Full metadata record
DC FieldValueLanguage
dc.contributorMulti-disciplinary Studiesen_US
dc.contributorDepartment of Electronic and Information Engineeringen_US
dc.creatorLee, Hok-leung Philip-
dc.publisherHong Kong Polytechnic University-
dc.rightsAll rights reserveden_US
dc.titleNeural network based fair queuing scheduler for video sourcesen_US
dcterms.abstractAs discussed in Reljin's paper [8], the neural networks can be very useful in the bursty traffic case. After that, there is no similar research on the performance of the Neural Network (NN) on other traffic sources. Therefore the dissertation investigates the cell loss rate of the NN based fair queuing (FQ) scheduler on the video data stream and its performance and compares it with the traditional round robbin (RR) scheduler. In the computer simulation, three types of the video data files are modeled, videoconference, videophone and CATV but only videoconference and videophone data files are used in the computer simulation because the bandwidth for the CATV is quite large (mean: 795Mbps and peak: 1.5Gbps) without compression. The simulation is based on the Autoregressive AR(1) model with Gamma random process characteristic and the scheduling controller is a competitive type neural network (NN) which is known as Kohonen learning. This type of NN is able to adapt to the changes of the input traffic. The training data for the Kohonen layer is assumed to consist of a sequence of input vectors x, drawn at random in accordance with a fixed probability density function p. As each of these vectors enter into the network, the Kohonen processing elements compete one another to determine the winner on the basic of minimum distance. This type of neuron is self-learning but not pre-trained as usual type. Therefore it is suitable for the data processing with no specific pattern. The FQ scheduler aggregates the input queue according to its behavior. Once any input queue gets busy, the NN temporarily takes control of the aggregation process to reduce the cell loss. Computer simulation shows that the proposed NN-based scheduler has a better performance specially in slow motion video sources such as videoconference and the performance remains well behaved even when the successively serving cell size is changed. The cell loss rate for a NN based scheduler is at least one order of magnitude better than that of an RR scheduler on light and half load condition. In full load condition or fast motion video, the NN based scheduler still provides some improvement over an RR scheduler.en_US
dcterms.extentviii, 82 leaves : ill. ; 31 cmen_US
dcterms.isPartOfPolyU Electronic Thesesen_US
dcterms.educationalLevelAll Masteren_US
dcterms.LCSHNeural networks (Computer science)en_US
dcterms.LCSHHong Kong Polytechnic University -- Dissertationsen_US
dcterms.accessRightsrestricted accessen_US

Files in This Item:
File Description SizeFormat 
b15177622.pdfFor All Users (off-campus access for PolyU Staff & Students only)2.46 MBAdobe PDFView/Open

Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show simple item record

Please use this identifier to cite or link to this item: https://theses.lib.polyu.edu.hk/handle/200/4755