PDA

View Full Version : CQI base optimization



amadeo
2010-12-29, 02:12 AM
Hi Guys,
It seems CQI can be a good soource to optimize HSPA+ network. However, i think containment of the coverage still the best way to improve 3G quality. The problem is, no 3gpp recommendation about how does UE calculating the CQI.
Does any of you have document about it?

cheers,

amadeo

tareq1010
2010-12-29, 03:09 AM
the CQI value that the UE reports does not just correspond to the Ec/No or the SIR the UE is experiencing. instead, the value reported is the function of the multipath enviroment, terminal receiver type, ration of the interference of the own own cell compared with other and expected the cell HSDPA power availability.

suver
2010-12-29, 01:06 PM
there will be many parameter to analyze.. better you see SINR

agenov
2010-12-29, 04:34 PM
Hi Guys,
It seems CQI can be a good soource to optimize HSPA+ network. However, i think containment of the coverage still the best way to improve 3G quality. The problem is, no 3gpp recommendation about how does UE calculating the CQI.
Does any of you have document about it?

cheers,

amadeo

3GPP provide complete freedom to different vendors for CQI calculation so you will not find anything there. What the specification define is the range from 0-30 and to corresponds to 10% BLER. So different vendors use different approaches inside the UE as well as CQI compensation algorithm in the NodeB from the Network side that is report to the RNC.

mon3em
2010-12-29, 05:30 PM
the common in calculation is Ec/No , but different vendors have different factors to calculate CQI.

Fernando
2010-12-29, 10:12 PM
UE calculate CQI as a mapping function with SNR and see the attached picture of SNR formula.

Obviously, the SNR is a function of HS-PDSCH power and the interference

However, because of dynamic power allocation used by many Vendor, UE could not know exactly how much is the power of HS-PDSCH. To be precise, the UE only knew at the beginning of the allocation as the BTS signalled it with the message at the allocation point, but once the connection was continue, the power changed and the UE did't get the update, he still use the power from the allocation moment to calculate the CQI.

So, the CQI we seen from the UE measurement is not totally 100% reliable, but anyway, the network once it received the CQI will put some compensate (Vendor specific) to adjust to the correct one but this correct CQI value, we can not see from the log file.

mardaphoca
2010-12-30, 12:26 AM
Hi,

I have found a paper where this formula is used to obtain CQI:

CQI = 0 if SINR<-3.96 dB
CQI = SINR/1.02 +4.81 if -3.96<SINR<26.04 dB
CQI = 30 if SINR >26.04 dB

I don't know if it is accurate, maybe some members could help with this :)

Fernando, I can't see your attachment, can you share the content of the image?

Fernando
2010-12-30, 11:23 PM
Hi,

I have found a paper where this formula is used to obtain CQI:

CQI = 0 if SINR<-3.96 dB
CQI = SINR/1.02 +4.81 if -3.96<SINR<26.04 dB
CQI = 30 if SINR >26.04 dB

I don't know if it is accurate, maybe some members could help with this :)

Fernando, I can't see your attachment, can you share the content of the image?

mardaphoca, the image is just a 1 line equation of the SNR that I've mention before (function of power HS-PDSCH and interference), it's just hard to type them in this text box.

mobilink999
2011-01-19, 06:53 PM
Actually I thought of CQI based optimization my self but its not much practical. Also the UE buffer is also a factor in throughput.

danikd
2011-01-19, 07:50 PM
UE calculate CQI as a mapping function with SNR and see the attached picture of SNR formula.

Obviously, the SNR is a function of HS-PDSCH power and the interference

However, because of dynamic power allocation used by many Vendor, UE could not know exactly how much is the power of HS-PDSCH. To be precise, the UE only knew at the beginning of the allocation as the BTS signalled it with the message at the allocation point, but once the connection was continue, the power changed and the UE did't get the update, he still use the power from the allocation moment to calculate the CQI.

So, the CQI we seen from the UE measurement is not totally 100% reliable, but anyway, the network once it received the CQI will put some compensate (Vendor specific) to adjust to the correct one but this correct CQI value, we can not see from the log file.

Can you elaborate the formula? What is NF and what is hi^2?

Regards

Tomas Delgado
2011-08-10, 03:03 AM
Already improved CQI reports and nothing happenned to througput, Huawei equipment.

PM-GIA
2011-08-12, 04:03 PM
I also face this problem, as UE has good CQI (>27) but throughput still low (max 12Mbps). Huawei HSPA+ network with bundle to 21Mbps max.

Please give me some idea to check what's really restrict in term of throughput!!!

Thanks,

danikd
2011-08-12, 09:56 PM
I also face this problem, as UE has good CQI (>27) but throughput still low (max 12Mbps). Huawei HSPA+ network with bundle to 21Mbps max.

Please give me some idea to check what's really restrict in term of throughput!!!

Thanks,

The main problem on Huawei HSPA network that I've observed is a dynamic HS code allocation problem. The maximal mentioned throughput can be observed only once all codes (14-15 SF16 codes) allocated for handling HS codes.

Typical algorithm used by Huawei RNC/NodeB for dynamic codes allocation does not crossing maximal 10 codes even if you have 15 codes configured in parameters.

The higher throughput you can achieve only if you configure fixed 14 HS codes. Why 14 and not 15? with 15 codes the HSUPA will be problematic since EDCH channels required additional codes from second SF16 code tree.

Another problem can be the user limitation for max throughput, which is defined at HLR.

In order to verify cell throughput you can monitor Cell Throughput and Codes allocation via LMT