Lancaster EPrints

Genetic Programming and Protocol Configuration

Grace, P. (2000) Genetic Programming and Protocol Configuration. Masters thesis, .

[img]
Preview
PDF (msc.pdf)
Download (324Kb) | Preview

    Abstract

    Dynamic protocol stacks have been identified as a method of improving the performance of process communication by constructing the protocol that meets it requirements with a minimum overhead. However, the design of these stacks is a complex process for human designers who must identify the correct elements and ordering. Genetic programming is a machine learning technique method for generating computer programs automatically. It has been applied to many areas including electronic circuit design, image classification and machine code creation. However, it has never been utilised in the area of communication protocols. The main aim of the project is to identify if genetic programming techniques can be applied to protocol construction using the JavaGroups toolkit and generate protocols automatically. This report describes the theory and application of genetic programming principles and also the technology behind and creation of dynamic protocols. There is also a description of the design and implementation of a genetic program that will create JavaGroups protocols to meet a set of requirements. The results from the experimentation of the implemented system identify that GP can automatically generate the correct protocol stack for a required communication type (e.g. reliable point-to-point). The uses of this method to generate any stack depending on the user’s requirement or within a changing real-time network situation are identified as future areas of work.

    Item Type: Thesis (Masters)
    Uncontrolled Keywords: cs_eprint_id ; 1712 cs_uid ; 361
    Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
    Departments: Faculty of Science and Technology > School of Computing & Communications
    ID Code: 41570
    Deposited By: ep_importer_comp
    Deposited On: 11 Aug 2008 14:57
    Refereed?: No
    Published?: Unpublished
    Last Modified: 27 Jul 2012 02:40
    Identification Number:
    URI: http://eprints.lancs.ac.uk/id/eprint/41570

    Actions (login required)

    View Item