Distributed multiple-access and the impact of feedback on communication systems

Two fundamental issues in communication systems are investigated in this thesis: (1) how to improve individual user's performance in a multi-user system using a distributed mechanism, and (2) what is the overall impact of feedback links on a practical communication system. We apply fundamental techniques in both information theory and queueing theory to gain insights into the structure of these problems.;Interference from multiple users arises naturally as different users in the same region attempt to transmit to the base station at the same time. The maximum transmission rate for each user is derived in the classic information theory literature. Numerous attempts have been made in the past several decades so that the users' transmission rates get closer and closer to the theoretic maximum. However, some of the standard approaches are suboptimal in terms of achievable rates. Other approaches can achieve the maximum rates, but they can not be implemented easily in practice. In the first part of this thesis, we propose a scheme called Distributed Rate Splitting to asymptotically achieve the optimal rates allowed by information theory in a practical manner.;Another fundamental problem which has not been investigated thoroughly by information theorists is the impact of feedback on a communication system, since feedback typically induces retransmission which leads to additional burstiness in the transmission buffer. Hence, a practical design of communication systems with or without feedback must take into account all system parameters such as transmission power, decoding error probability and packet delay. In the second part of the thesis, we assess the benefits using feedback from an integrated information theoretic and queueing theoretic perspective.