| Author | Kobchai Noimeesat |
| Call Number | AIT Thesis no.TC-01-11 |
| Subject(s) | Asynchronous transfer mode
|
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of
Engineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. TC-01-11 |
| Abstract | Self-similar traffic model is widely accepted as a potential tool for bursty traffic
modeling in the multimedia world. However, because of the significantly difference
between self-similar process and traditional traffic models, the performance estimation
from Poisson or related models are overly optimistic and not accurate for system
dimensioning. The search for more accurate performance evaluation of systems under
self-similar traffic has considerably attracted researchers in the last few years.
This thesis work studies the performance of ATM multiplexer under self-similar
traffic conditions based on multiplexing 2 to 8 homogeneous and heterogeneous input
traffic sources with 90% and 100% total offered load. The synthetic traffic trace is
modeled by using linear approximation Fractional Gaussian Noise and linear
transformation technique. The performance of A TM multiplexer is evaluated from several
viewpoints including Cell Loss Ratio (CLR), cell delay behavior, and the effect of Hurst
parameter (H) for different traffic cases and system configurations.
Simulation results reveal that the degree of the self-similarity of traffic has effect
on both cell loss ratio and delay behavior. For homogeneous case, the impact of selfsimilar traffic strongly depends on the Hurst value, time scale, traffic load, and especially
buffer size. When buffer size is not sufficiently large such that CLR > 10-3
, the effect of
the Hurst parameter is insignificant. This effect is due to the Long-Range Dependent
(LRD) property that occurs only when the observation time is sufficiently large. While
with the large buffer size, it increases not only the effect from Hurst parameter, but also
the self-similarity of output traffic. Time scale is another crucial parameter that adversely
affects on the performance. The performance of larger time scale (T > 10 milliseconds)
self-similar traffic tends to significantly degrade by the impact of the increasing Hurst
parameter.
In the case of multiplexing two heterogeneous streams and small time scale
(T = 100 cell time), both Cell Loss Ratio and Mean Cell Delay slightly different from
homogeneous case. However, performance gained when multiplexing two streams with
high Hurst values (H = 0.8 and 0.9) tends to be the same as when aggregating two stream
with H = 0.9. The simulations also show that when multiplexing a random traffic
(H = 0.5) with a self-similar traffic ( 0.5 < H < 1) the multiplexer output traffic becomes
self-similar traffic with Hurst value varying from 0.58 to 0.75.
|
| Year | 2001 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. TC-01-11 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Telecommunications (TC) |
| Chairperson(s) | Erke, Tapio; |
| Examination Committee(s) | Ahmed, Kazi M. ;Teerapat Sanguankotchakorn |
| Scholarship Donor(s) | Nokia Telecommunication Finland |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 2001 |