A service system of conflict flows formed in a random environment with finite number of states by a time-sharing algorithm with readjustments is considered. Primary customers generate branching flows of secondary customers whose probability distributions also depend on the random environment state. An average cost of waiting of all customers per system working cycle is the optimality criterion. Explicit formulae have been obtained for some stationary probabilities. Non-preemptive priority service has been shown to be the optimal one.
|
