LTE - Sinr - REM/CQI
1,440 views
Skip to first unread message
Arsalan Saeed
Jun 15, 2012, 1:19:54 PM6/15/12
to ns-3-users
Hello there,
Just a naive questions, may be the developers can comment on this.
As we know the amc module uses the SINR (_sinr) value to calculate the
Spectral Eff, and CQI so on.
where as when we are generating REM map, This also produces a SINR
value in ASCII file.
So now the question is, if a EnodeB is at position A and UE at
position B.
Should it not be that the SINR for calculating CQI for user at
PositionB, be the same as the SINR value in ASCII file from REM map at
Position B?
I am sure about my dB-Watt conversions, in both outputs. But I have
observed that the SINR values are not same in both cases. Or you can
say the RxPsd is not same.
Many Thanks,
Arsalan
Just a naive questions, may be the developers can comment on this.
As we know the amc module uses the SINR (_sinr) value to calculate the
Spectral Eff, and CQI so on.
where as when we are generating REM map, This also produces a SINR
value in ASCII file.
So now the question is, if a EnodeB is at position A and UE at
position B.
Should it not be that the SINR for calculating CQI for user at
PositionB, be the same as the SINR value in ASCII file from REM map at
Position B?
I am sure about my dB-Watt conversions, in both outputs. But I have
observed that the SINR values are not same in both cases. Or you can
say the RxPsd is not same.
Many Thanks,
Arsalan
Arsalan Saeed
Jun 15, 2012, 2:09:30 PM6/15/12
to ns-3-users
Just to add , there is only one EnodeB and one UE in the
consideration.
Or SINR = rxsignal / ((allsignal-rxsignal)+noise)
'allsignal' is equal to 'rxsignal', as there is no interference.
so Sinr = rxsignal/noise.
And i am using lena-dual-stripe example with minor modifications to
have fixed node locations.
Case1, Epc = true, Rem=false
Case2, Epc= fasle,Rem True
consideration.
Or SINR = rxsignal / ((allsignal-rxsignal)+noise)
'allsignal' is equal to 'rxsignal', as there is no interference.
so Sinr = rxsignal/noise.
And i am using lena-dual-stripe example with minor modifications to
have fixed node locations.
Case1, Epc = true, Rem=false
Case2, Epc= fasle,Rem True
Nicola Baldo
Jun 16, 2012, 10:53:18 AM6/16/12
to ns-3-...@googlegroups.com
In your scenario, if the positions are the same, then rxsignal is expected to be the same . Note that the noise power can be different, please refer to the documentation of the attributes of RadioEnvironmentMapHelper and LteUePhy for details.
Arsalan Saeed
Jun 17, 2012, 10:33:14 AM6/17/12
to ns-3-users
Hi Nicola,
Thanks for the reply, I understand your point, but there is a strange
behavior that I have observed. I might be wrong but I have read the
Documentation (Ns3-Model-Library) quite throughly and with a clean new
install of ns3.14 & also M4 lena. and only using the provided
examples. (lena-dual-stripe / lena-intercell-interference). I am
observing a major difference in Sinr from the REM maps & Sinr at amc
module to calculate CQI.
I would be thankful if someone can verify this at there end, as
following is my scenario
Distance between node 20m. ( 20m - 100m)
Tx power 17dbm*
Rem maps give a reasonable sinr output and is defiantly comparable to
the provided in the documentation.
Where as the sinr at the amc module gives a high sinr value (ranging
upto 30dB), and varying the distance between UE & EnodeB upto 100m
still gives MCS of 28. (This can be easily verified as an abnormal
behavior with theory,as a low power Txnode* cannot provide MCS - 28
upto 100m or more).
This behavior can also be double checked with a rough look at the REM
maps(running example or documentation figure) as the received signal
significantly decreases at 100m (more visible if one removes the
buildings / other node interferences), Hence the MCS / CQI can not be
maximum over this distance.
Please correct me if I am wrong, I was ignoring this behavior until I
fixed the positions of my node,removed buildings and considering only
one UE & EnodeB, which made this doubt mentionable.
If you consider there can be a possibility of having a look at this
issue, I can provide some codes to observe this aswell as some manual
calculations of NoisePSD/rxSignal/Indoor Loss itu-1238 etc.
NS3 Lte module is an excellent platform & I appreciate the work of the
developers. This post is just an effort to make the simulator more
reliable.
Thanks,
Arsalan
Thanks for the reply, I understand your point, but there is a strange
behavior that I have observed. I might be wrong but I have read the
Documentation (Ns3-Model-Library) quite throughly and with a clean new
install of ns3.14 & also M4 lena. and only using the provided
examples. (lena-dual-stripe / lena-intercell-interference). I am
observing a major difference in Sinr from the REM maps & Sinr at amc
module to calculate CQI.
I would be thankful if someone can verify this at there end, as
following is my scenario
Distance between node 20m. ( 20m - 100m)
Tx power 17dbm*
Rem maps give a reasonable sinr output and is defiantly comparable to
the provided in the documentation.
Where as the sinr at the amc module gives a high sinr value (ranging
upto 30dB), and varying the distance between UE & EnodeB upto 100m
still gives MCS of 28. (This can be easily verified as an abnormal
behavior with theory,as a low power Txnode* cannot provide MCS - 28
upto 100m or more).
This behavior can also be double checked with a rough look at the REM
maps(running example or documentation figure) as the received signal
significantly decreases at 100m (more visible if one removes the
buildings / other node interferences), Hence the MCS / CQI can not be
maximum over this distance.
Please correct me if I am wrong, I was ignoring this behavior until I
fixed the positions of my node,removed buildings and considering only
one UE & EnodeB, which made this doubt mentionable.
If you consider there can be a possibility of having a look at this
issue, I can provide some codes to observe this aswell as some manual
calculations of NoisePSD/rxSignal/Indoor Loss itu-1238 etc.
NS3 Lte module is an excellent platform & I appreciate the work of the
developers. This post is just an effort to make the simulator more
reliable.
Thanks,
Arsalan
Nicola Baldo
Jun 18, 2012, 4:42:54 AM6/18/12
to ns-3-...@googlegroups.com
Hi Arsalan,
On Sunday, June 17, 2012 4:33:14 PM UTC+2, Arsalan Saeed wrote:
yes, if you can send this it then we can take a look. It would also be of help if you could post a REM and also post a set of pairs (MCS, UE distance) for some relevant values of the UE distance from the eNB.
By the way, which propagation model are you using? It's a fundamental piece of information for the problem you are reporting, still you did not mention it so far.
Regards,
Nicola
On Sunday, June 17, 2012 4:33:14 PM UTC+2, Arsalan Saeed wrote:
If you consider there can be a possibility of having a look at this
issue, I can provide some codes to observe this aswell as some manual
calculations of NoisePSD/rxSignal/Indoor Loss itu-1238 etc.
yes, if you can send this it then we can take a look. It would also be of help if you could post a REM and also post a set of pairs (MCS, UE distance) for some relevant values of the UE distance from the eNB.
By the way, which propagation model are you using? It's a fundamental piece of information for the problem you are reporting, still you did not mention it so far.
Regards,
Nicola
Arsalan Saeed
Jun 20, 2012, 9:51:55 AM6/20/12
to ns-3-users
Just to update that this topic is being followed up on Bugzilla ( ID
1460)
1460)
Arsalan Saeed
Jun 18, 2012, 5:10:55 AM6/18/12
to ns-3-users
Hi Nicola,
Thanks for pointing that out, I am using the recently added
HybridBuildingPropogationLossModel, The one recently added with Lena-
Dual-Stripe.
More specifically the Itu-1238/Itu1411(shortrange) in
HybridBuildingProp -> the final result of the loss + Shadowing gets
accumulated in BuildingsPropagationLossModel.
I have verifed to this far with manual calculations that the Pathloss
from the ITU specifications are very correctly calculated in the
simulator and added to
BuildingsPropagationLossModel::DoCalcRxPower
I will post a complete set of test results regard this soon.
Many Thanks for your time.
Arsalan
Thanks for pointing that out, I am using the recently added
HybridBuildingPropogationLossModel, The one recently added with Lena-
Dual-Stripe.
More specifically the Itu-1238/Itu1411(shortrange) in
HybridBuildingProp -> the final result of the loss + Shadowing gets
accumulated in BuildingsPropagationLossModel.
I have verifed to this far with manual calculations that the Pathloss
from the ITU specifications are very correctly calculated in the
simulator and added to
BuildingsPropagationLossModel::DoCalcRxPower
I will post a complete set of test results regard this soon.
Many Thanks for your time.
Arsalan
Reply all
Reply to author
Forward
0 new messages