Bugzilla – Bug 1460
LTE - EPC/REM Different SINR values
Last modified: 2012-10-17 13:02:31 EDT
Created attachment 1415 [details]
The Sinr values for a fixed coordinate at REM maps should be similar to that produced as a sinr for a use simulating EPC. Use the example lena-dual-stripe & slightly modifying the example to 1UE & 1HomeENodeB produces different sinr results as compared to that for Rem maps.
This leads to a incorrect CQI (Unexpectedly very high) as the high Sinr value results in a higher spectral efficiency.
There are some tests performed to reproduce the problem which are attached to this post. Case2 clearly shows the unexpected behavior.
Example - lena-dual-stripe
number of nodes: 1UE, 1 EnodeB
distance: 30m(case1), 100m(case2), 500m/2000m(case3/4)
Propagation model: HybridBuildingPropagationLossModel->BuildingsLossModel(ITU1238/ITU1411/OH)
Bandwidth/Txpower/Noise/Shadowing: Kept same as in the example (Txpower for Txnode 17dBm)
The attached document would show the manual calculations for another test case. The calculations show the pathloss & shadowing values are rightly calculated in the simulator. A major expected cause could be the incorrect coordination of Use of bandwidth while calculation the RxPower. (BuildingLossModel.cc)
Or noisePSD/Rxpsd values for either EPC/REM should match.
1-Two relevant NS3 Users topics
2- A short presentation explaining the test cases & comparisons
3- Link to complete codes & fugures/ Manual calculations
<Rapid share link>
I wanted to have a look at this bug to see if we can fix it before the next release, but unfortunately the file you posted on rapidshare is not available any more. Could you please upload it to bugzilla instead?
Since the folder contains MacStats for all the cases, the file size is larger than Bugzilla's limit. Hence re uploading it on rapid share.