E. A. Vyskubova, N. P. Khmyrova

The possibility to create systems for reliable digital data transmission over a radio channel is defined in many respects by the efficiency of the method used for error-correcting coding. The paper considers the choice of the code structure for reliable transmission of digital messages which size can vary in a wide range but for the recipient only error-free information must be outputted with a specified probability. It is assumed that the transmission is performed over additive white Gaussian noise (AWGN) channel and the demodulator forms hard decisions with a specified error probability. It is shown that for large-sized messages composite codes with PC code as an outer code and a convolutional or block code as an inner code would be appropriate for the use. As the convolution code block can not be less than decoding depth that at least five times greater than the code constraint length, short block codes should be used for minimal-sized messages. Among short block codes the Golay code is the best one. The paper presents a procedure for selecting the code structure and method of calculating the error-free decoding probability. Besides the authors have developed the codec for AWGN channel in compliance with the following requirements: -error probability at the demodulator output: 0.02; -transmitted massage size: 1-8000 bytes -error-free decoding probability: 0.99998 The conclusion on the advantage of the coding method adaptation to the transmitted message size is made.