The traditional divide between information theory and networking is a heavily discussed problem. In many information theoretic formulations, the channel coupling is stochastic while the data source is traditionally infinitely backlogged. In many networking formulations, the data source is a stochastic process while the network is characterized as a group. The main question in this talk is - Can I study a system which incorporates both queuing dynamics and information theoretic encoding and decoding in its framework? This talk establishes that, for certain multiuser systems, such a coupling is indeed possible, in the following two senses:

a. A centralized algorithm exists such that any information theoretic achievable region in a network is also stable; and
b. A decentralized algorithm exists that can again render an information theoretic achievable region for the network stable.