This paper focuses on the design and analysis of scheduling approaches for Optical Flow Switching (OFS) serving high performance applications with very stringent time deadline constraints. In particular, we attempt to meet setup times only slightly longer than one roundtrip time with networks at moderate to high loading. This paper proposes three possible scheduling mechanisms for OFS connection setup in a WDM network: (i) a simple algorithm, which awards pre-emptive priority to applications requiring time deadline performance; (ii) a multi-path probing mechanism using only coarse average loading information (i.e., no detailed network state information) but without pre-emption; and (iii) a multi-path probing mechanism using periodically updated network state information and without pre-emption. The updating scheme calls for a slow control plane, which refreshes and broadcast network states only periodically on the order of seconds or longer. Our results show that for a low blocking probability, the update interval must be a fraction of the mean service time of transactions. We conclude that this algorithm, a combination of both slow centralized and fast distributed processes, delivers an efficient and scalable control design for a high-speed transport network of the future.

In our previous work, we presented optical flow switching (OFS) as a key enabler of scalable future optical networks. In the present work, we propose a practical scheduling algorithm and conduct an approximate throughput-delay analysis for OFS networks.

In this work, we consider access network design for the Optical Flow Switching (OFS) transport architecture [1]-[3]. Our work addresses the all-optical physical layer of the data plane in the context of tree-based networks. We consider the advantages and disadvantages of passive and active components - including optical amplifiers - and ultimately integrate these building blocks in the most economically attractive fashion for the required number of users.

We compare three optical transport network architectures - optical packet switching (OPS), optical flow switching (OFS), and optical burst switching (OBS) - based on a notion of network capacity as the set of exogenous traffic rates that can be stably supported by a network under its operational constraints. We characterize the capacity regions of the transport architectures, and show that the capacity region of OPS dominates that of OFS, and that the capacity region of OFS dominates that of OBS. We then apply these results to two important network topologies - bidirectional rings and Moore graphs - under uniform all-to-all traffic. Motivated by the incommensurate complexity/cost of comparable transport architectures, we also investigate the dependence of the relative capacity performance of the switching architectures on the number of switch ports per fiber at core nodes.

In this work, we evaluate an attractive candidate for optical network data transport: optical flow switching (OFS). We describe the operation and implementation of the architecture, characterize its capacity region and capacity-cost tradeoff, and compare it to other prominent optical network architectures.

In this work, we conduct a throughput-cost study of several optical network architectures: optical flow switching (OFS), tell-and-go (TaG), electronic packet switching (EPS), and generalized multiprotocol label switching (GMPLS). The simple, multi-tiered optical network that we consider comprises two groups of users, each in a distinct metropolitan-area network (MAN), which wish to communicate over a wide-area network (WAN). Our network cost model focuses on initial capital expenditure: transceiver, switching, routing, and amplification costs. Our results indicate that: OFS is the most scalable architecture of all, in that it is most cost-efficient when the average user data rate is high and the number of users in the network is large; EPS is most sensible when the product of the number of users and the average user data rate is low; the GMPLS architecture, which is conceptually intermediate to EPS and OFS, is optimal when the product of the number of users and the average user data rate is moderate; and, finally, there does not exist an optimal regime for TaG.

We compare three optical transport network architectures—optical packet switching (OPS), optical flow switching (OFS), and optical burst switching (OBS)—based on a notion of network capacity as the set of exogenous traffic rates that can be stably supported by a network under its operational constraints. We characterize the capacity regions of the transport architectures, and show that the capacity region of OPS dominates that of OFS, and that the capacity region of OFS dominates that of OBS. We then apply these results to two important network topologies—bidirectional rings and Moore graphs—under uniform all-to-all traffic. Motivated by the incommensurate complexity/cost of comparable transport architectures, we also investigate the dependence of the relative capacity performance of the switching architectures on the number of switch ports per fiber at core nodes.

In this paper, we consider the design of a physical network topology that meets a high level of reliability using unreliable network elements. We are motivated by the use of networks, and in particular, all-optical networks, for high-reliability applications which involve unusual and catastrophic stresses. Our network model is one in which nodes are invulnerable and links are subject to failure, and we consider both statistically independent and dependent link failures. Our reliability metrics are the all-terminal connectedness measure and the less commonly considered two-terminal connectedness measure. We compare in the low and high stress regimes, via analytical approximations and simulations, common commercial architectures designed for all-terminal reliability when links are very reliable with alternative architectures which are mindful of both of our reliability metrics and regimes of stress. Furthermore, we show that for independent link failures network design should be optimized with respect to reliability under high stress, as reliability under low stress is less sensitive to graph structure; and that under high stress, very high node degrees are required to achieve moderate reliability performance. Finally, in our discussion of correlated failure models we show the danger in relying on an independent failure model and the need for the network architect to minimize component failure dependencies.

In this paper, we consider the design of a physical network topology that meets a high level of reliability using unreliable network elements. We are motivated by the use of networks and, in particular, all-optical networks, for high-reliability applications which involve unusual and catastrophic stresses. Our network model is one in which nodes are invulnerable and links are subject to failure - a good approximation for optical networks with passive nodes and vulnerable fiber under stress of disconnection - and we focus on statistically independent link failures with initial steps taken toward generalization to dependent link failures. Our reliability metrics are the all-terminal connectedness measure and the less commonly considered two-terminal connectedness measure. We compare in the low and high stress regimes, via analytical approximations and simulations, common commercial architectures designed for all-terminal reliability when links are very reliable with alternative architectures which are mindful of both of our reliability metrics and regimes of stress. We derive new results especially for one of these alternative architectures, Harary graphs, which have been shown to possess attractive reliability properties. Furthermore, we show that for independent link failures network design should be optimized with respect to reliability under high stress, as reliability under low stress is less sensitive to graph structure; and that under high stress, very high node degrees and small network diameters are required to achieve moderate reliability performance. Finally, in our discussion of correlated failure models, we show the danger in relying on an independent failure model and the need for the network architect to minimize component failure dependencies.

Temporal skewing, a hitherto unexamined effect of chromatic dispersion on multiwavelength optical code-division multiple-access (MW-O-CDMA) systems, is explored through simulation of realistic networks. In addition, two novel methods for combating the deleterious effects of temporal skewing are proposed: optimum threshold detection and code-pattern preskewing. It is shown that for a practical system configuration with optimum threshold detection, MW-O-CDMA networks can maintain design bit error rate performance within an order of magnitude for link lengths of up to approximately 500 m. Our results show that the effect merits careful attention in the design and realization of any MW-O-CDMA system based on two-dimensional codes. Code pattern preskewing presents the possibility of completely eradicating the impact of temporal skewing.