| Summary: | In this Thesis we consider high-throughput rate compatible Joint Source-Channel Coding (JSCC) schemes based on Rate Compatible Modulation (RCM) codes. These codes achieve JSCC capabilities by embedding source compression into modulation through the generation of multi-level symbols from weighted linear combinations of the input bits. The smooth rate adaptation is achieved seamlessly by varying the number of generated symbols. These two properties make them advantageous over conventional Adaptive Coded Modulation (ACM) techniques, which usually rely on unrealistic instant and accurate channel estimations, and a limited set of coding and modulation combinations to choose from. The main drawback of RCM codes is that they experience performance degradation due to the presence of error floors at high Signal-to-Noise Ratios (SNRs). These error floors can be substantially improved by substituting a few RCM symbols by LDGM coded bits, forming an hybrid coding scheme in which the LDGM symbols correct residual errors produced by the RCM.
This work investigates new applications and design techniques of these family of codes for point-to-point and multi-user communications. For the point-to-point case, we propose an EXIT chart analysis and a bit error rate prediction procedure suitable for implementing RCM-LDGM codes. The developed EXIT charts speed up the design method of good codes, which otherwise requires the use of time-consuming simulations. We continue by considering the problem of implementing high-throughput JSCC schemes for the transmission of binary sources with memory over AWGN channels, for which we propose a coding scheme that makes use of the Burrows-Wheeler Transform (BWT) and the rate compatible RCM-LDGM codes. Finally, for the first time in the literature, we propose the use of RCM-LDGM codes for additive impulsive noise channels.
For multi-user communications we begin by considering that the information sources are uncorrelated and propose a new coding scheme based on the use of an irregular RCM encoder for each user. By properly designing the encoders and taking advantage of the additive nature of the MAC, the proposed scheme allows the simultaneous transmission of a large number of uncorrelated users at high rates, while the decoding complexity is the same as that of standard point-to-point RCM schemes. In the last part of the Thesis, we tackle the multi-user communication scenario in which the transmitted information sources are spatially correlated and also extend the use of LDGM codes in parallel with the proposed RCM systems in the MAC.
|
|---|