A Poisson limit for buffer overflow probabilities
Proceedings - IEEE INFOCOM - Tập 2 - Trang 994-1003 vol.2
Tóm tắt
A key criterion in the design of high-speed networks is the probability that the buffer content exceeds a given threshold. We consider n independent identical traffic sources modelled as point processes, which are fed into a link with speed proportional to n. Under fairly general assumptions on the input processes we show that the steady state probability of the buffer content exceeding a threshold b>0 tends to the corresponding probability assuming Poisson input processes. We verify the assumptions for a large class of long-range dependent sources commonly used to model data traffic. Our results show that with superposition, significant multiplexing gains can be achieved for even smaller buffers than suggested by previous results, which consider O(n) buffer size. Moreover, simulations show that for realistic values of the exceedance probability and moderate utilisations, convergence to the Poisson limit takes place at reasonable values of the number of sources superposed. This is particularly relevant for high-speed networks in which the cost of high-speed memory is significant.
Từ khóa
#Buffer overflow #Traffic control #Telecommunication traffic #Steady-state #High-speed networks #Costs #Tail #Asynchronous transfer mode #Probability #SwitchesTài liệu tham khảo
10.1287/moor.5.1.67
mocci, 1996, Broadband network teletraffic final report of action COST 242
dembo, 1992, Large Deviations Techniques and Applications
10.1109/49.400666
10.1017/S0305004100036781
cao, 2001, The effect of statistical multiplexing on the long range dependence of internet packet traffic
10.1109/26.486613
mandjes, 2001, Large deviations for small buffers: An insensitivity result, Queuing Systems, 10.1023/A:1010837416603
10.1109/49.400657
billingsley, 1968, Convergence of Probability Measures
daley, 1988, An Introduction to the Theory of Point Processes
10.1007/978-94-011-7801-3
10.1109/JSAC.1986.1146402
10.1145/378420.378440
ryu, 0, Point process approaches to the modeling and analysis of self-similar traffic: Part I-Model construction, Proc IEEE INFOCOM 1996, 1468
10.1239/aap/1013540352
cleveland, 1997, Trellis display: Modeling data from designed experiments
10.1109/90.491008
cao, 2002, A poisson limit for queueing functionals of superposed marked point processes
10.1109/90.554723
10.1109/90.392383
10.1239/jap/1029349455
10.1109/90.282603
10.2307/3215366
10.1007/BF01245322
konstantopoulos, 0, High variability vs long-range dependence for network performance, Proc 35th IEEE CDC Kobe Dec 1996, 1354
boxma, 1998, Fluid queues with long-tailed activity period distributions, Computer Communications, 21, 1509, 10.1016/S0140-3664(98)00219-9
ryu, 0, The importance of long-range dependence of VBR traffic in ATM traffic engineering: Myths and realities, Proc ACM SIGCOMM 1996, 3
10.2307/3215363