Stephan Schuhmann’s research while affiliated with Universität Stuttgart and other places

Complex pervasive applications need to be distributed for two main reasons: due to the typical resource restrictions of mobile devices, and to use local services to interact with the immediate environment. To set up such an application, the distributed components require spontaneous composition. Since dynamics in the environment and device failures may imply the unavailability of components and devices at any time, finding, maintaining, and adapting such a composition is a nontrivial task. Moreover, the speed of such a configuration process directly influences the user since in the event of a configuration, the user has to wait. In this article, we introduce configuration algorithms for homogeneous and heterogeneous environments. We discuss a comprehensive approach to pervasive application configuration that adapts to the characteristics of the environment: It chooses the most efficient configuration method for the given environment to minimize the configuration latency. Moreover, we propose a new scheme for caching and reusing partial application configurations. This scheme reduces the configuration latency even further such that a configuration can be executed without notable disturbance of the user.

As the size and complexity of Pervasive Computing environments increases, configuration and adaptation of distributed applications gains importance. These tasks require automated system support, since users must not be distracted by the (re-)composition of applications. In homogeneous ad hoc scenarios, relying on decentralized configuration schemes is obviously mandatory, while centralized approaches may help to reduce latencies in weakly heterogeneous infrastructure-based environments. However, in case of strongly heterogeneous pervasive environments including several resource-rich and resource-weak devices, both approaches may lead to suboptimal results concerning configuration latencies: While the resource-weak devices represent bottlenecks for decentralized configuration, the centralized approach faces the problem of not utilizing parallelism. Instead, a hybrid approach that involves only the subset of resource-rich devices is capable of rendering configuration and adaptation processes more efficiently. In this paper, we present such a resource-aware hybrid scheme that effectively reduces the time required for configuration processes. This is accomplished by a balanced-load clustering scheme that exploits the computational power of resource-rich devices, while avoiding bottlenecks in (re-)configurations. We present real-world evaluations which confirm that our approach reduces configuration latencies in heterogeneous environments by more than 30 % compared to totally centralized and totally decentralized approaches. This is an important step towards seamless application configuration.

Das Forschungsgebiet des Pervasive Computing hat in den letzten Jahren zunehmend an Bedeutung gewonnen. Hauptziel ist die Integration von Computertechnologie in Alltagsgegenstände und die Nutzung dieser hierdurch elektronisch angereicherten Benutzerumgebung, ihrer Geräte und Dienste für die Ausführung von Anwendungen. Dafür müssen diese Anwendungen in die Lage versetzt werden, sich dynamisch an wechselnde Umgebungen anzupassen, beispielsweise durch Verlagerung ihrer Funktionalität zwischen Geräten. Bedingt durch dynamische Umgebungen, Nutzermobilität sowie drahtlose Kommunikationstechnologien ist die Entwicklung von Anwendungen für Pervasive Computing Umgebungen hochkomplex. Daher wurden in diesem Projekt grundlegende Konzepte und Algorithmen entwickelt, um eine automatisierte Nutzerunterstützung in diesen Umgebungen zu ermöglichen. Hierbei lag der Schwerpunkt auf der Entwicklung von Algorithmen zur Selbstkonfiguration von Anwendungen mittels automatisierter Komposition und Adaption. Neben dynamischen, homogenen Ad Hoc Umgebungen sollten außerdem heterogene Umgebungen, in denen zusätzlich ressourcenstarke Infrastrukturgeräte vorhanden sind, berücksichtigt werden, um eine effiziente Ausführung von Konfigurationen und Adaptionen auch in solchen Umgebungen zu ermöglichen.

Two essential services of Ad-hoc networks are IP address autoconfiguration and intrusion detection systems. Since both autoconfiguration and intrusion detection may base their decisions on routing protocol anomalies, their interdependencies can cause problems. In this paper, we present our approach to efficiently combine autoconfiguration and intrusion detection, and present our enhancements in attack detection for an autoconfiguration system. We have identified anomalies indicating specific attacks, implemented suitable anomaly detectors, and evaluated our system. The results show that it is possible to detect both the attacks and IP address conflicts in an efficient way.

Numerous current research projects deal with the issue of automatic application configuration in pervasive computing scenarios. While completely distributed configuration is inevitable in infrastructure-less ad hoc scenarios, many realistic pervasive application scenarios are located in heterogeneous environments where additional computation power of resource-rich devices can be utilized. However, most of the current projects either focus solely on smart environments which rely on additional infrastructure devices, or they address ad hoc environments and treat all involved devices as equal. This leads to suboptimal results in case of present resource-rich devices, as their additional computation power is not exploited. In this paper, we present a framework that is based on clustering and allows the efficient and flexible support of automatic application configuration both in infrastructure-based and ad hoc environments. This is realized by a new concept called Virtual Container that enables the local emulation of remote devices functionalities and resources. In our evaluation results, we prove that this concept considerably reduces configuration delays noticeable by the user.

Location-awareness is highly relevant subject in ubiquitous computing, as many applications exploit location information to provide adequate services or adapt to a changing physical environment. While GPS provides reliable outdoor localization, indoor positioning systems present a bigger challenge. Many indoor localization systems have been proposed. However, most of them rely on customized hardware or presume some dedicated infrastructure. In this paper, we focus on WLAN-based localization in smart ubiquitous environments. We propose an improved scheme of the Weighted Centroid Localization (WCL) algorithm that is robust and provides higher location accuracy than the original WCL algorithm. The improvements are based on the use of dynamic weighting factors that are solely dependent on the correlation of the Received Signal Strength Indicators of the received beacon signals. Compared to the original WCL scheme, our approach does not increase requirements to the environment. Real-world experiments in a typical environment that we report on in this paper confirm that the increased location accuracy determined in previous calculations is reproducible in a realistic noisy environment. This provides a simple, cost-efficient, and battery-conserving, but yet adequate technique for getting the accurate location information of mobile devices.

The adaptation of pervasive applications is in the focus of many current research projects. While decentralized adaptation is mandatory in infrastructureless ad hoc scenarios, most realistic pervasive application scenarios are situated in heterogeneous environments where additional computation power of resource-rich devices can be exploited. Therefore, we propose a hybrid approach to application configuration that applies centralized as well as decentralized configuration as appropriate in the given environment. In this paper we introduce the Direct Backtracking algorithm that represents an efficient way for centralized configuration and adaptation of pervasive applications in heterogeneous scenarios. In our evaluation, we show that compared with other centralized algorithms, our algorithm significantly reduces adaptation latency as it avoids unnecessary adaptations that arise in many other backtracking algorithms, without significantly increasing memory waste. This is achieved by introducing two mechanisms: 1. proactive backtracking avoidance and 2. intelligent backtracking.