Bug 1515

Summary:	REM shows increased SINR when co-located eNBs have same EARFCN
Product:	ns-3	Reporter:	Nicola Baldo <nicola>
Component:	lte	Assignee:	Nicola Baldo <nicola>
Status:	REOPENED ---
Severity:	normal	CC:	hansgarry, ns-bugs
Priority:	P5
Version:	ns-3-dev
Hardware:	All
OS:	All
Attachments:	program that should reproduce the bug An example that shows the problem

Description Nicola Baldo 2012-10-17 12:27:00 UTC

https://groups.google.com/d/topic/ns-3-users/DwIo411ndBQ/discussion

Garry wrote:
If I do not enable EPC and simply plot the REM, I observe that if I set both the eNB devices to the same EARFCN, the SINR increases. Is this normal? I believe that this should not be happening.

Nicola wrote:
I agree it should not happen. However, the EARCNs works properly in the lena-dual-stripe example. Are you sure it's not due to your particular sim program? 

Garry wrote:
you are not observing the increase in SINR in dual-stripe, because there is only one LteEnbNetDevice per eNB. I am plotting the REM for the case where I have 2 such devices per eNB, and I set the same earfcn to both of them. 

Garry wrote:
I re-wrote my setup, and instead of defining 2 devices per eNB, I simply place a 2nd eNB in the same location and set it to a different EARFCN. However, the REM problem still shows up: I observe that if I set all eNBs to the same EARFCN, the SINR is increased (!).

Comment 1 Nicola Baldo 2012-10-17 13:55:32 UTC

Created attachment 1456 [details]
program that should reproduce the bug

Comment 2 Nicola Baldo 2012-10-17 14:02:19 UTC

The attached program creates a scenario with co-located enbs that takes the following parameters:
    --numEnbs: how many eNBs there are
    --earfcnStep: earfcn difference between successive eNBs

I ran it with different parameters parameters:

$ ./waf --run scratch/lena-rem-multiple-enbs --command="%s --numEnbs=2 --earfcnStep=0"

$ ./waf --run scratch/lena-rem-multiple-enbs --command="%s --numEnbs=2 --earfcnStep=1"

$ ./waf --run scratch/lena-rem-multiple-enbs --command="%s --numEnbs=2 --earfcnStep=50"

$ ./waf --run scratch/lena-rem-multiple-enbs --command="%s --numEnbs=2 --earfcnStep=100"

and here are the results:

$ cat 2enbs_0earfcnStep.rem 
-400	-300	0	0.376573
-400	-100	0	0.47042
-400	100	0	0.47042
-400	300	0	0.376573
-133.333	-300	0	0.583527
-133.333	-100	0	0.844632
-133.333	100	0	0.844632
-133.333	300	0	0.583527
133.333	-300	0	0.583527
133.333	-100	0	0.844632
133.333	100	0	0.844632
133.333	300	0	0.583527
400	-300	0	0.376573
400	-100	0	0.47042
400	100	0	0.47042
400	300	0	0.376573

$ cat 2enbs_1earfcnStep.rem 
-400	-300	0	0.379751
-400	-100	0	0.47539
-400	100	0	0.47539
-400	300	0	0.379751
-133.333	-300	0	0.591193
-133.333	-100	0	0.860789
-133.333	100	0	0.860789
-133.333	300	0	0.591193
133.333	-300	0	0.591193
133.333	-100	0	0.860789
133.333	100	0	0.860789
133.333	300	0	0.591193
400	-300	0	0.379751
400	-100	0	0.47539
400	100	0	0.47539
400	300	0	0.379751

$ cat 2enbs_50earfcnStep.rem 
-400	-300	0	0.604037
-400	-100	0	0.888289
-400	100	0	0.888289
-400	300	0	0.604037
-133.333	-300	0	1.40112
-133.333	-100	0	5.43633
-133.333	100	0	5.43633
-133.333	300	0	1.40112
133.333	-300	0	1.40112
133.333	-100	0	5.43633
133.333	100	0	5.43633
133.333	300	0	1.40112
400	-300	0	0.604037
400	-100	0	0.888289
400	100	0	0.888289
400	300	0	0.604037

$ cat 2enbs_100earfcnStep.rem 
-400	-300	0	0.604037
-400	-100	0	0.888289
-400	100	0	0.888289
-400	300	0	0.604037
-133.333	-300	0	1.40112
-133.333	-100	0	5.43633
-133.333	100	0	5.43633
-133.333	300	0	1.40112
133.333	-300	0	1.40112
133.333	-100	0	5.43633
133.333	100	0	5.43633
133.333	300	0	1.40112
400	-300	0	0.604037
400	-100	0	0.888289
400	100	0	0.888289
400	300	0	0.604037

To me the SINRs in the REMs look reasonable: the value decreases with increasing EARFCN difference, until the two eNBs are in completely orthogonal frequencies, at which point the SINR does not improve any more.

Comment 3 Nicola Baldo 2012-10-17 14:04:32 UTC

Based on the tests that I made, I am closing this bug as invalid. Feel free to reopen it if you can provide a test program that successfully reproduces the supposed buggy behavior.

Comment 4 Nicola Baldo 2012-10-17 14:06:38 UTC

(In reply to comment #2)
> the value decreases with increasing EARFCN difference

whoops, I meant "the value increases" of course.

Comment 5 Garry 2012-10-17 15:17:15 UTC

Created attachment 1457 [details]
An example that shows the problem

I attach a sample code that shows what I mean. If I set the DlEarfcn to 100 (check lines 68/69 in the code), the SINR in the REM increases. Maybe my code is wrong?

Comment 6 Garry 2012-11-20 11:46:14 UTC

Problem is still open.