A stochastic control approach to Slotted-ALOHA random access protocol

Abstract

ALOHA random access protocols are distributed protocols based on transmission probabilities, that is, each node decides upon packet transmissions according to a transmission probability value. In the literature, ALOHA protocols are analysed by giving necessary and sufficient conditions for the stability of the queues of the node buffers under a control vector (whose elements are the transmission probabilities assigned to the nodes), given an arrival rate vector (whose elements represent the rates of the packets arriving in the node buffers). The innovation of this work is that, given an arrival rate vector, it computes the optimal control vector by defining and solving a stochastic control problem aimed at maximising the overall transmission efficiency, while keeping a grade of fairness among the nodes. Furthermore, a more general case in which the arrival rate vector changes in time is considered. The increased efficiency of the proposed solution with respect to the standard ALOHA approach is evaluated by means of numerical simulations.

Keywords:

• Markov decision process
• stochastic control
• Slotted-ALOHA protocol

Acknowledgements

The work is partially based on research activities carried out in the framework of the EmerSat project, funded by the Italian Space Agency (ASI). The author wishes to thank Dr. Andrea Fiaschetti for his fundamental help in conceiving this work.

Notes

This is a common assumption in ALOHA protocol analysis; however, as suggested in Section 6, the proposed approach may be extended to the case of generic traffic model by means of model-free techniques.

For the sake of simplicity, it is assumed that each node buffer size is equal to B; the extension to the general case is straightforward by defining N different buffer sizes B_n, n = 1, 2, … , N.

Note that the algorithm, which computes the transmission probabilities is still centralised.