In this paper, we address the problem of routing and wavelength assignment of multicast sessions with subwavelength traffic demands. We consider this problem in the scenario of WDM ring networks. In order to support multicasting, individual nodes need to have the capability to duplicate traffic. We consider two different node architectures which perform the duplication in optical and electronic domain, respectively. As traffic duplication at the electronic level is much more expensive than the optical alternative, we study the problem of assigning routes and wavelengths to the multicast sessions so as to minimize electronic copying. The solution to this problem can be divided into three phases -- (1) routing of multicast sessions, (2) construction of circles by grouping non-overlapping arcs and (3) grouping these circles onto wavelengths. We propose a heuristic algorithm which implements the routing as well as circle construction phases simultaneously and then groups the circles. We present extensive simulation results to show that our approach leads to much lesser equipment cost than that obtained by routing each multicast session along its minimum spanning tree and then using the best known heuristic for circle construction.

We consider the problem of traffic grooming in WDM path, star, and tree networks. Traffic grooming is a variant of the well-known logical topology design problem, and is concerned with the development of techniques for combining low speed traffic components onto high speed channels in order to minimize network cost. Our contribution is two-fold. In the first part of the paper we settle the complexity of traffic grooming in path and star networks by proving that a number of variants of the problem are computationally hard. Since routing and wavelength assignment in these two topologies is trivial, these results demonstrate that traffic grooming is itself an inherently difficult problem. Our results have implications for ring and other more general topologies, which we explore. In the second part, we design practical grooming algorithms with provable properties. Specifically, for all three topologies, we obtain a series of lower and upper bounds which are increasingly tighter but have considerably higher computational requirements; the series of upper bounds form an algorithm for the traffic grooming problem with strong performance guarantees. We also present corresponding heuristics with good performance. Our work is a first step towards a formal and systematic approach to the grooming problem in general topologies that builds upon results and algorithms for more elementary networks.

This paper considers the problem of grooming multicast traffic in WDM networks, with arbitrary mesh topologies. The problem is different from grooming of unicast traffic, since traffic can be delivered to destinations through other destinations in the same set, or through branching points. The paper presents an optimal Integer Linear Programming (ILP) formulation in order to minimize the cost of the network in terms of the number of SONET Add/Drop Multiplexers (ADM). The formulation also minimizes the number of wavelength channels used in the network, and does not allow bifurcation of traffic. Since the ILP formulation is able to solve limited size problems, the paper also introduces a heuristic approach to solve the problem.

In this paper, we investigate the problem of enhancing dual-failure restorability in path-protected mesh-restorable optical wavelength division multiplexed (WDM) networks. A key finding of recent studies that have demonstrated the need to survive simultaneous dual-link failures is that designs providing complete (i.e. 100%) protection from all dual-failures may need almost thrice the spare capacity compared to a system that protects against all single-link failures. However, it has also been shown that systems designed for 100% single-link failure protection can provide reasonable protection from dual-link failures. Thus, the motivation of this work is to develop a hybrid mechanism that provides maximum (close to 100%) dual-failure restorability with minimum additional spare capacity. The system architecture considered is a circuit-switched WDM network with dynamic arrival of sessions requests. We also consider sparse wavelength conversion, where only some nodes have converters. We propose an adaptive mechanism, which we term active protection, that builds upon a pro-active path protection to provide complete single-failure restorability and adds dynamic segment-based restoration. The objective is to optimize network survivability (and minimize spare capacity needs) with ragard to dual-link failures while maintaining complete single-failure restorability. The basic premise of the algorithm is to identify scenarios in the dual-link failure model that necessitate additional spare capacity and provide protection for those scenarios only. Our findings indicate that the proposed scheme achieves close to complete (100%) dual-failure restorability with only maximum of 3% wavelength-links needing two backups even at high loads. Moreover, at moderate to high loads, our scheme attains close to 16% improvement over the base model that provides complete single-failure restorability.

This paper proposes a novel linear formulation for the problem of segment shared protection, where the switching/merging nodes and the least-cost link-disjoint working and protection segments corresponding to each switching/merging node-pair are jointly determined for a connection request. A novel approach of arc-reversal graph transformation is introduced. We verify the ILP and compare it with three reported approaches for solving the segment shared protection problem, namely CDR, PROMISE, and OPDA, by launching dynamic connection requests on two network topologies. From the experiment results, we observe that the ILP can always yield better results in terms of the total cost taken by the working and protection segments. We conclude that the proposed ILP formulation is a step ahead of the most state-of-the-art techniques in solving the shared protection problem, which provides a means of evaluating any other segment shared protection algorithms.

We study the forcer concept in the context of p-cycle based networks. A simple but efficient forcer analysis method is proposed specifically for span-restorable networks in general. Besides identifying forcers, the method is also capable of exploiting extra servable working channels given an initial network spare capacity budget designed for pre-existing working capacities. We find that a large number of extra working channels can be served with no increase in the pre-planned spare capacity budget. This attribute of p-cycle protected networks can be used to enhance their ability to serve unforeseen demand patterns or provide an expanded envelope of protected working capacity within which dynamic demand is servable without blocking due to exceeding the protected working capacity limits.

All-optical networks face the challenge of reducing slower opto-electronic conversions by managing assignment of traffic streams to wavelengths in an intelligent manner, while at the same time utilizing bandwidth resources to the maximum. This challenge becomes harder in networks closer to the end users that have insufficient data to saturate single wavelengths as well as traffic streams outnumbering the usable wavelengths, resulting in traffic grooming which requires costly traffic analysis at access nodes. We study the problem of traffic grooming that reduces the need to analyze traffic, for a class of network architecture most used by Metropolitan Area Networks; the star network. The problem being NP-complete, we provide an efficient twice-optimal-bound greedy heuristic for the same, that can be used to intelligently groom traffic at the LANs to reduce latency at the access nodes. Simulation results show that our greedy heuristic achieves a near-optimal solution.

Traffic grooming, in which low-rate circuits are multiplexed onto wavelengths, with the goal of minimizing the number of add-drop multiplexers (ADMs) and wavelengths has received much research attention from the optical networking community in recent years. While previous work has considered various traffic models and network architectures, protection requirements of the circuits have not been considered for the most part. In this paper, we consider survivable traffic grooming, or grooming traffic which contains a mix of circuits that need protection and that do not need protection. We assume a unidirectional ring network with all-to-all symmetric traffic with t greater than or equal to 1 circuits between each node pair, of which s require protection. As it turns out, survivable traffic grooming presents a significant tradeoff between the number of wavelengths and the number of ADMs, which is almost non-existent in non-survivable traffic grooming for this type of traffic. We explore this tradeoff by establishing bounds and optimal results for some specific cases in this paper.

Traffic grooming is a significant task in internetworking between an optical wavelength-routed core network that supplies "pipes" at the wavelength-granularity, and the attached client (e.g., IP) networks that usually require connections of sub-wavelength granularity. The focus of this study is to develop a theoretical performance analysis model for online traffic grooming in mesh optical networks. This paper first briefly discusses the difficulty in applying the analytic models developed for circuit-switched networks (including wavelength-routed optical networks) to the traffic grooming problem. It then develops a link blocking model based on the continuous time Markov chain and queueing theory, and finally conducts end-to-end performance analysis based on the Erlang fixed-point approximation. The results obtained from the analytic model are shown to match well with numerical results obtained from simulations.

In this paper, we study novel shared path protection schemes that use different policies to deal with different dual-link failure scenarios in order to minimize total bandwidth consumption (or spare capacity allocation). Our work addresses two open problems. The first is minimum total bandwidth allocation under deterministic routing (i.e., when three link-disjoint paths, one active path and two backup paths are pre-selected for each connection). For the dual-link failure case, this problem is much more complex than a similar problem for the single-link failure case because one needs to decide which one of the two backup paths for a connection is preferred to reroute the traffic along its active path affected by a single-link (or a simultaneously dual-link) failure. We propose a new concept of policy, which specifies the preferred backup path to use, and study different policies (including no policy) which will result in different performances in terms of amount of bandwidth needed and different implementation complexity. The second open problem we address in this paper is to minimize additional bandwidth to be allocated for each connection under several different policies and adaptive routing in on-line setting. In this case, the solution entails how to find optimal paths for each connection. From our simulation results, we find that shared path protection with a policy can considerably reduce bandwidth consumption compared with shared path protection without a policy.

We discuss the problem of designing translucent optical networks composed of restorable, transparent subnetworks interconnected via transponders. We formulate the problem of designing restorable subnetworks in translucent networks as an Integer Linear Programming (ILP) problem, where the subnetworks are determined subject to the constraints that each subnetwork satisfies size limitations and it is 2-connected. A greedy heuristic algorithm for the same problem is also proposed for planar network topologies. We propose section restoration for translucent networks where failed connections are rerouted inside the subnetwork which contains the failed link. The network design problem of determining working and restoration capacities with section restoration is formulated as an ILP problem. Numerical results show that section restoration generates fiber costs which are close to those with the path restoration technique for the mesh topologies used in this study. It is also shown that the number of transponders with the translucent optical network is substantially reduced compared to opaque networks.

The static provisioning problem in wavelength-routed optical networks has been studied for many years. However, service providers are still facing the challenges arising from the special requirements for provisioning services at the optical layer. In this paper, we incorporate some realistic constraints into the static provisioning problem, and formulate it under different network resource availability conditions. We consider three classes of shared risk link group (SRLG)-diverse path protection schemes: dedicated, shared, and unprotected. We associate with each connection request a lightpath length constraint and a revenue value. When the network resources are not sufficient to accommodate all the connection requests, the static provisioning problem is formulated as a revenue maximization problem, whose objective is maximizing the total revenue value. When the network has sufficient resources, the problem becomes a capacity minimization problem with the objective of minimizing the number of used wavelength-links. We give integer linear programming (ILP) formulations for these problems. Because solving these ILP problems is extremely time consuming, we propose a tabu search heuristic to solve these problems within a reasonable time. Experimental results are presented to compare the solutions obtained by an ILP solver, the tabu search heuristic and a divide-and-conquer greedy heuristic.

Traffic grooming in optical networks has gained significance due to the prevailing sub-wavelength requirement of end users. Optical networks get upgraded to the latest technology slowly with time, with only a subset of nodes being upgraded to the latest technology. The optical networks are thus comprised of nodes employing heterogeneous switching architectures. In this paper, we develop a framework, called Methodology for Information Collection and Routing in Optical Networks (MICRON), for connection establishment in WDM grooming networks with heterogeneous switching architectures. We illustrate with examples some of the information that could be collected from a link. The link information may be combined in a variety of ways to obtain different properties of a path. We complete the MICRON framework by providing a generic channel assignment procedure that may be adapted to implement different channel assignment schemes. The framework can be easily implemented with simple traffic engineering extensions to the already existing routing protocols in the wide-area networks.

Provisioning dynamic traffic is a major challenge in wavelength-routed networks. While the routes of connections are mainly determined by using global state information, different connections that share the same link may still compete for the same resources. These conflicts can cause significant blocking, a problem that becomes more serious as traffic becomes more dynamic and as the state information changes more frequently. Therefore, it is increasingly important to design appropriate signaling schemes to avoid blocking due to contention between different simultaneous connection requests. In this paper, we propose a new distributed signaling scheme, Intermediate-Node Initiated Reservation (IIR), for establishing dynamic lightpaths in wavelength-routed networks. In the framework of IIR, reservations may be initiated at any set of nodes along the path of a connection, and multiple wavelengths may be reserved. Two IIR implementations are designed for networks with no wavelength conversion and with sparse wavelength conversion. Extensive simulation results show that the IIR scheme outperforms the original schemes under various network and traffic conditions. Simulation results also show that the IIR scheme performs better in networks where number of wavelength per fiber is large, or in networks with sparse wavelength conversion.

We investigate new connection-provisioning algorithms to efficiently provide signal-quality-guaranteed connections in an all-optical WDM mesh network. In the all-optical network, signal degradations incurred by non-ideal transmission medium accumulate along a lightpath. When the signal degradation reaches a certain level, the connection is not usable and is blocked due to transmission impairments in the physical layer. To ensure high service quality of provisioned connections, it is essential to develop intelligent routing and wavelength assignment (RWA) algorithms which can combat the effects of impairments when setting up a connection. For this purpose, we propose two impairment-aware RWA algorithms, namely impairment-aware best-path (IABP) algorithm and impairment-aware first-fit (IAFF) algorithm. The optical signal-to-noise raito (OSNR) requirement and polarization mode dispersion (PMD) effect are used as signal-quality constraints to avoid setting up a connection with unacceptable quality due to the effects of transmission impairments. With the signal-quality consideration, as compared to algorithms that are not impairment aware in a realistic optical network, our proposed impairment-aware algorithms efficiently provide signal-quality-guaranteed connection while significantly reducing connection-blocking probability, better utilizing network resources, and having a reasonable computational requirement. Also, the effect of channel bit rate is studied in this paper.

With the advent of intelligent IP over optical networks, like GMPLS, connections can be protected against failures effectively; however, to capitalize the advantages, novel sophisticated methods are needed. This paper addresses the task of finding efficient mapping in a survivable multilayer network in order to ensure high availability for connections. Known methods (like running a shortest path algorithm) do not consider finding physically disjoint paths in the upper layer and thus cause failure propagation. Besides formulating the problem, we propose a randomized heuristic method to solve it. The quality of the solution is evaluated (1) by the number of node-pairs for which physically-disjoint path-pair can be found in the upper layer, or (2) by the number of spans used by both working and protection paths (i.e., failure propagation effect). It is shown with numerous simulations that our proposed method finds solution for significantly more node pairs (86% instead of 45% in the 35-node network) than traditional methods. Furthermore, it yields connection availabilities near to the optimum.

In resilient optical networks, there is a tradeoff between the amount of resources allocated for protection versus the probability that a failed working path can not be recovered, known as protection blocking probability. Often the network topology permits multiple protected groups of working paths (WPs) to share protection bandwidth and other network resources. The Common Control and Measurement Plane (CCAMP) working group in the IETF has defined an (M:N)ⁿ shared recovery scheme, in which defined n WP groups each consisting of N WPs and M backup paths (BPs) share some or all of the BP resources. In this paper, we present an analytical model that predicts protection blocking probability as a function of BP resource sharing for this shared recovery scheme. We also propose an algorithm that efficiently manages BP resources while doing protection assignments. We provide numerical results that highlight the benefits and tradeoffs involved. Our analytical model can assist in providing engineering guidelines to service providers so that they can effectively allocate resources and manage protection and restoration in their networks.

The amount of resource provisioning prior to failure events to support optical path recovery has a significant impact on network performance, and so designing recovery mechanisms for any large network necessitates balancing multiple (in some cases, competing) requirements. The Common Control and Measurement Plane (CCAMP) working group's Protection and Restoration Design Team conducted a qualitative analysis of path protection schemes using different degrees of backup resource pre-provisioning. The resulting analysis grid highlights some of the trade-offs between the different approaches. In this paper, we describe the results of a simulation study that we conducted using the NIST GMPLS/Lightwave Agile Switching Simulator (GLASS) simulation tool. By measuring network performance with respect to the metrics used by the design team, we were able to produce quantitative results that confirm the design team's qualitative analysis and provide additional information for carriers and service providers who are designing optical networks.

High susceptibility to adverse atmospheric conditions can severely limit the use of free-space optical systems for critical applications. This paper proposes a new architecture for lasercom systems for proactive adaptability during adverse atmospheric conditions. The hardware and software components of the proposed architecture are described in detail. We also present an in-field lasercom test-bed setup, the wireless channel propagation measurements recorded using the test-bed, and the results to validate the recommended design.

Leveraging the dominance of Ethernet in LANs and SONET/SDH in MANs and WANs, we propose a service called CHEETAH (Circuit-switched High-speed End-to-End Transport ArcHitecture). The service concept is to provide end hosts with high-speed, end-to-end circuit connectivity on a call-by-call shared basis, where a "circuit" consists of Ethernet segments at the ends that are mapped into Ethernet-over-SONET long-distance circuits. This paper focuses on the file-transfer application for such circuits. For this application, the CHEETAH service is proposed as an add-on to the primary Internet access service already in place for enterprise hosts. This allows an end host that is sending a file to first attempt setting up an end-to-end Ethernet/EoS circuit, and if rejected, fall back to the TCP/IP path. If the circuit setup is successful, the end host will enjoy a much shorter file-transfer delay than on the TCP/IP path. To determine the conditions under which an end host with access to the CHEETAH service should attempt circuit setup, we analyze mean file-transfer delays as a function of call blocking probability in the circuit-switched network, probability of packet loss in the IP network, round-trip times, link rates, and so on.

Packet switched based network architectures exhibit a high degree of resource sharing and consequently make very efficient use of the available bandwidth. On the other hand, they experience a great amount of transit traffic in IP routers, increasing costs. Wavelength switched based concepts can reduce this transit traffic, but have limited resource sharing and consequently need more resources (wavelengths) to avoid losses. We present a new hybrid network architecture, Overspill Routing In Optical Networks (ORION), which combines the benefits of wavelength switched networks and packet switched networks. An example node hardware design and corresponding control architecture is presented. A case study quantifying the benefits of ORION when compared to three other network architectures is also discussed.

Although uniform distribution of optical power is one of the main requirements for multi-access support in a wireless optical link, Multi Spot Diffusing Configuration (MSDC) provides uniformity along with spatial independence that allows spatial diversity techniques over the link. Independent spatial channels are generated by a multibeam transmitter producing spatially confined diffusing spots, and a multibranch receiver with narrow field-of-view branches. In this paper, we propose an orthogonal spatial coding technique that utilizes the independence of channels to reduce power and bandwidth requirements. The technique is based on treating the m channels between a transmitter and a receiver as an m-dimensional space. Thus, enabling data transmission through varying signal level and its location in space. Our study shows that using 8 channels, the proposed technique enables transmission at rates 10 and 12 times higher than combining when power per channel and user are constrained, respectively.

In high speed DWDM (Dense Wavelength Division Multiplexing) networks, meeting the targeted physical-layer performance is largely dependent on the total number and the placement of the optical amplifiers. An efficient design not only cuts down the network cost by minimizing the required number of amplifiers, but also effectively reduces the undesirable fiber nonlinearities and the end-to-end OSNR (Optical Signal-to-Noise Ratio) degradation on the propagating channels. By capturing major transmission impairments, such as fiber attenuation, nonlinear effects, amplifier and receiver noises, this paper proposes an algorithm for a given static lightpath scenario to find the most efficient solution of the amplifier placement that satisfies the performance constraints. We then adapt this algorithm to the dynamic traffic case. Numerical results show that our algorithm outperforms uniform placement schemes in controlling the nonlinear effects and providing targeted BER performance using the minimum number of amplifiers.

Packet-by-packet selective photonic add/drop multiplexer, of the finest data granularity, is experimentally demonstrated at 40Gbit/s. Optical code-label, attached to the packet, enables to determine in optical domain either to drop, cut through, or add packets.

In this paper, two switch models -- MOD1, employs a set of nondegenerate fiber delay-lines (FDLs) shared among the input lines, and MOD2, employs a set of degenerate FDLs and tunable wavelength converters (TOWCs) shared among the input lines, are proposed to handle contention. We demonstrate that for a prefixed packet loss probability constraints, e.g., 10^-6, and for nonbursty traffic, if the wavelengths per fiber is no less than 8, a very small number of FDLs is sufficient to obtain a reasonable packet loss probability for both models, i.e., without the need to use TOWCs for MOD2, used for MOD1 equals to those of the FDLs and TOWCs used for MOD2, MOD2 performs much better than MOD1 even if no TOWC is used. With the increase of the average burst length, the number of TOWCs needed by MOD2 needs to be increased so as to maintain a reasonable packet loss probability. However, even for the traffic with high degree of burstness, MOD2 is still a cost effective and robust solution.

Wavelength conversion has been shown as one of the key techniques that can improve the blocking performance in a wavelength-routed all-optical network. Given that wavelength converters nowadays are still very expensive, how to make effective use of the limited number of wavelength converters becomes an important issue. In this paper, we propose a novel sparse-partial wavelength conversion (SPWC) architecture with the inherent flexibility that can facilitate network carriers to migrate the optical backbone to support wavelength conversion. We demonstrate that this architecture can significantly save the number of wavelength converters while still achieving excellent blocking performance. We further investigate the wavelength converter placement problem. Simulation results indicate that, with appropriate wavelength assignment and wavelength converter placement scheme, the performance of the wavelength-routed all-optical network with only 1-5% of wavelength conversion capability is very close to that of the networks with full-complete wavelength conversion capability.

In current routing algorithms for wavelength switched optical transport networks, based on the Open Constrained Shortest Path First (OCSPF) concept, the effects of traffic dynamicity (the fact that lightpaths (a.k.a. optical connections or optical LSPs) are continuously being set-up and torn down) are not taken into account. Therefore, in this study four new link metrics are proposed, each of which could be used to replace the current metrics being used. Two metrics are based on the average load ρ on a link during a given timeframe. Two other metrics are based on an estimate of the number of setup requests n_br that will be blocked during the lifetime of the lightpath that is being set up. It is shown that the former two link metrics (i.e. those based on ρ) significantly reduce the blocking probability compared to the case in which a constant link metric is used. The latter two metrics (i.e. those based on n_br) however perform only slightly better or even worse than the original algorithm. In addition, it is observed that by taking into account the traffic dynamicity using the average load on a link, the amount of sustainable demand can be increased by about 10% for a given blocking probability.

Recently, optical burst switching and aggregated optical packet switching have gained significant exposure as possible future mechanisms for routing aggregated IP traffic over all-optical core networks. However, the limited buffering capacity in all-optical networks presents a major challenge, as current IP traffic displays strong self-similar properties. Reducing the burst loss rate of such long-range dependent traffic can be costly requiring a significant increase in either the network bandwidth or the buffer size of optical cross connects. In this paper, we revisit the possibility of using buffers to reduce self-similarity before the traffic is routed onto the all-optical core. The aim of this paper is to increase the understanding of the effect of packet/burst aggregation on the self-similarity measure of the traffic. In particular, we implement a simple burst assembly mechanism with two parameters, the maximum burst length L and the maximum burst delay d, so that incoming traffic is smoothed with a guaranteed delay bound. Unlike previous works, we simulate the burst assembler using more realistic input traffic sources, and analyze the results using both R/S plot and discrete wavelet analysis methods. Our detailed results show that buffering indeed reduces traffic self-similarity (an area of research controversy) when parameters L and d are set appropriately.

A switch matrix implemented as an optical crossbar using semiconductor optical amplifiers is able to accommodate extreme concentrations of data traffic. Due to the need to reduce optical guard band overhead it is beneficial to switch fixed size bursts of data cells on a time slot basis. The high capacity of the optical matrix supports multiple optical ports per burst card, and the implementation of multiple queue servers per burst card helps make better use of the multiplicity of ports. Problems associated with arbitrating multiple ports and multiple servers per burst card have been resolved by extending the operation of existing iterative, single server scheduling algorithms. The multiserver arbitration time will be in proportion to the number of servers -- corresponding to the channels of DWDM link -- unless a reconciliation stage is used after each iteration when an arbiter per server is used. The reconciliation stage sets the problem of broken data dependencies between server arbitrations in this case. Further, to address the time limitations for computing the scheduling solution, parallel arbiter implementations have been developed and tested against single arbiter designs. Again, the broken dependencies between iterations of an arbitration are addressed through the use of a grant reconciliation stage. The use of multiple queue servers per burst card also resolves some of the data loss problems related to polarized traffic. Simulations of the multiple server and parallel arbiter implementations have demonstrated their efficiency compared to previous implementations. Compounded to this problem is maintaining high throughput of the switch matrix while observing data transit time limits. This involves balancing two contradictory requirements; switch or line card efficiency and data transit times. To improve efficiency it is desirable to transmit only full packets. However, to prevent loss of data due to timeout it will be necessary to transmit some incomplete packets. We investigate three approaches -- thrifty, conservative, and greedy request policies. Using data content and age we demonstrate that unevenly distributed traffic can be handled better with multiserver switching matrices.

The paper introduces a new analytical approach for estimating blocking probabilities in all-optical networks. The assumptions of the classical Lee approximation are discussed and their effects are evaluated. The paper proposes a better substitute for the independent link assumption. The new assumption takes all active paths, as well as all free links, as independent objects on the network topology. The new model is shown to generate estimates that fit exactly the blocking probabilities obtained through simulations on linear topologies using Poissonian, spatially homogeneous traffic. An extension to regular meshes is presented that is asymptotically good for very small and very large nodal degrees. Finally, the independent wavelength assumption is also evaluated for two different wavelength assignment algorithms.

In this paper, we address the issue of dynamic load balancing in wavelength division multiplexing (WDM)-based optical burst switching (OBS) networks. We propose a load balancing scheme based on adaptive alternate routing whose objective is to reduce burst loss through load balancing. The key idea of adaptive alternate routing is to reduce network congestion by adaptively distributing the load between two pre-determined link-disjoint alternative paths based on the measurement of the impact of traffic load on each of them. Through extensive simulation experiments for different traffic scenarios, we show that the proposed dynamic load balancing algorithms outperforms the shortest path routing and static alternate routing algorithms.

For lack of optical random access memory, optical fiber delay line (FDL) is currently the only technology to implement optical buffering. Feed-forward and feedback are two types of FDL structures in optical buffering, both have advantages and disadvantages. In this paper, we present a novel architecture for WDM optical packet switches with an effective hybrid FDL buffering that combines the merits of both feed-forward and feedback schemes. The core of the switch architecture is the arrayed waveguide grating (AWG) and the tunable wavelength converter (TWC). It requires smaller optical device sizes and fewer wavelengths and has less noise than feedback architecture. At the same time feed-forward architecture can only do non-preemptive priority routing while ours supports preemptive priority routing. Our empirical results show that the new switch architecture significantly reduces packet loss probability.

We introduce Proactive Reservation-based Switching (PRS) -- a switching architecture for IP/WDM networks based on Labeled Optical Burst Switching. PRS achieves packet delay and loss performance comparable to that of packet-switched networks, without requiring large buffering capacity, or burst scheduling across a large number of wavelengths at the core routers. PRS combines proactive channel reservation with periodic shaping of ingress-egress traffic aggregates to hide the offset latency and approximate the utilization/buffering characteristics of discrete-time queues with periodic arrival streams. A channel scheduling algorithm imposes constraints on burst departure times to ensure efficient utilization of wavelength channels and to maintain the distance between consecutive bursts through the network. Results obtained from simulation using TCP traffic over carefully designed topologies indicate that PRS consistently achieves channel utilization above 90% with modest buffering requirements.

In this paper we present a dynamic bandwidth allocation algorithm for EPON, which makes use of the Multipoint Control Protocol (MPCP) with threshold reporting and with inter- and intra-ONU priority scheduling. Three varieties of this algorithm are compared, by means of a detailed simulation program, regarding average packet delay for several priorities, delay variation for constant bit rate (CBR) traffic and bandwidth utilization. We show that by introducing a specific intra-ONU priority scheduling algorithm, which takes the reported values into account, the bandwidth can be fully utilized. However, this scheduling algorithm causes an increased packet delay and delay variation for CBR traffic. In order to eliminate this drawback, we combine this scheduling algorithm with a rate-based scheme for the highest priority (CBR) traffic. This combined algorithm provides an interesting tradeoff between the efficiency, which is still near to the optimal, and the delay characteristics of time critical applications. Finally, we also include a comparison with a standard intra-ONU priority scheme.

In this paper we develop an analytical model for the FIFO delay-line buffer in asynchronous optical networks with any packet length distribution, under the assumption that arrivals are Poissonian. We consider that the incoming traffic is distributed among an infinite number of inputs (Aloha traffic) and show that this consideration represents very suitably a real system. The model enables the exact calculation of packet loss probability and average delay, which makes it a very powerful tool for performance evaluation and planning/dimensioning of networks that use this kind of contention resolution technique. The buffer performance as predicted by the model is compared with simulations and discussed for some packet length distributions.

A design method is proposed that minimizes the deployment cost of a network when only a limited number of OXCs is capable of 3-R regeneration. Optimization, based on a given static-traffic matrix, is carried out by routing lightpaths with constrained signal degradation. Constraint-based routing and wavelength assignment is performed considering accumulation of noise due to amplified spontaneous emission of optical amplifiers as the major cause of signal degradation. A general multilayer design and optimization procedure is developed and applied to some case-study networks. The effects of the variation of design parameters such as amplification-span length, wavelength conversion capability distribution and optical protection capability is investigated.

The problem of integrated routing in multifiber IP over WDM networks is studied in this paper. To solve this problem, a layered-graph is constructed and an algorithm, called integrated cost-based shortest path (ICSP) algorithm, is then proposed. ICSP can not only balance the traffic uniformly but also make a dynamic tradeoff between all the links in the layered-graph. A parameter r_l is introduced to characterize the resource richness of an IP over WDM Network. Simulation results show that ICSP outperforms other algorithms significantly in terms of blocking probability in all the cases we studied, and the performances of other algorithms are affected by r_l greatly. Because a multifiber network can be functionally equivalent to a single fiber network with limited wavelength conversion (WC), we make the first known attempt to investigate the impact of WC on dynamic integrated routing by studying the multifiber networks. Results show that the effect of WC depends on the granularities of label switched path (LSP) requests. If the granularity of each request is large, WC will improve the network performance; if the granularity of each request is small, WC will worsen the performance.

This paper presents an architecture and a realization of a burst reservation module for optical burst switching using the just-enough-time (JET) reservation scheme. JET is a reserve-a-fixed-duration reservation algorithm, i.e., wavelength channels are allocated exactly for the burst transmission time. As the exact start and end times of all bursts have to be recorded and processed for JET burst reservation, several publications assumed its realization to be prohibitively complex. This paper proposes an architecture for a hardware-based reservation module for JET. This architecture has been described in VHDL and synthesized on an FPGA representative for today's programmable logic technology. The proposed solution is evaluated under dynamic traffic based on timing and resource utilization results taken from the FPGA realization. The results of the performance evaluation prove that with this reservation module JET can even be realized for burst durations in the microsecond range.

In this paper the performance of low complexity optical CDMA systems is evaluated. Beat noise is the main performance limitation for an implementation using an incoherent receiver. Therefore, a channel model with both interference and beat noise is presented. From the channel model a maximum likelihood soft decoding method is derived. The system considered is a passive optical network with tree topology where the headend can use information about the number of active users to improve the receiver performance. In order to simulate systems with forward error correction (FEC) and soft decoding an efficient simulation method is used. The simulations show that the performance is highly overestimated when the beat noise is neglected. The results also show that it is better to use FEC to expand the bandwidth compared with only using CDMA spreading, in particular since the efficiency of FEC can be further increased with soft decoding.

In this paper, we investigate the problem of Preplanned Recovery with Redundant Multicast Trees (PRRMT) in optical networks. The redundant trees ensure the source node remains connected to all destination nodes for a multicast session request under single edge failures. Our objective is to minimize the total number of links used for both trees. We formulate PRRMT as an integer linear program (ILP), and also develop a heuristic algorithm. The ILP approach and heuristic algorithm are experimentally evaluated on 14-node NSFNET and 21-node Italian network. Experimental results show that: (1) ILP approach leads to optimal solutions but requires prohibitively long time, (2) Our heuristic algorithm yields optimal or near-optimal results in very short time, and (3) The edge-disjoint trees can protect the transmission for an edge failure.

We present and evaluate a novel protocol of traffic control that aims at solving the fairness issue typical of shared medium networks such as metropolitan rings. The proposed solution called TCARD (Traffic Control Architecture using Remote Descriptors) is based on a preventive mechanism to grant access to the resource -- i.e. free bandwidth is preserved by a node according to the traffic requirements from the other network nodes. A review of the existing methods used to manage fairness in ring networks is done and points out their inherent problems (performance issues, resource wastage or inadequacy). In contrast, we show how the new protocol addresses the above limitations, notably through a performance evaluation study. The major conclusion is the ability to avoid the degradation in performances and the resource sub-utilization while achieving fairness within the network.