A novel fault-localisation protocol is constructed based on the principles of distributed control and management mechanisms. The proposed protocol has high scalability and speed, but at the cost of increased computational complexity. To provide the maximum level of transparency, the protocol skips any optical power monitoring or spectrum analysis at the intermediate nodes of established lightpaths. Moreover, to narrow down the associated time and space complexities, it restricts the fault-localisation area to a small area called limited-perimeter. These functions are implemented by means of five phases, namely pausing, flooding, multicasting, matching, and concluding. Although the protocol has been developed to pinpoint single failures, it could track down multiple failures that occur in nonoverlapped limited-perimeters. To evaluate the performance of the proposed method, time and space complexities are calculated and compared with a counterpart protocol that does not limit the fault-localisation area.

Applications and users require different recovery services. Some applications are delay sensitive and demand fast restoration, while others are data sensitive and require 100% data recovery. Dynamic restoration techniques, on the other hand, have exclusive restoration abilities in terms of latency, restoration throughput, and complexity. It is therefore important to comprehend the diversity among dynamic restoration techniques in order to effectively mesh their aptitudes to applications requirements. Accordingly, in this paper we evaluate distinct types of dynamic restoration approaches, namely disjoint path, partially joint path, and link restoration to identify their specialty and deficiency in terms of blocking probability, restoration delay, and computational complexity. Our results reveal that partially joint path restoration always obtains the lowest blocking probability. Nevertheless, this scheme has the highest computational complexity. Link restoration is the fastest if fast fault localization is achieved ahead of the restoration process. Conversely, disjoint path restoration always has the highest average restoration delay. However, this approach is the simplest and obtains the lowest computational complexity.

This work proposes a new reservation protocol for enhancing the performance of wavelength-routed networks. To be more robust and reliable, the proposed approach employs distributed control mechanisms. The new method particularly focuses on wavelength-division multiplexed (WDM) core networks with distant end-nodes. It takes into account the considerable amount of data that can be transferred by high-speed WDM networks within limited reservation periods. To increase the throughput, the protocol consumes the unoccupied bandwidth of reservation phases by transferring nonreal-time data packets during these intervals. This scheme is implemented by applying a modified form of backward reservation protocol. To initiate a multihop reservation call, this protocol labels a path as reserved instead of locking it. Meanwhile, labeled nodes with single-hop requests will receive permission signals to send predetermined packet sizes. The length of packets transmitted is defined by the round-trip propagation delay between the current and the upcoming nodes along the path. In case a reservation fails, already labeled nodes will be notified by receiving a prevention signal, which will block them from transferring data packets.

In this paper, we propose and describe a new distributed reservation protocol for establishing lightpaths in WDM all-optical networks. Distributed control mechanisms are preferred and employed because of their advantages over centralized ones to set up virtual channels. The new protocol is a combination of the conservative and aggressive backward reservation protocols, which attempts to improve performance by adapting a reservation to network circumstances. On the one hand, the new protocol uses network circumstances and decides and applies a more conservative or aggressive approach. In other words, the protocol progressively fluctuates between those reservation protocols in order to capture their respective advantages. As a result, in extreme cases it acts exactly like either the conservative or the the aggressive reservation protocol. On the other hand, it considers the characteristics of a network to set a retry-list size. As a result, a retry-list size is not determined by a fixed number but modified based on the multiplexing degree of a network, which prevents imposing ineffective retries on a network with a small number of wavelengths and instead encourages more retries for a network with numerous channels. Therefore, the proposed protocol transforms the static nature of existing reservation protocols into a more adaptive one in order to enhance network performance.