References
- He , Q. 1996 . Queues with marked customers . Adv. Appl. Prob. , 28 : 567 – 587 .
- He , Q. 2001 . The versatility of the MMAP[K] and the MMAP[K]/G[K]/1 queue . Queueing Systems , 38 : 397 – 418 . [CROSSREF]
- He , Q. and Neuts , M.F. 1998 . Markov chains with marked transitions . Stochastic Processes Appl. , 74 : 37 – 52 . [CROSSREF]
- Latouche , G. and Ramaswami , V. 1999 . Introduction to Matrix Analytic Methods and Stochastic Modeling Philadelphia : SIAM .
- Meini , B. 1998 . Solving QBD problems: The cyclic reduction algorithm versus the invariant subspace method . Adv. Performance Anal. , 1 : 215 – 225 .
- Ramaswami, V. The generality of QBD processes. In Proceedings of the Second International Conference on Matrix Analytical Methods, Winnipeg, Manitoba, 1998, Notable Publications Inc
- Takine , T. 2001 . Queue length distribution in a FIFO single-server queue with multiple arrival streams having different service time distributions . Queueing Systems Appl (QUESTA) , 39 ( 4 ) : 349 – 375 . [CROSSREF]
- Takine , T. and Hasegawa , T. 1994 . The workload in a MAP/G/1 queue with state-dependent services: its applications to a queue with preemptive resume priority . Stochastic Models , 10 ( 1 ) : 183 – 204 .
- Van Houdt , B. and Blondia , C. 2002 . The delay distribution of a type k customer in a first come first served MMAP[K]/PH[K]/1 queue . J. Appl. Prob. , 39 ( 1 ) : 213 – 222 . [CROSSREF]
- Van Houdt , B. and Blondia , C. 2002 . The waiting time distribution of a type k customer in a FCFS MMAP[K]/PH[K]/2 queue. Technical Report
- #B. Van Houdt is a postdoctoral Fellow of the FWO Flanders.
- aHere, the delay is defined as the waiting time plus the service time. One can obtain the waiting time distribution from d k by means of a deconvolution, as the service time is independent from the waiting time.
- bNotice, we do not know which customer occupies which server. We can easily add this information to the MC by stating that the type k i customer occupies server i. However, this would imply that the condition k 1 ≥ k 2 is lost, hence, the number of states that are part of each level increases (up to a factor 2). Adding this additional info would however simplify the description of the transition probabilities.
- cOf course, both systems, have a different R matrix.