Optical WDMA network design and function placement

A Parallel Simulation Methodology for Speedup and Obtaining Performance Estimates with Specific Accuracy: Experiences of its Application in Studies of Metropolitan Area Networks

“…Including the delays of segments transmitted at the start of simulation in the point estimate gives an over-optimistic (i.e., smaller) estimate of the steady state mean segment delay. To avoid this mistake, for each Simulation Engine and for each estimated parameter, AKAROA uses a twin-gate procedure comprising of Gafarian's Heuristic [25] with k = 25 crossings to obtain a rough estimate of the end of the transient phase [1,6,8], followed by repeated applications of Schruben's test [26] Speedup i (P) measures the mean real-time speedup seen by an AKAROA user if he/she was estimating the mean access delay of the ith station in a DQDB network to a specific precision.…”

“…As mentioned, another pitfall in the simulation of telecommunication systems is that each simulated process traverses an initial transient (&dquo;warm-up&dquo;) period. Observations generated during this &dquo;warm-up&dquo; phase do not characterise the steady-state behaviour of the simulated process [1,3,4,6,8]. For example, a natural point estimator of the steady-state mean access delay encountered by segments at station Si before they are transmitted is the arithmetic average of the sequence of delay observations of segments transmit-ted from that station.…”

“…Thus simulation output data have to be properly, statistically analysed before any constructive conclusions about the investigated networks are made. As stated by J. Kleijnen, a world authority on simulation: &dquo;...computer runs yield a mass of data but this mass may turn into a mess&dquo; if the random nature of output data is ignored, and then &dquo;...instead of an expensive simulation model, a toss of the coin had better be used...&dquo; Unfortunately, simulations in electrical and telecommunications engineering are often computationally intensive and can require excessively long run times in order to obtain results at a desired level of precision [8,9,10]. Moreover, our simulation studies are often iterative, so once results are at hand, they often motivate questions which require more runs to answer, and so on.…”

“…Unfortunately, simulations in electrical and telecommunications engineering are often computationally intensive and can require excessively long run times in order to obtain results at a desired level of precision [8,9,10]. Moreover, our simulation studies are often iterative, so once results are at hand, they often motivate questions which require more runs to answer, and so on.…”

Automating parallel simulation using parallel time streams

“…Our main motivation for developing a package for automating parallel steady-state stochastic simulation was to overcome problems caused by long simulation times experienced in our ongoing research in performance evaluation of high-speed and integrated-services communication networks [Pawlikowski and Yau 1991;Yau and Pawlikowski 1992a;1992c;1996;1999a;1999b;Yau 1999]. It was assumed that the package should accept sequential (nonparallel) simulation programs written in C or Cϩϩ, and all further stages of stochastic simulation should be transparent for users.…”