Semantic Blueprints of Discrete Dynamic Systems: Challenges and Needs in Computational Modeling of Complex Behavior

How can one cope with the notorious problem of establishing the correctness and completeness of abstract functional requirements in the design of control-intensive software systems prior to actually building a system? The answer given here explores /abstract state machines/ (ASMs): a universal mathematical framework for semantic modeling of discrete dynamic systems. Combining common abstraction principles from mathematical and computational logic, ASMs provide a universal model of computation and an effective instrument for analyzing and reasoning about complex semantic properties of real-world systems. Widely recognized applications include semantic foundations of virtually all kinds of architectures, languages and protocols.

In this talk we focus on concurrent and reactive systems in automotive control, e-business, and advanced telecommunication services, and more recent work in computational criminology, safety and security.

Mapping of Nested Loop Programs onto Massively Parallel Processor Arrays with Memory and I/O constraints

Speaker: Prof. Dr. Jürgen Teich

Conventionally, special processors so called DSPs are used in the domain of digital signal processing in order to tackle high demands for performance, monetary and energy costs. Computationally intensive applications like video and image processing in consumer electronics and the rapidly evolving market of mobile and personal digital devices are the driving forces in this market and are greedy for new sophisticated architectures. Thanks to the steadily advances in semiconductor industry modern fabrication techniques allow today the implementing arrays of hundreds of 32-bit microprocessors on a single die. Only very few compilation techniques for such massively parallel Systems-on-a-Chip exist. These techniques are closely related to approaches from the DSP world, but they are not able to exploit the full parallelism of a given algorithm and the computational potential of a typical 2-dimensional processor array.

In this talk, we present our design methodology for mapping nested loop programs onto massively parallel arrays which is characterized by loop parallelization in the polytope model. We show how to adapt this methodology in order to target a given fixed architecture, i.e. the matching of architectural constraints with the parameters of mapping transformations is discussed. Here, hierarchical partitioning techniques are used in order to cope with the given memory and I/O constraints. Therefore, memory estimation formulas based on the tiling parameters are presented.

Using distributed systems for an efficient real-time control of mixed-model assembly lines

Speaker: Dr. Stefan Bock

In this presentation, a new real-time oriented control system for mass customized manufacturing processes is depicted. Originally, assembly lines were designed for the mass production of a single homogenous product. For those constellations, these systems yield substantial reductions of the variable production costs. But due to the increasing competition and the differentiated demands today, it is no longer sufficient to offer only standardized products. In order to combine the wishes and requirements of the customers with the advantages of an efficient flow line production, a large number of different product-variants are simultaneously produced on the same mixed-model assembly line. Therefore, the control of such kind of production processes becomes a complex task since an efficient production execution has to smooth the occurring oscillating work content of the different variants. Additionally, unforeseen disturbances of the production process may jeopardize its planned execution. In order to deal with these problems, the paper on hand proposes a new approach for an adaptive real-time control of assembly lines. It substantially extends the pure sequencing problem by the integration of specific line-balancing aspects and the mapping of consequences of possible disturbance scenarios. Such a scenario could be for instance the drop out of a worker, a material bottleneck or a machine breakdown. In order to guarantee in such a scenario an efficient continuation of the production process, the proposed controlling instrument reacts instantly on each occurring disturbance. By adapting the current plan in a few cycle times and in simultaneity to its execution, this approach was able to reduce the additional costs caused by the occurred disturbances substantially. Specifically, in several scenarios it was possible to avoid on the average up to 90 percent of these costs. For these fast plan adaptations conducted under real-time restrictions, specifically designed distributed optimization algorithms were applied. Their efficiency was validated by the analysis of several practical experiments executed in ordinary Local Area Networks of Personal Computers.