Brunnert A.,Fortiss GmbH |
Krcmar H.,TU Munich
Journal of Systems and Software | Year: 2016
Continuous delivery (CD) is a software release process that helps to make features and bug fixes rapidly available in new enterprise application (EA) versions. Evaluating the performance of each EA version in a CD process requires a test environment comparable to a production system. Maintaining such systems is labor intensive and expensive. If multiple deployments of the same EA exist, it is often not feasible to maintain test instances for all of these systems. Furthermore, not all deployments are known at the time of a release (e.g., for off-the-shelf products). To address these challenges, this work proposes the use of resource profiles which describe the resource demand per transaction for each component of an EA and allow for performance predictions for different hardware environments and workloads without the need to own corresponding test environments. Within a CD process, resource profiles can be used to detect performance changes in EA versions. Once a version is released, resource profiles can be distributed along with the application binaries to support capacity planning for new deployments. Three integrated experiments for a representative EA provide validation for these capabilities. © 2015 Elsevier Inc.
Natekin A.,Fortiss GmbH |
Knoll A.,TU Munich
Frontiers in Neurorobotics | Year: 2013
Gradient boosting machines are a family of powerful machine-learning techniques that have shown considerable success in a wide range of practical applications. They are highly customizable to the particular needs of the application, like being learned with respect to different loss functions. This article gives a tutorial introduction into the methodology of gradient boosting methods with a strong focus on machine learning aspects of modeling. A theoretical information is complemented with descriptive examples and illustrations which cover all the stages of the gradient boosting model design. Considerations on handling the model complexity are discussed. Three practical examples of gradient boosting applications are presented and comprehensively analyzed. © 2013 Natekin and Knoll.
Shah H.,Fortiss GmbH |
Knoll A.,TU Munich |
Akesson B.,Research Center
Proceedings -Design, Automation and Test in Europe, DATE | Year: 2013
The transition towards multi-processor systems with shared resources is challenging for real-time systems, since resource interference between concurrent applications must be bounded using timing analysis. There are two common approaches to this problem: 1) Detailed analysis that models the particular resource and arbiter cycle-accurately to achieve tight bounds. 2) Using temporal abstractions, such as latency-rate (LR) servers, to enable unified analysis for different resources and arbiters using well-known timing analysis frameworks. However, the use of abstraction typically implies reducing the tightness of analysis that may result in over-dimensioned systems, although this pessimism has not been properly investigated. This paper compares the two approaches in terms of worst-case execution time (WCET) of applications sharing an SDRAM memory under Credit-Controlled Static-Priority (CCSP) arbitration. The three main contributions are: 1) A detailed interference analysis of the SDRAM memory and CCSP arbiter. 2) Based on the detailed analysis, two optimizations are proposed to the LR analysis that increase the tightness of its interference bounds. 3) An experimental comparison of the two approaches that quantifies their impact on the WCET of applications from the CHStone benchmark. © 2013 EDAA.
Giuliani M.,Fortiss GmbH |
Knoll A.,TU Munich
International Journal of Social Robotics | Year: 2013
We present a robot that is working with humans on a common construction task. In this kind of interaction, it is important that the robot can perform different roles in order to realise an efficient collaboration. For this, we introduce embodied multimodal fusion, a new approach for processing data from the robot's input modalities. Using this method, we implemented two different robot roles: the robot can take the instructive role, in which the robot mainly instructs the user how to proceed with the construction; or the robot can take the supportive role, in which the robot hands over assembly pieces to the human that fit to the current progress of the assembly plan. We present a user evaluation that researches how humans react to the different roles of the robot. The main findings of this evaluation are that the users do not prefer one of the two roles of the robot, but take the counterpart to the robot's role and adjust their own behaviour according to the robot's actions. The most influential factors for user satisfaction in this kind of interaction are the number of times the users picked up a building piece without getting an explicit instruction by the robot, which had a positive influence, and the number of utterances the users made themselves, which had a negative influence. © 2013 Springer Science+Business Media Dordrecht.
Prehofer C.,Fortiss GmbH
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2016
In this paper, we consider extensions of state machines with additional functionality. We analyze how typical safety or liveness properties are affected when extending or refining the model. We identify several classes of extensions where properties are preserved. The extensions include adding new transitions at a state, refining transitions, as well as adding failure cases and adding additional, new functionality. We propose new concepts for refinements based on elimination of added behavior with context to capture property-preserving extensions in a precise and (mostly) syntactic way. © Springer International Publishing Switzerland 2016.