Lining Theory For Dummies

Queuing theory for dummies l.jpg
1 / 33
735 days ago, 379 views
PowerPoint PPT Presentation
Lining Hypothesis For Shams. Jean-Yves Le Boudec. Everything You Need to Think About Lining Hypothesis. Lining is fundamental to comprehend the conduct of complex PC and correspondence frameworks top to bottom investigation of lining frameworks is hard Luckily, the most imperative results are simple

Presentation Transcript

Slide 1

Lining Theory For Dummies Jean-Yves Le Boudec 1

Slide 2

All You Need to Know About Queuing Theory Queuing is fundamental to comprehend the conduct of complex PC and correspondence frameworks top to bottom investigation of lining frameworks is hard Fortunately, the most critical outcomes are simple We will concentrate this theme in two modules 1. straightforward ideas (this module) 2. lining systems (later) 2

Slide 3

1. Deterministic Queuing Easy yet capable Applies to deterministic and transient investigation Example: playback support estimating 3

Slide 4

Use of Cumulative Functions 4

Slide 5

Solution of Playback Delay Pb bits A(t) A'(t) D(t) d(t) (D2): r (t - d(0) - D ) (D1): r(t - d(0) + D ) time d(0) - D d(0) d(0) + D A. 5

Slide 6

2. Operational Laws Intuition: Say each client pays one Fr for each moment introduce Payoff per client = R Rate at which we get cash = N In normal λ clients every moment, N = λ R 6

Slide 7

Little Again Consider a recreation where you measure R and N. You utilize two counters responseTimeCtr and queueLengthCtr. At end of reproduction, gauge R = responseTimeCtr/NbCust N = queueLengthCtr/T where NbCust = number of clients served and T=simulation length Both responseTimeCtr=0 and queueLengthCtr=0 at first Q: When a landing or flight occasion happens, how are both counters refreshed ? A: queueLengthCtr += (t new - t old ) . q(t old ) where q(t old ) is the number of clients in line just before the occasion. responseTimeCtr += (t new - t old ) . q(t old ) in this way responseTimeCtr == queueLengthCtr and along these lines N = R . NbCust/T ; now NbCust/T is our estimator of  7

Slide 8

Other Operational Laws 8

Slide 9

The Interactive User Model 9

Slide 10

Network Laws 10

Slide 11

Bottleneck Analysis Example Apply the accompanying two limits (1) (2) 17 11

Slide 12

Throughput Bounds 12

Slide 13

Bottlenecks A 13

Slide 14

DASSA Intuition: inside one occupied period: to each flight we can connect one entry with same number of clients abandoned 14

Slide 15

3. Single Server Queue 15

Slide 16

i.e. which are occasion midpoints (versus time midpoints ?) 16

Slide 17


Slide 18


Slide 19


Slide 20

Non Linearity of Response Time 20

Slide 21

Impact of Variability 21

Slide 22

Optimal Sharing Compare the two regarding Response time Capacity 22

Slide 23

The Processor Sharing Queue Models: processors, organize joins Insensitivity: whatever the administration necessities: Egalitarian 23

Slide 24

PS versus FIFO PS FIFO 24

Slide 25

4. A Case Study Impact of limit increment ? Ideal Capacity ? 25

Slide 26

Methodology 26

Slide 27

4.1. Deterministic Analysis 27

Slide 28

Deterministic Analysis 28

Slide 29

4.2 Single Queue Analysis Assume no criticism circle: 29

Slide 30

4.3 Operational Analysis A refined model, with coursing clients Apply Bottleneck Analysis ( = Operational Analysis ) holding up time 1/c Z/(N-1) - Z 30

Slide 31


Slide 32


Slide 33

Conclusions Queuing is basic in correspondence and data frameworks M/M/1, M/GI/1, M/G/1/PS and variations have shut structures Bottleneck investigation and most pessimistic scenario examination are typically exceptionally basic and frequently give great bits of knowledge … it stays to see lining systems 33