Karsten Saller

Reducing feature models to improve runtime adaptivity on resource limited devices

Mobile devices like smartphones are getting increasingly important in our daily lifes. They are used in various environments and have to dynamically adapt themselves accordingly in order to provide an optimal runtime behavior. Naturally, adapting to continuously changing environmental conditions is a challenging task because mobile devices are always limited in their resources and have to adapt in real-time. In this paper, we introduce an approach that enables resource limited devices to adapt to changing conditions using dynamic software product lines techniques. Therefore, feature models are reduced to a specific hardware context before installing the adaptive mobile application on the device. This reduces the amount of possible configurations that are compatible with the device and, thereby, minimizes the costs and the duration of an adaptation during runtime.

Modularizing Triple Graph Grammars Using Rule Refinement

Model transformation plays a central role in Model-Driven Engineering. In application scenarios such as tool integration or view specification, bidirectionality is a crucial requirement. Triple Graph Grammars (TGGs) are a formally founded, bidirectional transformation language, which has been used successfully in various case studies from different applications domains. In practice, supporting the maintainability of TGGs is a current challenge and existing modularity concepts, e.g., to avoid pattern duplication in TGG rules, are still inadequate. Existing TGG tools either provide no support at all for modularity, or provide limited support with restrictions that are often not applicable. In this paper, we present and formalize a novel modularity concept for TGGs: Rule refinement, which generalizes existing modularity concepts, solves the problem of pattern duplication, and enables concise, maintainable specifications.

Context-aware DSPLs: model-based runtime adaptation for resource-constrained systems

Dynamic Software Product Lines (DSPLs) provide a promising approach for planning and applying runtime reconfiguration scenarios to adaptive software systems. However, applying DSPLs in the vital domain of highly context-aware systems, e.g., mobile devices, is obstructed by the inherently limited resources being insufficient to handle large, constrained (re-)configurations spaces. To tackle these drawbacks, we propose a novel model-based approach for designing DSPLs in a way that allows for a trade-off between precomputation of reconfiguration scenarios at development time and on-demand evolution at runtime. Therefore, we (1) enrich feature models with context information to reason about potential context changes, and (2) specify context-aware reconfiguration processes on the basis of a scalable transition system incorporating state space abstractions and incremental refinement at runtime. We illustrate our concepts by means of a smartphone case study and present an implementation and evaluation considering different trade-off metrics.

Benchmarking decentralized monitoring mechanisms in peer-to-peer systems

Decentralized monitoring mechanisms enable obtaining a global view on different attributes and the state of Peer-to-Peer systems. Therefore, such mechanisms are essential for managing and optimizing Peer-to-Peer systems. Nonetheless, when deciding on an appropriate mechanism, system designers are faced with a major challenge. Comparing different existing monitoring mechanisms is complex because evaluation methodologies differ widely. To overcome this challenge and to achieve a fair evaluation and comparison, we present a set of dedicated benchmarks for monitoring mechanisms. These benchmarks evaluate relevant functional and non-functional requirements of monitoring mechanisms using appropriate workloads and metrics. We demonstrate the feasibility and expressiveness of our benchmarks by evaluating and comparing three different monitoring mechanisms and highlighting their performance and overhead.

Designing benchmarks for P2P systems

In this paper we discuss requirements for peer-to-peer (P2P). benchmarking, and we present two exemplary approaches to benchmarks for Distributed Hashtables (DHT) and P2P gaming overlays. We point out the characteristics of benchmarks for P2P systems, focusing on the challenges compared to conventional benchmarks. The two benchmarks for very different types of P2P systems are designed applying a common methodology. This includes the definition of the system under test (SUT). and particularly its interfaces, the workloads and metrics. A set of common P2P quality metrics helps to achieve a comprehensive selection of workloads and metrics for each scenario.

A Framework for Bidirectional Model-to-Platform Transformations

Model-Driven Engineering (MDE) has established itself as a viable means of coping with the increasing complexity of software systems. Model-to-platform transformations support the required abstraction process that is crucial for a model-driven approach and are, therefore, a central component in any MDE solution. Although there exist numerous strategies and mature tools for certain isolated subtasks or specific applications, a general framework for designing and structuring model-to-platform transformations, which consolidates different technologies in a flexible manner, is still missing, especially when bidirectionality is a requirement.

Model-driven rapid prototyping with programmed graph transformations

Modern software systems are constantly increasing in complexity and supporting the rapid prototyping of such systems has become crucial to check the feasibility of extensions and optimizations, thereby reducing risks and, consequently, the cost of development. As modern software systems are also expected to be reused, extended, and adapted over a much longer lifetime than ever before, ensuring the maintainability of such systems is equally gaining relevance. In this paper, we present the development, optimization and maintenance of MoSo-PoLiTe, a framework for Software Product Line (SPL) testing, as a novel case study for rapid prototyping via metamodelling and programmed graph transformations. The first part of the case study evaluates the use of programmed graph transformations for optimizing an existing, hand-written system (MoSo-PoLiTe) via rapid prototyping of various strategies. In the second part, we present a complete re-engineering of the hand-written system with programmed graph transformations and provide a critical comparison of both implementations. Our results and conclusions indicate that metamodelling and programmed graph transformation are not only suitable techniques for rapid prototyping, but also lead to more maintainable systems.

Decentralized monitoring in peer-to-peer systems

From the early days, with the design of peer-to-peer overlays or with the decentralized storage and retrieval of content [2], researchers began to investigate how to control and manage these peer-to-peer systems. One important step towards the control and management of them is the assessment of the system’s performance. For this purpose, monitoring constitutes an inevitable and necessary element, because it provides the required data basis comprising information about the system and its participating peers. Given this information, the peers themselves or an overlay operator are able to adapt and improve the system according to changing parameters and conditions. Examples for the utilization of monitored data are manifold: (i) Bubblestorm [30] or Viceroy [21] use the monitored number of peers to influence the overlay construction, (ii) DASIS [1] improves a peer’s join process based on monitored data, and (iii) InfoEye [18] even monitors the access frequency of monitored data to reduce latency and cost for the provisioning of such information. Due to the versatile applicability, a multitude of decentralized monitoring mechanisms for peer-to-peer systems have been developed to provide meaningful statistics about the system and its participating peers. Each of these approaches satisfies different requirements with a varying performance. They can range from heuristic snapshots at low cost to detailed views of the system at higher cost, assuming static or highly dynamic peer-to-peer systems. Out of this set of mechanisms, the decision for the selection of an appropriate monitoring mechanism is a problem. Due to the varying requirements as well as achieved performance and resulting cost, a fair comparison between different solutions is hard to achieve, if not impossible. To overcome the lack of comparability between existing approaches, we present a benchmarking methodology for decentralized monitoring mechanisms in peer-to-

D4M, a Self-Adapting Decentralized Derived Data Collection and Monitoring Framework

Peer-to-peer systems are evolving as a viable distributed resource sharing paradigm on the Internet. The trend is growing towards the usage of such decentralized systems because they are more scalable and resource efficient than centralized systems. Current decentralized systems, like peer-to-peer networks, lack functionality to adapt the transmission of certain information artifacts, according to their access patterns. Additionally, there is still no approach for an efficient dependency management between distributed and dependent information in decentralized networks. This paper presents our ideas of D4M, a framework for the management of distributed derived data in decentralized systems, and how such data can be handled in an efficient manner.