Beta 1

Title Design and Implementation of Analysis and Optimization Tool for Embedded Systems aiming at Increased Interoperability
Author Kordianowski, Adam
Supervisor Pop, Paul (Embedded Systems Engineering, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Master's thesis
Year 2010
Abstract Embedded systems are nowadays present in every area of people lives. Their complexity is constantly increasing and they have tight requirements in terms of performance, energy consumption, reliability, size and cost. Hence, the task of designing embedded systems is getting more difficult. Designers today use complex tools that help them in synthesizing an implementation that fulfills all the requirements. Several tools have to work in conjunction to produce the desired result. In this thesis we evaluate three interoperability approaches: (i) full integration at the source-code level using shared data structures, (ii) tool integration using XML files and (iii) tool integration using object-oriented databases for data exchange. To evaluate the interoperability approaches, we have focused on three design tasks: (i) mapping, (ii) voltage scaling and (iii) schedulability analysis. The first design task is to find the best mapping of functions to the processing elements in the given architecture. Dynamic voltage scaling, which is the second design task, allows to choose appropriately voltage levels for all functions so that the power consumption is minimized without performance or reliability degradation. The third design task, schedulability analysis, is used to determine whether all the functions in the system will meet their desired deadlines. We have designed and implemented an analysis and optimization tool that performs all three design tasks. The optimization is performed using a Simulated Annealing meta-heuristic. The integration of the tool modules implementing the analysis, design transformation and optimization functions has been performed using all three mentioned interoperability approaches. We have discussed both advantages and disadvantages of each considered interoperability approach and have evaluated their impact on the performance of the implementation in terms of runtime and the quality of the produced solution.
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-M.Sc.-2010-60
Original PDF ep10_60_net.pdf (0.61 MB)
Admin Creation date: 2010-09-20    Update date: 2010-09-20    Source: dtu    ID: 267010    Original MXD