Abstract

A novel joint source and channel coding (JSCC) scheme is proposed, which we refer to as the reordered Elias gamma error correction (REGEC) code. Like the recently proposed unary error correction (UEC) code and EGEC code, the proposed code facilitates the practical near-capacity transmission of source symbol values that are randomly selected from a large or infinite alphabet. However, in contrast to the UEC code, both the EGEC and our proposed REGEC codes are universal codes, facilitating the transmission of source symbol values that are randomly selected using any monotonic probability distribution. However, the EGEC code has a complicated structure comprising two parts, where unequal error protection is required to balance the two parts with the aid of a specific parameterization that must be tailored to the source distribution, preventing its employment for unknown or non-stationary sources. By contrast, the proposed REGEC code does not need unequal error protection, and hence its parameterization does not have to be tailored to the particular source distribution, and thus the REGEC code is a more attractive scheme. More explicitly, our REGEC code has a simple structure comprising only a single part, which does not suffer from the delay and loss of synchronization that are associated with the two parts of the EGEC code. In a particular practical scenario, where the source symbols obey a specific Zeta probability distribution, our REGEC scheme is shown to offer gains of up to 0.9 dB over the best of JSCC and separate source and channel coding (SSCC) benchmarkers, when QPSK modulation is employed for transmission over an uncorrelated narrowband Rayleigh fading channel. In the scenario where the source symbols obey the distribution produced by the H.265 video codec, our REGEC scheme is shown to offer a gain of 0.7 dB over the SSCC benchmarker. These gains are achieved for free, without increasing the required transmit-duration, transmit-bandwidth, transmit-energy, or decoding complexity.

Highlights

  • Multimedia codecs such as H.264 [1] and H.265 [2] typically produce symbols that have a wide range of values

  • We previously proposed a pair of Joint Source and Channel Coding (JSCC) schemes for the near-capacity transmission of source symbols that are randomly selected from a large alphabet, namely the Unary Error Correction (UEC) code [3] and the Elias Gamma Error Correction (EGEC) code [43]

  • This paper proposes a universal JSCC scheme, which we refer to as the Reordered Elias Gamma Error Correction (REGEC) code

Read more

Summary

INTRODUCTION

Multimedia codecs such as H.264 [1] and H.265 [2] typically produce symbols that have a wide range of values. If the actual source distribution is unknown or it is non-stationary, it will typically fail to match the distributions, causing further capacity loss Against this background, this paper proposes a universal JSCC scheme, which we refer to as the Reordered Elias Gamma Error Correction (REGEC) code. This paper proposes a universal JSCC scheme, which we refer to as the Reordered Elias Gamma Error Correction (REGEC) code This has a simple structure, which facilitates the near-capacity transmission of symbol values that are randomly selected from large alphabets using any arbitrary monotonic probability distribution at a low complexity. Since it is a universal code, the applicability of the REGEC code is not limited to any particular source symbol distribution like the UEC.

SYMBOL VALUE SETS HAVING A LARGE CARDINALITY
REORDERED ELIAS GAMMA CODE
REORDERED ELIAS GAMMA ERROR CORRECTION ENCODER
Reordered Elias Gamma encoder
Reordered Elias Gamma Error Correction trellis encoder
Integration of the REGEC encoder into a transmitter
REORDERED ELIAS GAMMA ERROR CORRECTION
Integration of Reordered Elias Gamma Error Correction decoder into a receiver
Reordered Elias Gamma Error Correction trellis decoder
Reordered Elias Gamma decoder
PARAMETRIZATION OF THE REORDERED ELIAS GAMMA ERROR CORRECTION CODE
Reordered Elias Gamma Error Correction codebook extension
Performance analysis
REGEC codebook candidate selection
EXIT charts of the REGEC candidate codebooks and the best matching URCs
REGEC C8
Error floor analysis
PERFORMANCE COMPARISON WITH THE BENCHMARKERS
Parametrization
SER comparison with the benchmarkers
VIII. CONCLUSIONS
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call