Steffen Rothkugel

World Wide Web caching: the application-level view of the Internet

The Internet has fallen prey to its most successful service, the World Wide Web. The networks do not keep up with the demands incurred by the huge amount of Web surfers. Thus, it takes longer and longer to obtain the information one wants to access via the World Wide Web. Many solutions to the problem of network congestion have been developed in distributed systems research in general and distributed file and database systems in particular. The introduction of caching and replication strategies has proven to help in many situations; therefore, these techniques are also applied to the Web. Although most problems and associated solutions are known, some circumstances are different with the Web, forcing the adaptation of known strategies. This article gives an overview of these differences and currently deployed, developed, and evaluated solutions.

WACA: A Hierarchical Weighted Clustering Algorithm Optimized for Mobile Hybrid Networks

Clustering techniques create hierarchal network structures, called clusters, on an otherwise flat network. In a dynamic environment-in terms of node mobility as well as in terms of steadily changing device parameters-the clusterhead election process has to be re-invoked according to a suitable update policy. Cluster re-organization causes additional message exchanges and computational complexity and it execution has to be optimized. Our investigations focus on the problem of minimizing clusterhead re-elections by considering stability criteria. These criteria are based on topological characteristics as well as on device parameters. This paper presents a weighted clustering algorithm optimized to avoid needless clusterhead re- elections for stable clusters in mobile ad-hoc networks. The proposed localized algorithm deals with mobility, but does not require geographical, speed or distances information.

Multimedia content distribution in hybrid wireless networks using weighted clustering

Fixed infrastructured networks naturally support centralized approaches for group management and information provisioning. Contrary to infrastructured networks, in multi-hop ad-hoc networks each node acts as a router as well as sender and receiver. Some applications, however, requires hierarchical arrangements that¿for practical reasons¿has to be done locally and self-organized. An additional challenge is to deal with mobility that causes permanent network partitioning and reorganizations. Technically, these problems can be tackled by providing additional uplinks to a backbone network, which can be used to access resources in the Internet as well as to inter-link multiple ad-hoc network partitions, creating a hybrid wireless network. In this paper, we present a prototypically implemented hybrid wireless network system optimized for multimedia content distribution. To efficiently manage the ad-hoc communicating devices a weighted clustering algorithm is introduced. The proposed localized algorithm deals with mobility, but does not require geographical information or distances

Dynamic multi-hop clustering for mobile hybrid wireless networks

In mobile wireless networks communication is often improved by sending messages along a stable backbone of more reliable communication paths. Building such a backbone requires efficient clustering algorithms which aggregate network nodes into logical groups, each group being managed by a clusterhead and any two neighboring clusters being interconnected by at least one gateway node or gateway path. In this concept k-hop clustering refers to cluster structures where cluster members are at most k hops away from their clusterhead. While the dynamicity of mobile wireless network is often considered as a challenge, in this work we explicitly exploit node mobility in order to support cluster formation and maintenance of k-hop clusters. The described KHOPCA algorithm consists of a set of easy to implement rules which form and maintain k-hop sized clusters in a purely localized way. In a static network cluster formation is limited to a constant number of messages exchanges among neighboring nodes. In dynamic networks the localized nature of the described rules promise a fast cluster convergence and low communication complexity in case of mobility triggered cluster reconfiguration.

Topology-based Clusterhead Candidate Selection in Wireless Ad-hoc and Sensor Networks

Clustering techniques create hierarchal network structures, called clusters, on an otherwise flat network. Neighboring devices elect one appropriate device as clusterhead. Due to the dynamic environment, clusterhead selection becomes an important issue. We consider the problem of appropriate clusterhead selection in wireless ad-hoc networks and sensor networks. This work presents topological criteria for robust clusterhead candidate selection, resilient to sporadic node mobility and failure as well as for efficient information dissemination. One of the main ideas of our approach is to avoid selecting nodes located close to the network partition border as such nodes are more likely to move out of the partition, thus causing a clusterhead re-election. We conducted experiments both for static topologies as well as for cases in the presence of node mobility. Our results showed that the frequency of clusterhead re-election and average shortest path length from the clusterhead decrease when considering topological criteria. Additionally, the clusters tend to be robust to clusterhead failure. The presented mechanisms rely on local topological information only and do not require geographical data.

NLWCA Node and Link Weighted Clustering Algorithm for Backbone-Assisted Mobile Ad Hoc Networks

This paper introduces a clustering algorithm for applications running in backbone-assisted mobile ad hoc networks. The algorithm assigns weights both to the network nodes and to the communication links in order to elect local leaders, so called clusterheads.

Architecture-Centric Software Development Based on Extended Design Spaces

The realization of software projects can be significantly eased by extending the focus of reuse to architectural aspects instead of concentrating on separate software elements. Yet in any case, operational techniques are required to support the retrieval and selection of reusable items. To this end, we extend on the concept of design spaces which allows to describe the relevant properties of software elements in a semi-formal way. Moreover, we show how the concept of extended design spaces can be deployed in tools supporting component-and framework-based software development.

Hycast- podcast discovery in mobile networks

Podcasts are a popular way to provide multimedia information about certain topics. A multitude of podcast servers exist in the Internet, allowing people to subscribe to them. Typically, podcasts are downloaded onto desktop computers and copied on mobile devices to be played while being on the move. In this paper, we extend the idea of podcasts, making them available in mobile network environments. In particular, HyCast does not rely on central podcast directories. Instead, HyCast also allows discovering, subscribing to, and downloading podcasts and episodes in the local neighborhood. For the dissemination of podcast information, we introduce and evaluate two different strategies. One is based on peer-to-peer communication between one-hop neighbors. The second one employs clustering to reduce the overhead of the podcast information dissemination.

LSWTC: A local small-world topology control algorithm for backbone-assisted mobile ad hoc networks

The prevalence of the small-world phenomenon in numerous efficient networks, such as social networks, Internet, nervous systems, implies that small-worlds are an evolutionary solution for locally growing networks. These networks require short communication distances between their nodes in spite of their potentially large network diameter but at the same time are robust against randomly occurring failures. Integrating these properties into human-made communication networks drastically increases their efficiency and performance but represents an enormous challenge in the case of mobile ad hoc networks since their communication graphs rely on local construction principles. In this paper, the focus is on backbone-assisted ad hoc networks. In such networks devices connect arbitrarily to other devices within their transmission range, and dedicated devices are able to connect to a backbone network. This construction principle leads to a geometric graph that has a small average path length, but that is sensitive to random attacks or failures. The challenge in evoking small-world properties in backbone-assisted mobile ad hoc networks is to build a topology control algorithm that works with localized data (i.e. using neighboring information) despite knowledge of average path length and clustering coefficient requiring global information. Such a topology control algorithm is described here. Empirical results show that depending on network density, small-world properties can be efficiently evoked.

HyTrace Backbone-Assisted Path Discovery in Hybrid Networks

HyTrace is a hybrid system providing stable paths between mobile ad hoc nodes. The ad hoc network thereby is organized in clusters where the cluster heads maintain uplinks to a backbone network. The backbone keeps track of available clusters and stable connection between them. The backbone also inter-links different network partitions thus increasing the reliability of the whole system. The underlying Weighted Cluster Path Discovery algorithm still keeps working in settings where no backbone network is available. Broadcasting as well as message exchange between arbitrary nodes is used to further motivate the usefulness of the system.