Capacity sharing in an integrated voice/data packet-switched multiplexer: optimal and sequential solutions

The authors consider an integrated packet-switched system performing voice/data multiplexing with a fixed frame duration. The optimal allocation of the channel capacity is formulated as a Markovian decision process problem with a long-run average cost to minimize subject to a hard constraint. A Lagrange multiplier method was used to develop an algorithm, based on dynamic programming, for the computation of the optimal policy. The authors also propose an easily implementable but suboptimal policy and compare the performances of these policies by simulation. Numerical results show the feasibility and the reliability of the dynamic programming procedure.<>