Interacting random access communication networks

In this dissertation, we study random access systems. The first part deals with interacting groups of users. In the second part, we study the effect of capture. The goal is to devise and analyze protocols that are suitable for such environments. We repeatedly find that the use of the ALOHA protocol leads to instabilities of operation and that the use of an appropriate retransmission control algorithms leads to stable operation. The first model studied consists of two groups of users, independently transmitting packetized messages over two broadcast channels that interact. The interaction occurs when the number of transmissions in any group exceeds a threshold. The system is shown to be unstable if ALOHA type protocols are employed. Therefore a two-dimensional retransmission control procedure is employed and analyzed. We next study a random access system that consists of two classes of users with different priorities. The two-dimensional Markov chain with the number of backlogged users as states is shown to be non-ergodic. Hence the throughput in steady state is zero. In a capture channel, differences in the levels of received power and time of arrivals enable the receiver to successfully receive a packet even when two or more have been transmitted simultaneously. Several authors have already studied this capture channel based on the statistical equilibrium approximation. We avoid this approximation and prove that the infinite population model with capture is unstable for a large class of capture statistics. As in the first part, we stabilize the system and derive the channel capacity as a function of capture parameter. The performance improvement, compared to the conventional ALOHA systems, is noticeable even when the capture probability is low. We extend the capture model to communication systems that employ Carrier Sensing as well Code Division Multiplexing techniques. We show that neither the possibility of capture nor the ability to sense the channel is enough to stabilize the system. We stabilize the system by employing a retransmission control algorithm and analyze the performance. We find that as long as the propagation delay is fairly large the retransmission control system performs better than the traditional CSMA system (52). Finally, we study the statistics of capture as it arises in radio communication and spread spectrum multiaccess systems. We extensively analyze the Nakagami fading model since it describes a wide range of fading statistics; Rayleigh, Rician, half-Gaussian and lognormal, etc. Computational results are included for a broad range of system parameter. (Abstract shortened with permission of author.)