Page 423 - A First Course In Stochastic Models
P. 423

418             ALGORITHMIC ANALYSIS OF QUEUEING MODELS

                         Table 9.8.4  The minimal buffer size in the M/G/c/c + N queue
                                      ρ = 0.5                    ρ = 0.8
                          α  10 −2  10 −4  10 −6  10 −8  10 −10  10 −2  10 −4  10 −6  10 −8  10 −10
                  c=1
                   2   1
                  c =    exa  4    9    15   20   25     10   26   41   57    73
                   S   2
                         app  4    10   15   20   25     10   26   41   57    73
                   2
                  c = 2 exa   7    16   26   35   45     19   49   80  110   141
                   S
                         app  7    17   25   36   46     19   50   80  111   140
                  c=10
                   2   1
                  c =    exa  1    7    12   17   23     8    24   39   55    71
                   S   2
                         app  1    7    13   17   23     8    24   39   55    71
                   2
                  c = 2 exa   1    10   20   29   39     14   44   74  105   135
                   S
                         app  1    10   20   29   39     14   45   75  106   135
                assumed that the number of time slots needed to serve a packet has a geometric
                distribution {(1 − r) n−1 r, n ≥ 1}. The case of deterministic services is included as
                a special case (r = 1). In many telecommunication applications the service time
                of a packet is deterministic and equals one time slot. A served packet leaves the
                system at the end of the time slot in which the service is completed. The numbers of
                packets arriving in the system during consecutive time slots are independent non-
                negative random variables with the common probability distribution {a n , n ≥ 0}.
                It is assumed that the packets arrive individually during the time slots and that
                an arriving packet is rejected when it finds the buffer full upon arrival. It is no
                restriction to use the convention of individual arrivals provided that the partial
                rejection strategy applies when arrivals actually occur in batches. The load factor
                ρ is defined as
                                                  λµ
                                              ρ =   ,
                                                  c
                           ∞

                where λ =     na n is the arrival rate of new packets and µ = 1/r is the mean
                           n=1
                service time of a packet. Let
                          P rej = the long-run fraction of packets that are rejected.
                Under the assumption of ρ < 1 an approximation to P rej can be given in terms
                of the state probabilities in the corresponding infinite-buffer model. Assuming that
                                                                 (∞)
                ρ < 1, define for the infinite-buffer model the probability u  by
                                                                j
                     (∞)
                    u   = the long-run fraction of time slots at whose beginnings there are
                     j
                          j packets in the system
                for j = 0, 1, . . . . By the assumption of geometrically distributed service times, the
                process describing the number of packets present at the beginning of a time slot is
   418   419   420   421   422   423   424   425   426   427   428