Volume 31, Issue 1

In the last decade, we witnessed the unparalleled success of communication networks such as the Internet or wireless mobile communication networks. The unprecedented scale of the aforementioned networks does not mark the end of the growth, but the emergence of wireless sensor networks and the “Internet of Things” are going to further add to the complexity of today's communication networks.

ZUSAMMENFASSUNG Die Erforschung und Anwendung von autonomen Systemen ist momentan in der Informatik ein Themengebiet von wachsendem Interesse. Die Aussicht, mit autonomen Verfahren komplexe Systeme handhabbar zu machen und Kosteneinsparungen bei deren Betrieb zu erzielen, hat bereits die Softwareindustrie auf dieses Thema gelenkt und zu neuartigen Produkten geführt. Andererseits darf die Verlässlichkeit eines Systems nicht aufgrund eines autonomen Verfahrens herabgesetzt werden. Dieses Spannungsfeld ist ein Fokus verschiedener Forschungsbemühungen, um autonome Systeme alltagstauglich zu machen. Beim Gebiet der autonomen Systeme handelt es sich um ein junges Themengebiet, welches noch nicht durch allgemein akzeptierte Definitionen geprägt ist. Dadurch entsteht der Bedarf einer terminologischen Basis, die sich momentan in der Phase der Etablierung befindet. Diese Arbeit beteiligt sich an diesem Prozess und schlägt ein konzeptionelles Modell vor. Dieses Modell benennt die grundlegenden Termini und zeigt deren Zusammenhänge auf. Es beschreibt eine Interpretation der relevanten Begriffe und leitet daraus Relationen ab. Auf diese Weise fördert es das gemeinsame Verständnis und erleichtert die Kommunikation bezüglich spezifischer Fragestellungen innerhalb dieses Gebietes.

ABSTRACT Until recently, it has been hard to obtain a predictable performance from wireless sensor networks, not to mention performance guarantees. In particular, a widely accepted and established methodology for modelling the performance of wireless sensor networks is missing. In the last three years we have made a step into the direction of an analytical framework for the performance modelling of wireless sensor networks based on the theory of network calculus, which we customized towards a so-called sensor network calculus [Jens B. Schmitt and Utz Roedig: Sensor Network Calculus – A Framework for Worst Case Analysis. In: Proc. of IEEE/ACM Int. Conf. on Distributed Computing in Sensor Systems (DCOSS '05), pages 141-154. Springer, LNCS 3560, June 2005.]. We believe the sensor network calculus to be especially useful for applications which have timing requirements. Examples for this class of applications are factory control, nuclear power plant control, medical applications, and any alerting systems. In general, whenever the sensed input may necessitate immediate actions, the sensor network calculus may be the way to go. In this paper we summarize these activities and discuss the open issues for such an analytical framework to be widely accepted.

ABSTRACT Dynamic overlay networks on the Internet and especially ad hoc networking scenarios often rely on self-organizing schemes for distributed communication and for construction and modification of the topology. When there are no central management facilities and search indices available, flooding is a standard method to collect knowledge about the network structure and the content on the nodes. In recent time, random walks have attracted attention as an alternative search method whose efficiency has been demonstrated by simulation studies. In this paper we use transient analysis as a simple and scalable approach to examine the properties of random walks. The convergence to steady state and the coverage of the network in the course of the random work are main characteristics of interest. In our evaluations we consider randomly build networks in comparison to structured networks of different type and size. In addition to basic random walks, an improvement by avoiding a direct return to the previous node is included.

ABSTRACT Maintenance overhead of a structured peer-to-peer network under churn depends on many factors. One of these factors is the created and maintained overlay topology. We describe these structured overlays as sets of possible overlay graphs for both stable and transient states due to network node leave and join events. We use the graph edit distance concept to determine the minimal communication cost of moving from a transient state to a stable state. Based on these communication costs we establish a theoretical lower bound on the maintenance overhead as the function of the churn rate. By using our model it is possible to determine whether the overlay topology maintained by the self-organizing algorithm of a structured peer-to-peer system is a bottleneck under dynamically changing environments. We analyze the Chord and the eQuus systems as examples.

ABSTRACT This article describes the application of the theory of retrial queues in capturing certain aspects of self-organizing behavior that arises, for example, in Peer-to-Peer networks. It can be shown that retrials have a fair effect on the performance of such self-organizing systems, and thus, should be taken into account adequately during the design and evaluation of these systems. Moreover, it can be shown that there is a notable difference between finite-source and infinite-source retrial queueing models. The main goal of this paper is to show the practical applicability of retrial queues and some of their varieties.

ABSTRACT Computing technology quickly pervades our living environment. We are getting used to carrying with us powerful devices with computing and communication capabilities. The execution context of applications on these devices changes continuously and applications need to adapt in a self-organizing manner, to retain their usefulness. Parameter and compositional adaptation are two frequently used mechanisms for self-adaptive applications in pervasive computing environments. Both of them have advantages as well as disadvantages. Our aim is to provide a comprehensive model-driven application development approach that combines the two concepts, in order to benefit from the advantages of both mechanisms and to reduce development effort. In this paper, we present a modeling approach for self-adaptive applications that combines parameter and compositional adaptation in an integrated, homogeneous fashion. The resulting adaptation model is automatically transformed to source code which is used by the middleware to manage the adaptation at application run-time. The approach was successfully tested in the development of two comprehensive pilot applications.

ABSTRACT Recent peer to peer protocols have mainly been investigated and tested by discrete-event network simulators, e.g. ns-2. These software products usually model the network topology in great details, thus giving reliable picture about a possible run of the modeled protocol. In most cases they can estimate performance measures like throughput, traffic intensity, probability of failure, etc. These protocols are often analyzed in many ways, non of them are standard. In this paper, we elaborate on a new approach based on cellular automata designed to prove correctness of distributed protocols in some cases. Cellular automata are well known for their complex global behaviors, although they are constructed from simple and similar pieces. We shall demonstrate how standard distributed protocols can be modeled by cellular automata. The correctness of the protocols can also be proved by this technique, or – in some cases –counter-examples may be found. Our approach is supported by a simulator that can help researchers to model their protocols by the proposed method.

ABSTRACT Performance assessment is a key issue of the development of self-organizing systems. To this end, both analysis and simulation techniques, augmented by physical experiments, can be used. These approaches have their specific challenges, but are in fact complementary. In this paper, we use design models of self-organizing systems specified with SDL directly for performance simulations. To obtain accurate performance assessments, our approach is to simulate all resource bottlenecks influencing system performance simultaneously. For this purpose, we have devised PartsSim, a component-based simulator that is capable of evaluating the performance of SDL system models, taking limitations of both hardware platform and network into account. We assess the credibility of simulation models by real-world experiments. Several performance simulations of a simple, self-organizing MicaZ network scenario show the value of the additional accuracy achieved with PartsSim. Finally, we discuss how this approach scales to larger systems.

ABSTRACT In a wireless environment, guaranteeing QoS is challenging because applications at multiple devices share the same limited radio bandwidth. In this paper we introduce and study a resource management model for centralized wireless networks, using feedback control theory. Before applying in practice, the proposed model is evaluated using the well-known 20-sim dynamic system simulator. The experimental results show that flexible and efficient resource allocation can be achieved for a variety of system parameters and WLAN scenarios; however, care should be taken in setting the control parameters and coefficients.

ZUSAMMENFASSUNG, EINLEITUNG HPC, also High Performance Computing, ist ein zunehmend bedeutenderes Instrument zur wissenschaftlichen Erkenntnisgewinnung in allen Disziplinen. Simulationen gewinnen aber auch in Wirtschaft und Industrie an Gewicht, weil Spitzenprodukte oft nur mittels numerischer Simulation entworfen werden können. Zumindest für die sogenannten Grand Challenges ist jedoch die benötigte IT-Infrastruktur so teuer, dass sich eine Einzelinstitution die Beschaffung und den Betrieb von Capability-Systemen nicht leisten kann. Deshalb bedarf es regionaler, nationaler und europäischer HPC-Konzepte, die die auf dem Ausbildungs- und Entwicklungssektor erforderlichen Maßnahmen koordinieren und eine Finanzierung von Investition und Betrieb einer HPC-Infrastruktur ermöglichen. Der Beitrag schildert die derzeitige Situation in der BRD und die europäischen Aktivitäten. Dazu muss auch auf die jeweilige Vorgeschichte eingegangen werden, um die Herangehensweise an die möglichen Konzepte zu verstehen. Wichtige Aspekte betreffen das Gauss Centre for Supercomputing (GCS), die zu gründende HPC-Allianz und das europäische Vorhaben PRACE. (Dieser Beitrag basiert auf dem gleichlautenden Vortrag, den der Autor im Namen von GCS bei dem ZKI-Herbsttreffen 2007 in Wilhelmshaven hielt.)