Scribd
Upload
52 views3 pages

Mg1 Numerical Inversion Methods

This document discusses a numerical inversion method using MATLAB to obtain the probability density function (pdf) for the delay distribution of an M/G/1 queue. It involves sampling the frequency domain representation of the delay distribution and applying the inverse fast Fourier transform. The cumulative distribution function is then obtained through numerical integration of the pdf. Sample data involves 6 stations generating Poisson arrivals of messages with an average rate of 0.05 arrivals/second and an average message length of 6 packets.

Uploaded by

Koushik Kashyap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
52 views3 pages

Mg1 Numerical Inversion Methods

This document discusses a numerical inversion method using MATLAB to obtain the probability density function (pdf) for the delay distribution of an M/G/1 queue. It involves sampling the frequency domain representation of the delay distribution and applying the inverse fast Fourier transform. The cumulative distribution function is then obtained through numerical integration of the pdf. Sample data involves 6 stations generating Poisson arrivals of messages with an average rate of 0.05 arrivals/second and an average message length of 6 packets.

Uploaded by

Koushik Kashyap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

M/G/1 Delay Distribution

Numerical Inversion Method with


Matlab

Numerical Inversion for the Delay


Distrib. D(s) of an M/G/1 Queue

In order to obtain numerically the pdf that corresponds to D(s) for the
FIFO discipline, we change variable s = j2f, where j is the imaginary
unit (j2 = - 1). Then, we sample in the frequency domain D(f) = D(s =
j2f) with interval fc and we apply the Inverse Fast Fourier
Transform (IFFT) algorithm, by considering their scaled versions by
1/Tc, where Tc is the sampling interval in the time domain.

We make the approximation that the Fourier components are negligible


for f > fmax. We determine fmax so that Tc = 1/(2fmax) is the smallest
possible delay value (i.e., service time).

We expect that the pdf vanishes to zero for high t values. Hence, the
number of samples of the IFFT, N, can be determined considering the
approximation that the pdf is equal to zero for t > NTc. Since the pdf is
unknown, we will use suitable values for NTc and we will a posteriori
verify that the obtained pdf is negligible for t > NTc.

The cumulative distribution function (cdf) has been obtained through


numerical integration of the pdf (e.g., cumsum function in Matlab).

Dati usati:
6 stazioni ciascuna
generante arrivi di
messaggi secondo un
processo di Poisson con
tasso medio 0.05 arrivi/s
Lunghezza media dei
messaggi di 6 pacchetti.
Tempo di trasmissione di
un pacchetto di 0.2346 s

Cumul. distribution function (cdf) Prob. density function (pdf)

M/G/1 Delay Distribution:


Results
M/G/1 Theory: Numerical Derivation of Delay Distributions

10

geometric
fixed

-2

10

-4

10

-6

10

10
0

10
time [s]

10

10

geometric
fixed
-1

10

-2

10

10

10
time [s]

10

You might also like