A new LP formulation of the admission control problem modelled as an MDP under average reward criterion.
Antonio PietrabissaPublished in: Int. J. Syst. Sci. (2011)
Keyphrases
- average reward
- admission control
- optimality criterion
- markov decision processes
- optimal policy
- long run
- control policy
- end to end
- discounted reward
- quality of service
- reinforcement learning
- semi markov decision processes
- policy iteration
- linear programming
- model free
- linear program
- web server
- resource management
- total reward
- dynamic programming
- state and action spaces
- finite state
- production system
- average cost
- markov chain
- hierarchical reinforcement learning
- markov decision problems
- optimal solution
- infinite horizon
- partially observable
- resource consumption
- decision problems
- state space
- reward function
- management system
- least squares
- markov decision process
- multistage
- np hard
- fixed point
- objective function