Decoding of Repetition Codes for Transmissions of Compressed Data

Abstract

This paper proposes a modification on the existing coding scheme of Repetition codes for compressed data. We introduce a new variable taking the ratio of probability of ”0” and ”1” appearance of the compressed source, which is usually omitted by assuming that the source has equiprobable occurence of ”0” and ”1”. However, in this paper we observe that the omitting of the statistic of the source is inappropriate for compressed data. We use both practical compressed source of Indonesian text and artificial text having an extreme appearance probability ”0” and ”1” as the baseline compression. This new variable is expected to improve the error capability once the source statistic is known and do not change during the transmissions. We evaluate the performances of the proposed decoding in terms of bit-error-rate (BER) performances under the additive white Gaussian noise (AWGN), block Rayleigh fading, and fast Rayleigh fading channels using a series of computer simulations. The results show that the proposed modified Repetition codes provide better BER performances under all channel models, where the improvement is linear to value. The results are expected to make contributions to the development of transmitting compressed sources.