Capacity, Error Exponent, and Structural Results for Communication Networks

Abstract

In various multi-terminal communication scenarios, contrary to point-to-point communication, characterization of fundamental limits such as capacity and error exponent is still an open problem. We study such fundamental limits and the structure of optimality achieving codes. This thesis consists of two parts: in the first part, we investigate the role of algebraic structures in multi-terminal communications. We show the necessity of various types of algebraic structure in capacity achieving codes and argue that the lack of such structures in the conventional random codes leads to their sub-optimality. We develop a new class of partially structured codes called quasi-structured code (QSC). Such codes span the spectrum from completely structured to completely unstructured codes. It is shown that the application of QSCs leads to improvements over the current coding strategies for many problems including distributed source coding and multiple-access channel (MAC) with feedback.

In the second part of the thesis, we study the optimal error exponent in various multi-terminal communication scenarios. We derive a lower and upper bound on the error exponent of discrete memoryless MAC with noiseless feedback and variable-length codes (VLCs). The bounds increase linearly with respect to a specific Euclidean distance measure defined between the transmission rate pair and the capacity boundary. The bounds are shown to be tight for specific classes of MACs.