Franz Rambach

A multilayer cost model for metro/core networks

A capital expenditure model is a key requirement to evaluate multilayer (ML) metro and core network architectures. Based on the Internet protocol/multiprotocol label switching (IP/MPLS), MPLS-transport profile, optical transport network and WDM technology, a detailed capital expenditure model developed by researchers from system vendors and network operators is introduced. Besides current equipment and corresponding prices for the different layers, it also contains predictions for technology evolution and pricing until 2018. We show how to determine the price of anML node by constructing it from components specified in the cost model. We use the model in a case study where we benchmark the price of an integrated IP/MPLS/WDM solution.

Reach-dependent capacity in optical networks enabled by OFDM

In todays optical networks, the channel capacity is designed for the systems maximum reach. By using OFDM, reach dependent capacity can be realized which improves the network performance substantially.

Cost Comparison of Networks Using Traditional 10 and 40 Gb/s Transponders Versus OFDM Transponders

OFDM allows flexible adjustment of transmission data rate according to channel properties. We propose to use OFDM transponders and calculate the cost at which OFDM networks result in lower CAPEX compared to traditional networks.

Efficient Protection in Single-Domain Networks Using Network Coding

Quality of service requirements and the increasing amount of data transmission demand for efficient protection mechanisms. Existing mechanisms like 1+1 and 1+N path protection offer instantaneous recovery but they suffer from high capacity needs and nodal degree requirements, respectively. The high nodal degree requirement makes an implementation of the 1+N mechanism in transport networks difficult. In this paper, we develop two new mechanisms by applying network coding on a virtualized protection scheme. These mechanisms are more resource-efficient compared to 1+1, and are implementable in transport networks even under stringent nodal degree constraints. The difference between the two mechanisms lies in complexity and performance. Therefore, either of them can be preferred as a good compromise between 1+1 and 1+N in different scenarios according to the needs and available resources.

An adaptive inter-domain PCE framework to improve resource utilization and reduce inter-domain signaling

Upcoming broadband commercial and scientific applications are now demanding high bandwidth pipes across multiple domains with guaranteed Quality of Service (QoS). Recent research initiatives such as the Path Computation Element (PCE) framework are focusing on the development of scalable multi-domain QoS provisioning frameworks, especially within the emerging carrier grade transport technologies based on layer-2 tunnels. QoS provisioning across multiple domains requires that QoS parameters for available transit paths inside a domain be advertised in the inter-domain routing algorithms, while the dynamic inter- and intra-domain connections vary the available resource, and hence require frequent inter-domain updates. The signaling load on the other hand hampers the scalability of the inter-domain routing mechanisms. We propose the use of an adaptive partitioning framework, which can effectively use network resources and at the same time stabilize the advertised domain topologies and thus path advertisements. Our method partitions network resources by pre-reserving resources for inter-domain transit traffic, and uses policies to modify the resource partitioning in order to maintain the available transit capacity between specified bounds. We show by simulations that the proposed mechanism can reduce inter-domain signaling load by 10%-20% and reduce overall blocking inside a domain by creating a trade-off between available resources for intra-domain connections and inter-domain transit connections. The reduction in inter-domain signaling and blocking can be used as a building block to design scalable QoS routing systems for carrier grade transport networks.

Network Coding for Protection against Multiple Link Failures in Multi-Domain Networks

Many networks today are composed of numerous subnetworks or domains. The connection points of these domains are important and vulnerable parts of the communication. Therefore, one of the key points of protection schemes today is being able to protect these interconnections. Hence, in this type of networks a protection mechanism is applied, where there is 1+1 protection inside each domain and dual homing at the borders. With this mechanism, the breakdown due to the failure of the interconnection elements is prohibited. This system enables the network to survive any node or link failure in each domain. However, it suffers from the high capacity need on the border links, where data has to be transmitted bidirectionally for each single demand. We are showing that combining network coding with this protection mechanism gives a realization, which outperforms the existing mechanism by providing high resource savings. We present the theoretical basis and the application methodology of the proposed method.

Assessment and performance evaluation of PCE-based inter-layer traffic engineering

Multi-layer (ML) and multi-domain networks require path computation elements (PCE) for constraint-based path calculation. In this paper we introduce and evaluate qualitatively as well as quantitatively the PCE scenarios newly proposed by the IETF for PCE-based inter-layer traffic engineering. Requirements on additional communication, on hardware and on optimality of path computation serve as the qualitative metrics in our comparison. The path setup delay is derived analytically and serves as the quantitative metric. We derive the results using simulations on the ML German reference network with 17 nodes and back-up our results by two different ML TE routing algorithms. We show that one single ML PCE performs best in the overall qualitatively and quantitatively comparison.

Embedding optical ethernet services within the path computation element framework: The 100GET approach

We address the issues of inter-domain optical service provisioning with carrier-grade Ethernet and propose a multi-layer PCE based control plane architecture to implement adaptive advance reservation, which can significantly improve resource utilization and reduce blocking.

Adaptive advance reservation based inter-domain framework

There is a growing demand for end-to-end quality of service in present day multi-domain networks. IP/BGP is not equipped to support end-to-end QoS required by new upcoming services, and scalable QoS paradigms are required in lower layers to support future applications. Frequent fluctuations of available inter-domain resources inside a domain hinder the development of a scalable QoS routing paradigm. To address this, we propose a mechanism to create stable inter-domain topologies by employing adaptive advance reservation for inter-domain connections inside a domain. Advance reservation implies that capacity is pre-reserved for inter-domain traffic in the form of layer-2 tunnels, and simple policies are introduced to control the capacity of these tunnels. We study this proposal within the concept of the path computation element (PCE) framework and evaluate the performance of dynamic and adaptively advance-reserved inter-domain connections and topologies.

Availability of Connections in Ethernet over DWDM Core Networks

This paper investigates the availability of connections in core networks which use Ethernet directly over the optical DWDM layer. We study transparent, opaque and hybrid connections. We show that the availability of a protected hybrid connection can be accurately estimated through calculating the availability of a protected transparent path or protected opaque path. The selected optical node architecture has even a higher influence on the reliability compared to the choice to route opaque or transparent. To achieve availabilities over 99.999% - often required by providers - is only possible if the end nodes are completely doubled.