliary graph (LBAG) is proposed to compute a route for a requested or failed connection, which is a multihop dynamic grooming algorithm and is called integrated grooming based on LBAG (IGL). Secondly, this thesis presents two kinds of restoration protocols, connection restoration protocol (CRP) for CDRS and lightpath restoration protocol (LRP) for LDRS. Finally, on this basis, the analytical expressions for the restoration switching times of the two restoration schemes are presented.Simulation results show that since CDRS provides more flexibility, it has a better performance of restoration success rate and resource utilization ratio than LDRS, and in terms of restoration switching time, CDRS also achieves a shorter performance. But LDRS meets the requirement of large granularity recovery and has better scalability. Therefor, it is necessary to adopt appropriate dynamic restoration method in the different appliacations. Traffic grooming technology helps to improve the restoration success rate as well as to reduce restoration switching time. Compared to traditional DPR, the results show that performance of CDRS is fully superior to the DPR. Compared to MPAC, CDRS has a better resource utilization ratio performance and achieves a high restoration success rate but less than MPAC, and the restoration switching times of the two algorithms are very close. Above all, CDRS can achieve a better balance between restoration success rate and resource utilization ratio. By setting the appropriate configuration of network resource, when the network load is not big, CDRS can achieve ideal restoration success rate as well as a high resource utilization ratio.