Diane Prisca Onguetou

A new approach to node-failure protection with span-protecting p-cycles

Recent work has revealed a new, relatively simple and possibly cost-effective, approach to achieve combined protection of optical networks against both node and span failures. The resulting network designs use only a single set of p-cycle structures that have the same or only slightly more capacity than a corresponding optimal set of p-cycles for span protection. The new principle is based on a generalization of how nodes in a BLSR-ring or p-cycle (to date) derive survivability through loop-back at the nearest two neighbour-nodes on the same ring. The generalization views any combination of node failure and an affected transiting path from the standpoint of the two-hop segment defined by the failure node, and the nodes immediately adjacent on the affected path. We then ask whether these nodes are found together within the same p-cycle as the failure node, or another p-cycle entirely. In any case where they are, we show that the transiting path affected by the node failure is inherently restorable by ordinary p-cycle switching actions whether the respective two-hop segment is on-cycle, straddling, or partially on-cycle and partially straddling. We explain the principle and characterize its effectiveness in terms of network-wide single node failure restorability (R1-node) in networks designed only for minimum spare capacity, networks designed for enhanced R1-node (at min capacity) and networks designed strictly for R1-node = 1.

Approaches to p-cycle network design with controlled optical path lengths in the restored network state

In a transparent optical network it is desirable to have design control over the length of normal working paths and over the end-to-end length of paths in any restored network state. An obvious approach with p-cycles is to limit the maximum allowable circumference of candidate cycles considered in the network design. But this is somewhat inefficient and does not directly control the end-to-end length of paths in a restored state; it only controls the maximum length of protection path-segments that might be substituted into a working path on failure. Another basic strategy is now considered. It consists of systematically matching shorter working paths with longer protection path-segments through p-cycles, and vice versa, with direct consideration of the end-to-end length of paths in the restored network state during the design. This complementary matching notion is studied through an integer linear programming (ILP) model to minimize cost while intelligently associating longer working paths with shorter protection path-segments and vice versa. The basic ILP is adapted in one case to minimize the average restored state path lengths; in another to achieve the least possible longest path length; and, finally, to also constrain all restored path lengths under a fixed limit. Each variation can also be subject to a requirement of using only the theoretically minimal spare capacity or, through bi-criteria methods, a minimal amount of additional spare capacity for the corresponding objective on path lengths. Taken overall the work provides the means to design an entire transparent survivable island that respects the transparent reach limits of a given ultra-long-haul technology. A heuristic combination of ILP and genetic algorithm methods is also developed to solve some of the larger problems and is shown to perform well.

Comparative study of fully pre-cross-connected protection architectures for transparent optical networks

Protection architectures that have the property of pre-cross-connection are advantageous to the implementation of transparent optical paths. A pre-cross-connected protection path can be in a known-good working condition before use, whereas on-the-fly assembly of a protection path through transparent concatenation of optical channels may not rapidly satisfy the optical path integrity objective. In this study, several pre-cross-connected architectures were compared on the basis of spare capacity cost for 100% single failure restorability, the dual failure restorability of these designs, and the ability of each architecture to support a feasible wavelength assignment and limits on the maximum length of transparent optical paths. The architectures considered were p-cycles, failure independent path-protecting (FIPP) p-cycles, demand-wise shared protection (DSP), pre-cross-connected trails (PXTs), span-protecting p-trees, and path-protecting p-trees. The results give insight into the relative merits and demerits of these architectures.

p-cycle network design: From fewest in number to smallest in size

An idea seems to have spread that p-cycle networks are always based on a single Hamiltonian cycle. The correct understanding is that while they can be based on a Hamiltonian, network designs involving multiple p-cycles are far more capacity-efficient in general. In fact, from an optical networking standpoint one would probably like to work with p-cycles of the smallest size (circumference) possible, to satisfy optical reach considerations, and in this case the number of p-cycles might be even more numerous than a pure minimum capacity design. However, the fact that an entire network could be protected by a single cyclic structure could be attractive from another viewpoint simply because only one logical structure has to be managed. Thus, different recent orientations have brought us to realize the need for a study of p-cycle network designs that vary systematically across the range between the smallest size p-cycles, to using the fewest number of p-cycles. Questions include: What are the design models for p-cycle networks that use the fewest number of distinct structures? What are the capacity implications of a design restricted to a specific maximum number of structures? Can a capacity-optimal design be ¿nudged¿ into using fewer structures in total without requiring any extra capacity? What happens to the number of structures if the smallest possible p-cycles are insisted upon? Accordingly, we offer a systematic study of the optimal p-cycle network design problem addressing such questions about how the logical number of p-cycle structures present or allowed in a design interacts with the minimum spare capacity required for the design to be 100% restorable.

Near-optimal FIPP p-cycle network designs using general path-protecting p-cycles and combined GA-ILP methods

Recent work on failure independent path-protecting p-cycles (FIPP) has revealed some new, relatively simple and possibly cost-effective approaches for FIPP p-cycle network design. The first step of the proposed strategy consists of solving a more general path-protecting p-cycle (GPP) problem in which the constraint of failure independence is relaxed. The second step consists of imposing the failure independence constraint onto the GPP solution and identifying the working paths that become unprotected as a result. A FIPP p-cycle solution is extracted by capacitating additional cycles to protect these paths, the number of which the results revealed to never exceed three. Another contribution of this work is the adaptation of the novel combination of genetic algorithms with integer linear programming (GA-ILP) to the GPP concept, which allowed us to solve large GPP problem instances. GA-ILP solutions were typically within 1% of optimality for smaller networks for which the exact solutions were known. The GPP and FIPP solutions obtained with the assistance of GA-ILP were considerably better (by as much as 23%) than those obtained by the FIPP disjoint route set (DRS) method. Furthermore, the results obtained in this paper also showed that relaxing the disjoint route set constraint in FIPP p-cycle networks can result in as much as 9% decrease in spare capacity cost. Also in this paper, we ventured to provide a true comparison of span-protecting p-cycles with FIPP p-cycles, from the capacity efficiency perspective.

p-Cycle protection at the glass fiber level

The cost and complexity of wavelength assignment, wavelength conversion and wavelength-selective switching are always of primary considerations in the design of survivable optical networks. This proposal recognizes that as long as the loss budgets are adequate, entire DWDM wavebands could be restored with no switching or manipulation of individual lightpaths; so that the DWDM layer would never know the break happened. And environments where fiber switching devices are low cost, and ducts are full of dark fibers provide a very low cost alternative to protect an entire DWDM transport layer (or working capacity envelope) against the single largest cause of outage. Yet, while nodes and single DWDM channels may fail, a pre-dominant source of unavailability is the physical damage of optical cables. Thus, with the objectives of reducing the overall real CapEx costs and removing the complexity due to wavelength assignment and wavelength continuity constraints when configuring p-cycles in a fully transparent network context, this paper addresses the subsequent questions: if it is ultimately glass that fails, what if just the glass is directly replaced? More specifically, what if p-cycles were used to rapidly, simply and efficiently provide for the direct replacement of failed fiber sections with whole replacement fibers?

A Two-Hop Segment Protecting Paradigm that Unifies Node and Span Failure Recovery under p-Cycles

This letter explains that protecting segments of exactly two hops within an optical network is alone enough to prevent both single span and node failure events. More specifically, we show how to transform a network (topology and working routes) from the two-hop segment perspective and then perform a p-cycle design for the new network. Experimental results typically suggest that two-hop protecting p-cycles are preferable to span- and path-protecting p-cycles for degree-3 or more network instances. The overall two-hop segment protection framework is applicable to any other span-survivability scheme.

CAPEX Costs of Lightly Loaded Restorable Networks under a Consistent WDM Layer Cost Model

Shared protection promises the benefits of lower network capacity utilization without sacrificing the availability level of dedicated protection. Shared protection is often evaluated in terms of its spare capacity efficiency, but seldom in terms of real-world CAPEX (capital expenditure) costs. In this paper, we investigate the design of several shared protection architectures under a standardized cost model for the WDM layer, developed recently within the European NOBEL project. This paper presents a comparative study of the implementations and costs of these architectures under this model. Findings show a counterintuitive relationship between network capacity utilization and design cost when the network is lightly loaded.

Solution of a 200-Node p-Cycle Network Design Problem with GA-Based Pre-Selection of Candidate Structures

As a research challenge we have sought to create and solve p-cycle network design problems involving 200 or more nodes. At such problems sizes, the space of all simple cycle structures on the network graph cannot even be known in practice, let alone put into a conventional ILP problem instance. The approach being taken is a combination of GA methods with ILP; GA is guided by a subsidiary ILP surrogate problem to preselect a set of collectively high merit candidate cycles to populate a size-reduced final fully detailed ILP. Feasible solutions of high quality have been obtained for an initial 200- node test case. Comparison of the result by other workers is invited.

Whole fiber switched p-cycles

In the design of survivable optical networks, the cost and complexity of wavelength assignment and conversion and wavelength-selective switching is always a dominant consideration. And yet, while nodes and single DWDM channels may fail, a pre-dominant source of unavailability is physical damage to optical cables. Thus, we have considered: If it is ultimately glass that fails, what if we just replace the glass directly? More specifically, what if p-cycles were used to rapidly, simply and efficiently provide for the direct replacement of failed fiber sections with whole replacement fibers? As long as the loss budgets are adequate, entire DWDM wavebands could be restored with no switching or manipulation of individual lightpaths. Following a substitution transient, the DWDM layer would never know the break happened. In environments where fiber switching devices are low cost, and ducts are full of dark fiber, this could provide a very low cost alternative to protect an entire DWDM transport layer (or working capacity envelope) against the single largest cause of outage. Here, we make a first proposal of considering this approach. A main motivation is to remove the complexity due to wavelength assignment and wavelength continuity constraints when configuring p-cycles in a fully transparent network context. Another objective is the overall real CAPEX and OPEX cost reductions.