(A) new decoding scheme for reed-solomon code in TDMB receiver

Abstract

The T-DMB system is a derivation of the DAB system in which the main ser- vice is digital audio with the required BER = $10^{-4}$. The main service in T-DMB is video with the required BER = $10^{-8}$ much better than that of DAB. To satisfy this requirement, a Forward Error Correction (FEC) block is added to obtain the better BER. The FEC block consists of the Reed-Solomon(204,188,8) encoder/ de- coder block and the convolutional interleaver/deinterleaver at the transmitter and the receiver respectively. In this thesis work, a new method is presented to improve the performance of the T-DMB system further by exploiting the capability of the Reed-Solomon(204, 188, 8) code to correct up to 16 erasures. The Reed-Solomon(204, 188, 8) code used in T-DMB system can correct up to 8 errors or up to 16 erasures. In the conventional RS decoder at the T-DMB receiver, it purely dectects and corrects errors only, so it can correct maximum 8 errors. When the number of errors is more than 8 but less than 16, it cannot correct any error. To correct errors in this case, the new method is to try to predict erasures as many as possible in advance before decoding. Based on that the FEC block at the T-DMB receiver follows the convolutional decoder, the characteristic of convolutional decoder that the errors, if existing after the convolutional decoder, are burst errors and the known relationship between symbols after and before convolutional deinterleaver, the prediction can be made. Therefore, it will improve the performance of T-DMB receiver. The simulation results of the new method have shown that 0.5dB - 1dB SNR (Eb/No) is saved to obtain the required BER = $10^{-8}$ compared with the required SNR at the conventional T-DMB receiver.