Notes Project Output Archive Links
Index table
+Intro
+Prephase
+TimePlan
+Outline
+Research Phase
+-28 Feb.-
+-5 Mar.-
+-12 Mar.-
+-Odysee
+-19 Mar.-
+-26 Mar.-
+-2 Apr.-
+Develope Tests Spec.
+- 9 Apr.-
+Test Phase
+- 16 Apr.-
+- 23 Apr.-
+- > 30 Apr.-
+Review Results
+- 7 May.-
+Prep. Report
+- 14 May.-
+- 21 May.-
+After Project

Testing special parameters

ChangedRetransmissionCounterPicture As a big idea and question of me was, if the IEEE802.11 protocol, implemented in this special way with retransmission to decreas the BER. Will have problems with VoIP data, which emphasizes on delay and not BER. For that reason I changed the retransmission counter and got the results like shown in the pictures to the right.

The chosen capacity, coverage & quality test

Finally I did only 2,4 and 8 clients tests, didn t define a threshold, but draw the average delay at the y axis. Draw 3 diagrams (for 2,4 and 8 clients), used the distance (measured by the received power in dBm) as x axis and draw for each quality a line. This made altogether about 3 days testing. However, the problem occured at looking at the delay, time graph, which showed a steady increasing delay.

Increased delay problem

Having the decision to stop the tests and fix the problem or continue afer a short investigation. I chose to test short if this delay is part of the wireless system or of the measuring equipment, by using a wired network connection. The symptom was the same, as it is always possible to recalculate influenced data to gain the real values I decided to continue the tests. Later on my supervisor was not convinced by the correctness of the results and I used only the direction into my report and postboned the prove, what caused the delay, to later. The time starts to become short. One part of proving, that the measureing I am doing is not completely wrong was done by setting up a real WLLAN setup and measuring the delay of a simplex and duplex channel, correcting the values and comparing the curves. The expected Gausian curve, which shifts toward a longer average delay in case of a heavier loaded channel could be seen - my software worked in this point correct.
A fast sketch of the result, without bothering to investigate time for a more formal diagram, is shown belowe.
SketchofComprehensiveTestResult
© and created by Jörg Abendroth