The Complexity of Optimal Queuing Network Control
- 1 May 1999
- journal article
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Mathematics of Operations Research
- Vol. 24 (2), 293-305
- https://doi.org/10.1287/moor.24.2.293
Abstract
We show that several well-known optimization problems related to the optimal control of queues are provably intractable—independently of any unproven conjecture such as P ≠ NP. In particular, we show that several versions of the problem of optimally controlling a simple network of queues with simple arrival and service distributions and multiple customer classes is complete for exponential time. This is perhaps the first such intractability result for a well-known optimization problem. We also show that the restless bandit problem (the generalization of the multi-armed bandit problem to the case in which the unselected processes are not quiescent) is complete for polynomial space.Keywords
This publication has 10 references indexed in Scilit:
- Performance bounds for queueing networks and scheduling policiesIEEE Transactions on Automatic Control, 1994
- Markov Decision ProcessesWiley Series in Probability and Statistics, 1994
- Optimization of Multiclass Queueing Networks: Polyhedral and Nonlinear Characterizations of Achievable PerformanceThe Annals of Applied Probability, 1994
- On an index policy for restless banditsJournal of Applied Probability, 1990
- Branching Bandit ProcessesProbability in the Engineering and Informational Sciences, 1988
- Restless bandits: activity allocation in a changing worldJournal of Applied Probability, 1988
- Games against natureJournal of Computer and System Sciences, 1985
- AlternationJournal of the ACM, 1981
- Time-Sharing Service Systems. ITheory of Probability and Its Applications, 1975
- Word problems requiring exponential time(Preliminary Report)Published by Association for Computing Machinery (ACM) ,1973