• DocumentCode
    1991996
  • Title

    Adaptive Multi-Code Chaos-Based CDMA with Unequal Error Protection Using Reduced Support Vector Machine

  • Author

    Kao, Johnny W H ; Berber, Stevan M. ; Kecman, Vojislav

  • Author_Institution
    Dept. of Electr. & Comput. Eng., Univ. of Auckland, Auckland, New Zealand
  • fYear
    2010
  • fDate
    6-10 Dec. 2010
  • Firstpage
    1
  • Lastpage
    5
  • Abstract
    In this paper, we propose a support vector machine receiver for a multi-code Chaos-based CDMA (MC-CB-CDMA) system. Unlike the conventional multi-code CDMA, the main aim for the assigning multiple spreading codes for each user is to provide unequal error protection on the transmitted data without using further error correction coding (ECC). In addition, the proposed SVM receiver is further simplified by the recursive feature elimination algorithm so that it has a less complexity than the conventional correlator receiver. The modified receiver is also very flexible in controlling the level of error protection in the system. Simulation results are compared with correlator receiver and huge performance gain is achieved due to the multi-user detection capability from the SVM receiver. Theoretical probability of error for both the correlator and the SVM receiver have been derived and confirmed with the simulation results.
  • Keywords
    code division multiple access; error correction codes; support vector machines; MC-CB-CDMA; SVM receiver; adaptive multicode chaos; multiple spreading codes; multiuser detection capability; recursive feature elimination algorithm; reduced support vector machine; support vector machine receiver; unequal error protection; Bit error rate; Complexity theory; Correlators; Fading; Multiaccess communication; Receivers; Support vector machines;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Global Telecommunications Conference (GLOBECOM 2010), 2010 IEEE
  • Conference_Location
    Miami, FL
  • ISSN
    1930-529X
  • Print_ISBN
    978-1-4244-5636-9
  • Electronic_ISBN
    1930-529X
  • Type

    conf

  • DOI
    10.1109/GLOCOM.2010.5683673
  • Filename
    5683673