We consider a Markov Chain in which the state space is partitioned into sets where both transitions within sets and between sets have a special structure. Transitions within each set constitute a finite level dependent quasi-birth-and-death-process (LDQBD), and transitions between sets are restricted to six types of transitions. These latter types are needed to preserve the sets structure in the reduction step of our algorithm. Specifically, we present a successive censoring algorithm, based on matrix analytic methods, to obtain the stationary distribution of this system of connected LDQBD-processes.
- Connected level dependent QBD-processes
- Exact aggregation/disaggregation
- Matrix analytic methods
- Steady state analysis
- Successive censoring algorithm