Andreas Drakos

Optical packet routers: how they can efficiently and cost-effectively scale to petabits per second

Todays data-centric networks are becoming progressively dynamic with respect to the transported traffic volume, to the spatial and temporal variations of traffic patterns, and to the subsequent interconnection request patterns. Optical packet routers (OPRs) aim to provide a viable answer to these requirements by rendering the optical layer adaptable, reconfigurable at will, and cost-effective by means of statistical multiplexing of the network resources while satisfying end-to-end quality-of-service requirements. We study the role of OPRs in two different network approaches: an OPR adopting the legacy of a telecommunication solution and a solution pertinent to the role and the functionality of an IP world. To benchmark the two scenarios, a multilayer, multigranular OPR architecture is presented and its potential to cost-effectively scale toward petabit-per-second throughput is justified by means of physical layer performance and power consumption estimations for each case. Furthermore, the blocking and frame-loss performance of the proposed OPRs is presented, proving the viability of the proposed solutions.

A Hybrid Optical Switch Architecture with Shared Electronic Buffers

In this paper we present a hybrid opto-electronic switch architecture, which can be enhance the performance of optical packet switching (OPS) schemes. The proposed architecture is based on the use of electronic line-cards and packet cross-connect to a limited degree utilizing the large buffering capacity available only by means of electronic subsystems. This hybrid architecture achieves efficient utilization of optical network resources, which can be assigned dynamically in small time scales, limiting loss probability with controlled average latency.

QoS performance benchmarking of networking paradigms in core networks

We benchmark OCS, OFCS, OBS and CANON against QoS parameters. The results demonstrate that sub-walength switching alone cannot guarantee efficiency. We show that CANON can simultaneously provide for both, statistical multiplexing gains and QoS guarantees outperforming the other paradigms.

Performance benchmarking of core optical networking paradigms

The sustainability of Future Internet critically depends on networking paradigms able to provide optimum and balanced performance over an extended set of efficiency and Quality of Service (QoS) metrics. In this work we benchmark the most established networking modes through appropriate performance metrics for three network topologies. The results demonstrate that the static reservation of WDM channels, as used in IP/WDM schemes, is severely limiting scalability, since it cannot efficiently adapt to the dynamic traffic fluctuations that are frequently observed in todays networks. Optical Burst Switching (OBS) schemes do provide dynamic resource reservation but their performance is compromised due to high burst loss. It is shown that the CANON (Clustered Architecture for Nodes in an Optical Network) architecture exploiting statistical multiplexing over a large scale core optical network and efficient grooming at appropriate granularity levels could be a viable alternative to existing static as well as dynamic wavelength reservation schemes. Through extensive simulation results we quantify performance gains and we show that CANON demonstrates the highest efficiency achieving both targets for statistical multiplexing gains and QoS guarantees.

Evaluation of optical core networks based on the CANON architecture

Clustering of nodes in optical networks has been proven to be an efficient way to serve end-to-end connectivity. However, clustering requires specific topological characteristics, or alternatively the introduction of significant alterations of an existing topology to achieve the expected performance improvements. The comparison of future dynamic optical networking technologies should therefore include in the set of initial assumptions, apart from the statistical properties of the traffic load, the network topology to draw conclusions regarding the efficiency as well as feasibility and scalability of the proposed solutions. In this article, we show how node clustering under the CANON architecture can be applied in real-life core networks and provide superior performance compared to conventional burst switching techniques in terms of blocking, resource utilization and power consumption.

A Node Clustering Algorithm for the CANON Architecture

In this article, we present an algorithm for clustering a mesh network implementing the Clustering Architecture of Nodes in Optical Networks (CANON). In order to evaluate the impact of the resulting clustering solutions on actual network efficiency, specific cost criteria are used.

Impact of node clustering on network cost and resource utilization

In this work we benchmark through appropriate performance metrics the legacy Optical Circuit Switching (OCS) mode in an optical network of nodes following a mesh topology interconnection vs. a clustered optical network architecture (CANON). The results demonstrate that the static reservation of WDM channels, as used in IP/WDM schemes, is severely limiting scalability, since it cannot efficiently adapt to the dynamic traffic fluctuations that are frequently observed in todays networks, while the hierarchical clustered architecture with dynamic reservations can exploit statistical multiplexing efficient grooming traffic at appropriate granularity levels leading to improved performance and resource utilization. We quantify through computer simulation the performance gains and the smoothing of traffic profile achieved when implementing the CANON architecture as an upgrade of existing infrastructure by using the reference network of a European operator. The effect of traffic grooming, when employing CANON and its impact on performance and cost are evaluated using a national backbone network as a case study.

Dynamic resource allocation with service guarantees over large scale optical networks

The lack of effective contention resolution mechanisms in the optical domain presents dynamic optical switching architectures with a hard dilemma between high loss (when adopting one-way reservations) and high delay (when using two-way reservations). In this work we evaluate an alternative hierarchical network architecture, which achieves a satisfactory compromise by partitioning the network into a number of geographically limited domains operating in slot and frame synchronous mode where two-way reservations are effective.

Multi-granular optical cross-connects for the CANON network scenario

Todaypsilas core networks deploy wavelength division multiplexing (WDM) to offer high capacity of the order of 1 Tb/sec and long reach transmission of the order of 1000 km. In such systems, the information is transmitted optically on different wavelengths but it is transferred across the network through fibre links that are terminated by SONET/SDH equipment. In such networks, layer crossing is inevitable, requiring deep data inspection and complex protocol processing, resulting in limited scalability and high installation, operation and maintenance cost. It is however the only reliable end-to-end service delivery scenario available today that utilises the optical bandwidth and guarantees provisioning of resources. In this paper we review the alternative concept of CANON (clustered architecture for nodes in optical networks), a scenario that utilises clustering of nodes in ring topologies to create a hierarchy in transporting traffic in a dynamic, future proof and scalable manner. Here we emphasize on the optical crossconnect architectures that can serve such a network scenario.

An Efficient Optical Switch Architecture with Controlled Latency for GRID Networks

In this paper we evaluate a hybrid optoelectronic switch architecture, which can be used under optical burst switching (OBS) or even optical packet switching (OPS) schemes. The proposed architecture incorporates electronic switching equipment to a limited degree utilizing the large buffering capacity available only by means of electronic subsystems. This hybrid architecture achieves efficient utilization of optical network resources, which can be assigned dynamically in small time scales, limiting loss probability with controlled average latency. Thus, the proposed switch architecture is suitable for critical networking applications with high capacity and dynamically changing requirements like GRID networking.