ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

CUDA implementation of a fuzzy collecitve behaviour model

Anže Čuk (2016) CUDA implementation of a fuzzy collecitve behaviour model. EngD thesis.

Download (720Kb)


    This thesis consists of an implementation of a fuzzy model using the graphics card and the CUDA environment. Our fuzzy model consists of entities, which are all the same type. Entity interaction occurs only between neighbouring entities, where the probability of an interaction is inversely proportional to the distance between the neighbouring entities. For this reason an optimisation of the process of gathering neighbouring entities has been made by dividing the simulation area into bins. The algorithm that drives the entity interaction is implemented with multiple kernels, which are described in detail in this thesis. For the purpose of ensuring the correctness of our results, JUnit tests were used along with the jFuzzyLogic library, which serves as a reference fuzzy logic system. For each entity we tested if the choice of the neighbouring entity was valid. We also tested the individual membership functions and the final crisp output. We conclude this thesis by comparing the execution times of the individual implementations and measuring the speed up value of our graphic card implementation.

    Item Type: Thesis (EngD thesis)
    Keywords: fuzzy logic, CUDA, collective behaviour, optimisation
    Number of Pages: 46
    Language of Content: Slovenian
    Mentor / Comentors:
    Name and SurnameIDFunction
    doc. dr. Iztok Lebar Bajec269Mentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=51012&select=(ID=1536883395)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 3306
    Date Deposited: 25 Mar 2016 12:29
    Last Modified: 21 Apr 2016 13:00
    URI: http://eprints.fri.uni-lj.si/id/eprint/3306

    Actions (login required)

    View Item