SEAMS 2015

The diversity of software components (e.g., browsers, plugins, fonts) is a wonderful opportunity for users to customize their platforms. Yet, massive customization creates a privacy issue: browsers are slightly different from one another, allowing third parties to collect unique and stable fingerprints to track users. Although software diversity appears to be the source of this privacy issue, we claim that this same diversity, combined with automatic reconfiguration, provides the essential ingredients to constantly change browsing platforms. Constant change acts as a moving target defense strategy against fingerprint tracking by breaking one essential property: stability over time.

We leverage virtualization and modular architectures to automatically assemble and reconfigure software components at multiple levels. We operate on operating systems, browsers, fonts and plugins. This work is the first application of software reconfiguration to build a moving target defense against browser fingerprint tracking. The main objective is to automatically modify the fingerprint a platform exhibits. We have developed a prototype called Blink to experiment the effectiveness of our approach at randomizing fingerprints. We have assembled and reconfigured thousands of platforms, and we observe that all of them exhibit different fingerprints, and that commercial fingerprinting solutions are not able to detect that the different platforms actually correspond to a single user.

Today’s system developers and operators face the challenge of creating software systems that make efficient use of dynamically allocated resources under highly variable and dynamic load profiles, while at the same time delivering reliable performance. Benchmarking of systems under these constraints is difficult, as state-of-the-art benchmarking frameworks provide only limited support for emulating such dynamic and highly variable load profiles for the creation of realistic workload scenarios. Industrial benchmarks typically confine themselves to workloads with constant or stepwise increasing loads. Alternatively, they support replaying of recorded load traces. Statistical load intensity descriptions also do not sufficiently capture concrete pattern load profile variations over time.

To address these issues, we present the Descartes Load Intensity Model (DLIM). DLIM provides a modeling formalism for describing load intensity variations over time. A DLIM instance can be used as a compact representation of a recorded load intensity trace, providing a powerful tool for benchmarking and performance analysis. As manually obtaining DLIM instances can be time consuming, we present three different automated extraction methods, which also help to enable autonomous system analysis for self-adaptive systems. Model expressiveness is validated using the presented extraction methods. Extracted DLIM instances exhibit a median modeling error of 12.4% on average over nine different real-world traces covering between two weeks and seven months. Additionally, extraction methods perform orders of magnitude faster than existing time series decomposition approaches.

A promising avenue to control energy-related costs in enterprise data centers is to investigate power-aware resource management strategies. In this study we investigate techniques to schedule resources adaptively with the sole aim of reducing power consumption. Our approach is based on a characterization of energy usage and resource utilization patterns obtained by monitoring energy consumption in an enterprise data center. We propose an adaptive feature extraction method to classify resource utilization patterns from energy consumption data. Improved classification results are obtained through signal feature extraction prior to the training stages for cascading classifiers for at least 14 different energy usage patterns. Adaptive feature extraction prior to classifier training improved class identification even further. The identified patterns can now be used as a basis for adaptive resource scheduling within a power-smart data center. The classification method that performed best is part of our proposed energy runtime model and controller which manages and controls the energy consumption in the data center according to usage patterns.

Autonomous robots have a great potential to help in disaster scenarios. Nevertheless, each robot is faced with the variety of these scenarios and the difficulties of cooperation with heterogeneous robots and human actors of other rescue forces. Furthermore, approaches using predefined adaptation models are not flexible enough for these scenarios. Therefore, we propose an adaptive configuration of the information processing to enable the collaboration of heterogeneous groups. We present two contributions in this paper: a) An ontology to describe the semantics of information processing components and the structure of information; b) The composition of these components to an adaptation model at run-time. To coordinate the information sharing and integration of new discovered information sources at run-time, we further provide a general framework. Finally, we describe how to apply our approach in a fictitious scenario.

Designers of self-adaptive systems often formulate adaptive design decisions, making unrealistic or myopic assumptions about the system’s requirements and environment. The decisions taken during this formulation are crucial for satisfying requirements. In environments which are characterized by uncertainty and dynamism, deviation from these assumptions is the norm and may trigger “surprises”. Our method allows designers to make explicit links between the possible emergence of surprises, risks and design trade-offs. The method can be used to explore the design decisions for self-adaptive systems and choose among decisions that better fulfil (or rather partially fulfil) non-functional requirements and address their trade-offs. The analysis can also provide designers with valuable input for refining the adaptation decisions to balance, for example, resilience (i.e. satisfiability of non-functional requirements and their trade-offs) and stability (i.e. minimizing the frequency of adaptation). The objective is to provide designers of selfadaptive systems with a basis for multi-dimensional what-if analysis to revise and improve the understanding of the environment and its effect on non-functional requirements and thereafter decision-making. We have applied the method to a wireless sensor network for flood prediction. The application shows that the method gives rise to questions that were not explicitly asked before at design-time and assists designers in the process of risk-aware, what-if and trade-off analysis.

Self-adaptive systems overcome many of the limitations of human supervision in complex software-intensive systems by endowing them with the ability to automatically adapt their structure and behavior in the presence of runtime changes. However, adaptation in some classes of systems (e.g., safetycritical) can benefit by receiving information from humans (e.g., acting as sophisticated sensors, decision-makers), or by involving them as system-level effectors to execute adaptations (e.g., when automation is not possible, or as a fallback mechanism). However, human participants are influenced by factors external to the system (e.g., training level, fatigue) that affect the likelihood of success when they perform a task, its duration, or even if they are willing to perform it in the first place. Without careful consideration of these factors, it is unclear how to decide when to involve humans in adaptation, and in which way. In this paper, we investigate how the explicit modeling of human participants can provide a better insight into the trade-offs of involving humans in adaptation. We contribute a formal framework to reason about human involvement in self-adaptation, focusing on the role of human participants as actors (i.e., effectors) during the execution stage of adaptation. The approach consists of: (i) a language to express adaptation models that capture factors affecting human behavior and its interactions with the system, and (ii) a formalization of these adaptation models as stochastic multiplayer games (SMGs) that can be used to analyze human-system- environment interactions. We illustrate our approach in an adaptive industrial middleware used to monitor and manage sensor networks in renewable energy production plants.

Self-adaptive systems (SAS) can reconfigure at runtime to mitigate uncertainties posed by environments for which they may not have been explicitly designed. High-assurance SAS applications must continually deliver acceptable behavior for critical services, enabling the need for run-time validation techniques. To this end, run-time testing can provide additional assurance that an SAS will continue to behave as expected while executing under unknown conditions. This paper introduces Proteus, a framework for adaptive run-time testing on an SAS. Proteus facilitates both execution and adaptation of run-time testing activities to ensure that the SAS continues to execute according to its requirements and that both test plans and test cases continually remain relevant to changing operating conditions. We demonstrate our approach by applying it to a simulated self-adaptive remote data mirroring network that must efficiently diffuse data while experiencing adverse operating conditions. Experimental results suggest that Proteus can reduce the number of executed irrelevant, false positive, and false negative test cases at run time to ensure that online testing activities remain relevant as the SAS encounters uncertainty.

Future-generation self-adaptive systems will need to be able to optimize for multiple interrelated, difficult-tomeasure, and evolving quality properties. To navigate this complex search space, current self-adaptive planning techniques need to be improved. In this position paper, we argue that the research community should more directly pursue the application of stochastic search techniques—search techniques, such as hill climbing or genetic algorithms, that incorporate an element of randomness—to self-adaptive systems research. These techniques are well-suited to handling multi-dimensional search spaces and complex problems, situations which arise often for self-adaptive systems. We believe that recent advances in both fields make this a particularly promising research trajectory. We demonstrate one way to apply some of these advances in a search-based planning prototype technique to illustrate both the feasibility and the potential of the proposed research. This strategy informs a number of potentially interesting research directions and problems. In the long term, this general technique could enable sophisticated plan generation techniques that improve domain specific knowledge, decrease human effort, and increase the application of self-adaptive systems.