F. De Greve

Managing Ethernet aggregation networks for fast moving users

Delivering broadband services to fast moving users (e.g., in trains) involves both wired and wireless networks. In this article we demonstrate that a hierarchical wired Ethernet aggregation network in combination with Ethernet-based wireless access networks is an interesting approach for realizing broadband access to fast moving users. The focus of the article is on fast and autonomie setup and removal of tunnels in the Ethernet aggregation network for providing broadband services to trains. The resulting moving tunnels realize seamless connectivity for the users on a train with an efficient resource reservation in the aggregation network, A hybrid approach is presented, which allows tunnel-setup times to be minimized and provides accurate tunnel-setup triggers. A testbed has been set up to evaluate the hybrid approach and the results of this work are intended to be used as a guideline for network service providers on how to set up an Ethernet-based aggregation network for fast moving users

Towards Ethernet-Based Wireless Mesh Networks for Fast Moving Users

In this paper, we examine the capabilities of switched Ethernet for building wireless mesh networks (WMNs) and more specific for the support of fast moving users. We motivate that switched Ethernet with wireless extensions is a promising auto-configurable WMN technology. An inherent Ethernet feature, topology modification according to the actual link rates, will be used to avoid poor wireless links in the active topology. We demonstrate that fast recovery is feasible in a WMN by extending the standard spanning tree protocols. Simulations and analytical results will reveal some basic differences if the wired links of switched Ethernet are replaced by wireless links and this leads to a parameter evaluation of the IEEE 802.11e protocol for optimizing the bandwidth usage of the wireless medium

On the management of aggregation networks with rapidly moving traffic demands

In this paper, the focus is on the management aspects for delivery of multimedia services to fast moving users (e.g., users in trains, cars or airplanes). The considered network architecture consists of two parts: an aggregation network part and an access network part. We designed a management system for the aggregation network. The main functionalities of this management system are: (i) pre-configuration of the QoS (quality of service) tunnels, based on the results of the network capacity planning process, (ii) activation of the QoS tunnels when required and (iii) tunnel reconfiguration when the activated tunnels are insufficient to carry the instantaneous demand or in case of other failures. The main difference with existing management systems is that it is designed to deal with the rapidly moving nature of the traffic conditions and the generation of triggers for tunnel activations or reconfigurations. The management system focuses on multimedia service delivery to trains through high speed Ethernet aggregation networks. An extended GVRP (GARP (generic attribute registration protocol) VLAN registration protocol) and a new GARP protocol (G2RP) were designed and implemented as protocols for the automatic tunnel pre-configuration and activation, respectively. The management system makes use of optimization algorithms for network dimensioning and tunnel path determination. These algorithms are described in detail and their evaluation results are compared for a basic train scenario. Finally, measurement results on the performance of the designed protocols for tunnel pre-configuration and activation, are presented in detail.

A new carrier grade aggregation network model for delivering broadband services to fast moving users

In this article, we present the research challenges that are associated with designing a cost-effective network architecture for delivering broadband services to fast moving users (e.g. in trains). We specifically extended the standard Switched Ethernet technology towards a truly Carrier Grade network solution for fast moving users. Prototype implementations allow us to evaluate dynamic tunnel setup mechanisms and to prove that fast Ethernet recovery is feasible by extending the existing spanning tree mechanisms. For architectures with multiple spanning trees the problem arises as to how the spanning trees have to be configured. Therefore, we propose time-efficient algorithms which solve the problem of aggregating paths into a minimal set of spanning trees. In the performance evaluation section, we compare vulnerable centralized backup systems to systems relying on distributed spanning tree-based recovery and it is shown that the former require more spanning tree instances to be configured than the latter for the same set of backup paths. The presented methods and results show that Ethernet technologies are well suited for building flexible and robust network solutions that can support fast moving users. Copyright

Rapidly recovering Ethernet networks for delivering broadband services on the train

With the currently emerging trials for Internet solutions on the train, it is a matter of time before best-effort Internet on the train becomes a reality. However, designing the aggregation network which transports data from fast moving users to service providers, requires extended research because an aggregation network responsible for the reliable transport of broadband multimedia traffic has not been studied before. We propose a flexible recovery strategy for Ethernet networks with dynamic VLANs. It is shown that fast recovery can be achieved in aggregation networks if standard Ethernet protocols are adjusted to co-operate with the fast and efficient link probe mechanism. Additionally a cost-effective design method is presented for networks of realistic size which aims at combining requirements of the commuters (such as end-to-end delay variations) with reliability constraints

Cost-effective Ethernet routing schemes for dynamic environments

Design of aggregation networks for delivering multimedia services to fast moving users is an interesting topic. Research has already been devoted to the dimensioning taking into account the movement of the users. However, the effects of packet loss and packet reordering when switching paths have not been taken into account before. This paper presents a design method that aims at minimizing packet loss and packet reordering in the dimensioning and routing process. Consequently, various routing techniques are presented and their performance is thoroughly compared. The results show that for some routing strategies only a slight extra installation cost is required for guaranteeing minimal packet loss and reordering. Deployment of routing schemes in Ethernet networks requires mapping of the calculated routes on different spanning tree instances. A path aggregation algorithm is presented that forms a minimal set of spanning tree instances out of a set of routing paths.