Catálogo de publicaciones - libros

This work-in-progress paper surveys error detection techniques for transient, timing, permanent and logical errors in system-on-chip (SoC) design and discusses their applicability in the design of monitors for our Autonomic SoC architecture framework. These monitors will be needed to deliver necessary signals to achieve fault-tolerance, self-healing and self-calibration in our Autonomic SoC architecture. The framework combines the monitors with a well-tailored design methodology that explores how the Autonomic SoC (ASoC) can cope with malfunctioning subcomponents.

Self-optimization is a promising approach to cope with the increasing complexity of today’s automation networks. The high complexity is mainly caused by a rising amount of network nodes and increasing real-time requirements. Dynamic hardware reconfiguration is a key technology for self-optimizing systems, enabling, e.g., Real-Time Communication Systems (RCOS) that adapt to varying requirements at runtime. Concerning dynamic reconfiguration of an RCOS, an important requirement is to maintain connections and to support time-constrained communication during reconfiguration. We have developed a dynamically reconfigurable Ethernet switch, which is the main building block of a prototypic implementation of an RCOS network node. Three methods for reconfiguring the Ethernet switch without packet loss are presented. A prototypical implementation of one method is described and analyzed in respect to performance and resource efficiency.

Coordination of altruistic agents to solve optimization problems can be significantly enhanced when inter-agent communication is allowed. In this paper we present an evolutionary approach to learn optimal communication structures for groups of agents. The agents learn to solve the , but our ideas can easily be adapted to other problem fields. With our approach we can find the optimal communication partners for each agent in a static environment. In a dynamic environment we figure out a simple relation between each position of agents in space and the optimal number of communication partners. A concept for the establishment of relevant communication connections between certain agents will be shown whereby the space the agents are located in will be divided into several regions. These regions will be described mathematically. After a learning process the algorithm assigns an appropriate number of communication partners for every agent in an - arbitrary located - group.

Consider a (robotic) explorer starting an exploration of an unknown terrain from its base station. As the explorer has only limited communication radius, it is necessary to maintain a line of robotic relay stations following the explorer, so that consecutive stations are within the communication radius of each other. This line has to start in the base station and to end at the explorer.

In the simple scenario considered here we assume an obstacle-free terrain, so that the shortest connection (the one which needs the smallest number of relay stations) is a straight line. We consider an explorer who goes an arbitrary, typically winding way, and define a very simple, intuitive, fully local, distributed strategy for the relay stations — our strategy — to maintain a line from the base station to the robot as short as possible.

Besides the definition of this strategy, we present an analysis of its performance under different assumptions. For the static case we prove a bound on the convergence speed, for the dynamic case we present experimental evaluations that show the quality of our strategy under different types of routes the explorer could use.

Self-optimizing mechatronic systems react autonomously and flexibly to changing conditions. They are capable of learning and optimize their behavior throughout their life cycle. The paradigm of self-optimization is originally inspired by the behavior of biological systems. The key to the successful development of self-optimizing systems is a conceptual design process that precisely describes the desired system behavior. In the area of mechanical engineering, active principles based on physical effects such as friction or lever are widely used to concretize the construction structure and the behavior. The same approach can be found in the domain of software-engineering with software patterns such as the broker-pattern or the strategy pattern. However there is no appropriate design schema for the development of intelligent mechatronic systems covering the needs to fulfill the paradigm of self-optimization. This article proposes such a schema called Active Patterns for Self-Optimization. It is shown how a catalogue of active patterns can be derived from a set of four basic active patterns. This design approach is validated for a networked mechatronic system in a multiagent setting where the behavior is implemented according to a biologically inspired technique — the neuro-fuzzy learning method.

Self-optimizing mechatronic systems have the ability to adjust their goals and behavior according to changes of the environment or system by means of complex real-time coordination and reconfiguration in the underlying software and hardware. In this paper we sketch a generic software architecture for mechatronic systems with self-optimization and outline which analogies between this architecture and the information processing in natural organisms exist. The architecture at first exploits the ability of its subsystems to adapt their resource requirements to optimize its performance with respect to the usage of available computational resources. Secondly, the architecture achieves, inspired by the acute stress response of a natural being, that in the case of an emergency it makes all recources available to address a given threat in a self-coordinated manner.

The Self Distributing Virtual Machine (SDVM) is a middleware concept to form a parallel computing machine consisting of a any set of processing units, such as functional units in a processor or FPGA, processing units in a multiprocessor chip, or computers in a computer cluster. Its structure and functionality is biologically inspired aiming towards forming a combined workforce of independent units (“sites”), each acting on the same set of simple rules.

The SDVM supports growing and shrinking the cluster at runtime as well as heterogeneous clusters. It uses the work-stealing principle to dynamically distribute the workload among all sites. The SDVM’s energy management targets the health of all sites by adjusting their power states according to workload and temperature. Dynamic reassignment of the current workload facilitates a new energy policy which focuses on increasing the reliability of each site.

This paper presents the structure and the functionality of the SDVM.

This paper presents a tool, one component of the Teleworkbench system, for analyzing experiments in multi-robotics. The proposed tool combines the video taken by a web cam monitoring the field where the experiment runs and some computer generated visual objects representing important events and information as well as robots’ behavior into one interactive video based on MPEG-4 standard. Visualization and data summarization enables the developer to quickly grasp a situation, whereas the possibility of scrolling through the video and selectively activating information helps him analyzing interesting events in depth. Because of the MPEG-4 standard used for the output video, the analysis process can be done in a wide range of platforms. This trait is beneficial for education and research cooperation purposes.

We present a socially inspired approach that allows agents in Multi-Agent Systems to speed up their own learning process through communication. Thereby, they are able to trade off impact of knowledge by mutation dependent on the recent performance of the interacting agents. This is inspired by social interaction of humans, where the opinions of experts have greater impact on the overall opinion and are incorporated more exactly than those of newbies. The approach is successfully evaluated in a simulation in which mobile robots have to accomplish a task while taking care of timely recharging their resources.

Despite the advances in wireless, energy-constrained ad hoc networks, there are still many challenges given the limited capabilities of the current hardware. Therefore, our aim is to develop a lightweight, yet powerful operating system (OS) for these networks. We reject the brute force method of provisioning all necessary OS services at each node of the system. Instead, our approach aims to distribute the set of requested OS services over the network to reduce and balance load, improve quality of service, increase fairness and predictability. To limit the burden imposed on the network by the service distribution mechanism, only a subset of nodes, the coordinators, chosen by an underlying state-of-the-art topology control, are concerned with this task. Coordinators observe the state of nodes and OS services within their one-hop vicinity, i.e. their decision area, incorporating different aspects, such as energy, utilisation, or available resources in their decisions. Although each coordinator acquires information and triggers migrations of service states only locally within its decision area, a global-level result emerges, as decision areas naturally overlap. In this manner, an increased amount of work load e.g. in one decision area “floats” to the surrounding decision areas attracted by better conditions. In ns-2 simulations we demonstrate that the mechanism of emergence, which produces many fascinating results in natural systems, can successfully be applied in artificial systems to considerably increase the efficiency and quality of OS service distribution.