[LENA] Question regarding the SINR to CQI mapping in Uplink Scheduler

[Vikram Chandrasekhar]

Hello NS3-Experts,

This email is to raise an issue that I believe significantly impacts LENA peak throughput performance.

In the uplink scheduler, I noticed that the CQI over uplink is derived from the SINR using the following derivation. I also noticed a similar implementation in the "PiroEW2010" model [where the m_ber parameter is set to default value 0.00005] inside lte_amc.cc. 

          // translate SINR -> cqi: WILD ACK: same as DL

          double s = log2 ( 1 + (std::pow (10, minSinr / 10 )  / ( (-std::log (5.0 * 0.00005 )) / 1.5) ));

Above expressions translate to a Gap to capacity of approximately 7.42 dB.  Note that in practical systems e.g. LTE, since there is a Turbo code FEC, the actual gap to Shannon capacity is significantly smaller than 7. 42 dB (typically a gap of 3 dB is used to be conservative].  More details on the source of the above derivation is given below.

I would like to point out that this excessively pessimistic Gap approximation implies that the UL scheduler will significantly underestimate CQI which can be impact peak data-rates [medium-high SNR scenarios].

Any clarification would be appreciated.

- Vikram

PS: 

The original source for the derivation can be found at "Variable-rate Variable-power M-QAM for Fading Channels", IEEE Trans. on Comm., October 1997, Vol. 45, No. 10 by Goldsmith and Chua. I would like to point out that the above expressions have been originally derived while calculating the maximum consellation size of uncoded QAM, for a given target bit error rate, over an AWGN channel. Note the emphasis on the word "uncoded". 

[Marco Miozzo]

Hi Vikram,

the uplink scheduler has been implemented using the code of the downlink one and therefore might have some assumptions quite conservative.

It would be great if you can open a bugzilla issue and provide a more detailed explanation and a patch of the code with the relative documentation/references.

Many thanks.

Best regards,
marco.

--
You received this message because you are subscribed to the Google Groups "ns-3-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ns-3-users+...@googlegroups.com.
To post to this group, send email to ns-3-...@googlegroups.com.
Visit this group at http://groups.google.com/group/ns-3-users.
For more options, visit https://groups.google.com/groups/opt_out.

[Vikram Chandrasekhar]

Hi Marco,

I understand that the uplink scheduler has been implemented with conservative assumptions. It will be good to parameterize the "gap" parameter [currently set equal to - log (5 BERTarget)/1.5 ] within the UL scheduler, just as it is performed within the lte-amc.cc

I will follow-up on bugzilla. 

- Vikram

[Nicola Baldo]

Hi all,

I've filed a bug report on bugzilla for this issue:
https://www.nsnam.org/bugzilla/show_bug.cgi?id=1840

lets' continue the discussion there.

Regards,

Nicola

[Vikram Chandrasekhar]

Thanks Nicola. I have posted my response on bugzilla; I can volunteer to help out fix this bug.

Regards

Vikram

[akin soysal]

Hello,

I have looked at the bug report but could not find a fix. My observation is that the UL CQI is still very pessimistic that I get a lot of 0 UL CQIs. Has a fix been provided?

Best Regards,
Akın

[Reeta Gaokar]

Hello

Have you fixed the uplink cqi bug. Can you help me for the same.

Thanks and Regards