Queue Performance in Presence of Long-Range Dependencies – An Empirical Study

摘要: The queuing, as an inherent feature of each real system, plays a very important role in teletraffic networks well. It is known that the queuing system consists three basic components: input stream requests, service processes and queue discipline. optimal performance can be achieved only case when all components are “matched” together. models describe early telephone services (e.g. Poisson Markov) have been successful. One main reasons for this success with models, which accurately resembles real-world behavior, has due to highly static, low varying nature development phase voice traffic. This means whole environment treated simple system. features such systems described by: thermodynamic quasi-equilibrium, homogenous topologies, random graph theory, short-range dependencies, distribution, circuit switching, etc. However, evolution from new data traffic integrated inevitably leads vastly different statistical characteristics much more irregular variable result undermine basis traditional models. In paper we show taking into account complex approach existence long-range dependencies (expressed by 1/f noise) influence performance. done basing on analytical presentation dependent model also experimental study behavior computer network during daily workload generated user.