Transmission of Gaussian Source Using Ternary Convolutional Low Density Generator Matrix Codes

Abstract

In this paper, we propose a new joint source-channel coding (JSCC) scheme to transmit a Gaussian source over an additive white Gaussian noise (AWGN) channel, where the channel input constraint is on energy per source symbol rather than per channel symbol. The JSCC scheme is based on an embedded Lloyd-Max quantizer with adjustable quantization levels and the ternary convolutional low density generator matrix (LDGM) codes, which are easily configurable in the sense that any rational code rate can be achieved without complicated optimization. In our scheme, first, a Gaussian source is quantized into ternary symbols, which can be formatted into different symbol planes according to their contribution to the distortion. Second, the symbol planes are encoded by a ternary convolutional LDGM code, where the relatively important symbol plane is encoded with a lower code rate. Third, the coded sequence is modulated in 3-PAM and then transmitted over the channel. Numerical results show that the proposed scheme performs close to the Shannon limit, and that the distortion can be lowered down by increasing the number of quantization levels.