Optimal Hop-By-Hop Flow Control Policies with Multiple Transmitters

We formulate the flow control between one receiving node and its adjacent transmitting nodes in a computer network as a Markovian decision process. The objective is to minimize the total number of messages waiting for transmission subject to the evolution of the state and to the buffer constraints of the receiving node. The control action consists of an allocation of time slots among the transmitting nodes. We derive several structural properties which are useful in characterizing the optimal policy.