Stabilizing Queuing Networks With Model Data-Independent Control

Classical queuing network control strategies typically rely on accurate knowledge of model data, i.e., arrival and service rates. However, such data are not always available may be time-variant. To address this challenge, we consider a class data-independent (MDI) policies that only traffic state observation topology. In article, focus the MDI can stabilize multiclass Markovian networks under centralized decentralized policies. Control actions include routing, sequencing, holding. By expanding routes constructing piecewise-linear test functions, derive an easy-to-use criterion to check stability given policy. For stabilizable networks, show centralized, stabilizing policy exists. single-class further decentralized, addition, for both settings, construct explicit attain maximal throughput present numerical examples illustrate results.