• Convolutional Versus LDPC and Turbo Codes on the Rayleigh Fading Channel

      Ryan, William E.; Marcellin, Michael W.; Jagiello, Kristin; Cooper, Charlie; University of Arizona (International Foundation for Telemetering, 2009-10)
      We consider the performance of low-density parity-check (LDPC) codes, turbo codes and convolutional codes over the binary-input AWGN channel with flat Rayleigh fading. LDPC and turbo codes are capacity-approaching codes for long codewords. For short and medium codewords we seek to determine if they still outperform the industry-standard memory-6, rate-1/2 convolutional code. For a fixed SNR, the probability of error for the codes of interest are plotted as a function of codelength. We find that for very short codewords, the convolutional code performs best.
    • Isually Lossless Coding for Color Aerial Images Using PEG

      Marcellin, Michael W.; Bilgin, Ali; Oh, Han; Kim, Yookyung; University of Arizona (International Foundation for Telemetering, 2009-10)
      This paper describes a psychophysical experiment to measure visibility thresholds (VT) for quantization distortion in JPEG2000 and an associated quantization algorithm for visually lossless coding of color aerial images. The visibility thresholds are obtained from a quantization distortion model based on the statistical characteristics of wavelet coefficients and the deadzone quantizer of JPEG2000, and the resulting visibility thresholds are presented for the luminance component (Y) and two chrominance components (Cb and Cr). Using the thresholds, we have achieved visually lossless coding for 24-bit color aerial images at an average bitrate of 4.17 bits/pixels, which is approximately 30% of the bitrate required for numerically lossless coding.
    • Low-Density Parity-Check Codes Which Can Correct Three Errors Under Iterative Decoding

      Vasic, Bane; Marcellin, Michael W.; Krishnan, Anantha Raman; Chilappagari, Shashi Kiran; University of Arizona (International Foundation for Telemetering, 2009-10)
      In this paper, we give necessary and sufficient conditions for low-density parity-check (LDPC) codes with column-weight four to correct three errors when decoded using hard-decision message-passing decoding. We then give a construction technique which results in codes satisfying these conditions. We also provide numerical assessment of code performance via simulation results.