Abstract
Efficient design on multiple access is fundamental in networkingenvironment where resource (bandwidth, service time, etc.) must be shared by numbers of users. Since 1970's, extensive research effort has been put on the development of effective multiple access schemes to exploit the ultimate efficiency achievable by practical solutions. Those results were then categorized into deterministic (e.g., time-division multiple access (TDMA) and frequency-division multiple access (FDMA)) and stochastic methods, and the most well-known protocols of the latter include ALOHA, tree and CSMA (carrier sensing multiple access).A general modeling technique to achieve comprehensive knowledge on multiple access analytically should be much more essential than new protocol proposition given such a long-time dedication on certain field of research. We therefore propose MULCAR (multi-layer collision avoidance/resolution) as such a tool for random access in centralized networks (either infrastructural or Ad-hoc), which highlights the core mechanism of modern networking. The development of this modeling is begun by understanding the spirit of efficient random access and thefeasibility of implementation techniques, enriched by the approval of earlier recognized protocols, and completed by an analytic module addressing all alternatives of protocol structure and channel statistics. The effect of some basic parameters for this general modeling is also investigated with this analytic module.Furthermore, we propose the γ-VO algorithm as a systematicmethodology for optimization with constraints. It is appropriate to apply this algorithm on the general analytic model of MULCAR to optimize the performance of a network, with either global or diverse objectives. We demonstrate its efficiency on power-efficient (global objective) as well as integrated-traffic (diverse objectives) control protocol design with both analysis and simulations. The contributionof the MULCAR modeling in theory and practical protocol design is hence substantiated.