Hardware-Software Partitioning at the Knowledge Level

2000

The synthesis of mixed analog/digital systems is a relatively complex task because of the heterogeneity of these systems. This paper describes how major principles of expert systems for construction tasks can be used to deal with the design space of mixed analog/digital systems at the system level: 1) conception-hierarchies can represent the system level in an elegant way, which views the partitioning problem as a well-de ned task; 2) constraint nets support the distribution of limited design resources. Both together with design methodologies for the subblocks and strategies allow the construction of such systems. This is su cient to make important design decisions at system level, like the determination of sample frequencies and bit widths, for example.

Computación Y Sistemas, 2013

Hardware/Software partitioning (HSP) is a key task for embedded system co-design. The main goal of this task is to decide which components of an application are to be executed in a general purpose processor (software) and which ones, on a specific hardware, taking into account a set of restrictions expressed by metrics. In last years, several approaches have been proposed for solving the HSP problem, directed by metaheuristic algorithms. However, due to diversity of models and metrics used, the choice of the best suited algorithm is an open problem yet. This article presents the results of applying a fuzzy approach to the HSP problem. This approach is more flexible than many others due to the fact that it is possible to accept quite good solutions or to reject other ones which do not seem good. In this work we compare six metaheuristic algorithms: Random Search, Tabu Search, Simulated Annealing, Hill Climbing, Genetic Algorithm and Evolutionary Strategy. The presented model is aimed to simultaneously minimize the hardware area and the execution time. The obtained results show that Restart Hill Climbing is the best performing algorithm in most cases.

IEEE Transactions on Systems, Man, and Cybernetics, 1997

This paper presents the knowledge representation schemes adopted in MICKEY, a knowledge based system for designing microprocessor based systems. MICKEY is essentially a hybrid expert system, using rules and procedures for achieving the different design tasks. We briefly describe the hierarchy of tasks in this problem domain, and emphasize on the refinement paradigm, constraint propagation, conflict resolution and task management strategies adopted in MICKEY. Next, we dwell upon the different knowledge sources and their functions, with respect to the particular design domain. Finally, we present an industrial design, achieved by MICKEY, to demonstrate its applicability.

2006

The use of Application Specific Instruction-set Processors (ASIP) is a solution to the problem of increasing complexity in embedded systems design. One of the major challenges in ASIP design is Design Space Exploration (DSE), because of the heterogeneity of the objectives and parameters involved. Typically DSE is a multiobjective search problem, where performance, power, area, etc. are the different optimization criteria. The output of a DSE strategy is a set of candidate design solutions called a Pareto-optimal set. Choosing a solution for system implementation from the Paretooptimal set can be a difficult task, generally because Pareto-optimal sets can be extremely large or even contain an infinite number of solutions. In this paper we propose a methodology to assist the decision-maker in analysis of the solutions to multi-objective problems. By means of fuzzy clustering techniques, it finds the reduced Pareto subset, which best represents all the Pareto solutions. This optimal subset will be used for further and more accurate (but slower) analysis. As a real application example we address the optimization of area, performance, and power of a VLIW-based embedded system.

ACM Transactions on Design Automation of Electronic Systems, 2003

This paper presents an in-depth study of several system partitioning procedures. It is based on the appropriate formulation of a general system model, being therefore independent of either the particular co-design problem or the specific partitioning procedure. The techniques under study are a knowledge-based system and three classical circuit partitioning algorithms (Simulated Annealing, Kernighan&Lin and Hierarchical Clustering). The former has been entirely proposed by the authors in previous works while the later have been properly extended to deal with system level issues. We will show how the way the problem is solved biases the results obtained, regarding both quality and convergence rate. Consequently it is extremely important to choose the most suitable technique for the particular co-design problem that is being confronted.

HW/SW techniques make it possible for the system designers to validate their design, assign modules to be implemented in either hardware or software in the early stages of the system design life cycle. In addition, those techniques provide powerful mechanism for continuous system validation until the final product is done. Partitioning the system into either hardware or software, in the system early stages, is vital decision that has to be done iteratively and accurately. Many techniques have been proposed for HW/SW partitioning: conventional circuit partitioning techniques, simulated annealing, expert systems, and even genetic algorithm techniques. The partitioning problem has been proved to be and NP-Hard problem, thus AI, ANN and GA techniques can find a rich playground to apply their techniques. This paper presents a novel approach to use Bayesian Belief Networks as the tool that does the partitioning decision when provided by simulation parameters that measure certain character...

IEEE Transactions on Industrial Electronics, 1994

The widespread use of microprocessors in industrial applications such as process control, data logging, monitoring, etc., demand that the design of such systems be automated. Algorithmic methods are inadequate for this task, as knowledge from several sources need to be combined to produce the resulting design. In this paper we present a knowledge-based approach to the design of such systems, which includes the design of the hardware configuration as well as the application software. The knowledge requirements and the functional modules of the design task are elicited, and practical designs are demonstrated.

Coma 13, 2013

Knowledge-Based Engineering (KBE) is a smart approach used in product development in order to shorten the duration of the engineering design phase. It consists of using computational intelligence to capture the design rules of a product family in order to generate several design variants from a single generative model. The KBE main objective is the automation of routine tasks that constitute an important part of the design phase. Such tasks can be performed several times as part of the design customization and optimization process. The use of KBE during the last three decades resulted in a significant reduction of the design duration in some companies. However, despite the impressive results obtained through the use of KBE in the automotive and aeronautical industry, there are still very few companies that make use of this approach. The review of relevant literature showed that the lack of an effective methodology of implementation is one of the major stumbling blocks to the expansion of KBE. Current methodologies do not seem to propose an efficient method for knowledge processing which is a very important phase of the implementation of the KBE approach. This paper discusses a detailed qualitative method that addresses the issue of knowledge processing in KBE. Based on the system engineering approach and a logical classification of the design information, this method enables the efficient capture and presentation of design knowledge. The strength of this method lies in its ability to represent design knowledge in a form that makes it understandable to both engineers and programmers. This suitable representation has the potential shortens the duration of the knowledge processing and facilitates the knowledge encoding phase. A practical example is also presented to illustrate implementation of the suggested method and to show its advantage over other methods.

Fuzzy Sets and Systems, 1998

Much of engineering design can be characterised as putting together variants of existing mechanisms to meet novel requirements. This paper describes an approach to engineering design in which fuzzy sets are used to represent the range of variants on existing mechanisms. Membership functions are chosen to reproduce the distinctions and classifications used by experienced designers. Methods are introduced to calculate the fuzzy performance range achievable by each component type, and a metric is suggested for the ranking of design candidates against design requirements. The underlying approach to design evaluation derives from that developed by Antonsson, Wood and Otto. If the components model the design domain accurately, this architecture will always find a successful design, where one exists. However, finding this design may involve exhaustive search of the design space, and the time required for such an exhaustive search is typically far greater than is available for the task. The architecture is therefore augmented by the introduction of design agents, embodying heuristic rules to direct the search intelligently, so that a satisfactory design is found within a reasonable length of time. As an example, the method is applied to preliminary design of a Stirling engine heat exchanger.

Fuzzy Sets and Systems, 2004

This paper describes hierarchical modeling of fuzzy logic concepts that has been used within the recently developed model of intelligent systems, called OBOA. The model is based on a multilevel, hierarchical, general object-oriented approach. Current methods and software design and development tools for intelligent systems are usually di cult to extend, and it is not easy to reuse their components in developing intelligent systems. The OBOA model tries to reduce these deÿciencies. The model starts with a well-founded software engineering principle, making clear distinction between generic, low-level intelligent software components, and domain-dependent, high-level components of an intelligent system. This paper concentrates on modeling and implementation of fuzzy logic concepts within the hierarchical levels of the OBOA model. The fuzzy components described are extensible and adjustable. As an illustration of how these components are used in practice, a practical design example from the domain of medical diagnosis is shown. The paper also suggests some steps towards future design of fuzzy components and tools for intelligent systems.