BLER vs MCS

[abdulkadir köse]

Hi,


Some test cases have been generated under phy error model in ns3-model-library.pdf. The 2 of them as follows:

1. UE placed 1800 meters far from the eNB, which implies the use of MCS 2 (SINR of -5.51 dB) and a TB of
1088 bits, that in turns produce a BLER of 0.02 (see point C in figure BLER for tests 1, 2, 3.).

4. UE placed 600 meters far from the eNB, which implies the use of MCS 12 (SINR of 4.43 dB) and a TB of
4800 bits, that in turns produce a BLER of 0.3 (see point D in figure BLER for tests 4, 5.).

I run lte-test-phy-error-model.cc for 2 cases above  the results as follows :

• Test with 1 user(s) at distance 1800 expected BLER 0.02

              User 0 imsi 1 DOWNLINK pkts rx 986 tx 1000 BLER 0.014 Err 0.006 expected rx 980 difference 6 tolerance 33

• Test with 1 user(s) at distance 600 expected BLER 0.3

              User 0 imsi 1 DOWNLINK pkts rx 696 tx 1000 BLER 0.304 Err 0.004 expected rx 700 difference 4 tolerance 38



 Why we see higher BLER for higher MCS?  (I would expect that received packets for UE at high distance should be less than for UE at low distance) .
 
 Did I misunderstand the BLER (ratio of the number of erroneous blocks received to the total number of blocks sent) ?

Thanks in advance.

Regards,
Kadir

[Tommaso Pecorella]

Hi,

you misunderstood (or forgot) one thing.

The higher the MCS, the higher the [potential] bitrate.

What the algorithms do is to use an MCS that keeps the the BLER below a given threshold, preferring an higher bitrate over a lower one.

T.

[abdulkadir köse]

Thanks a lot Tommaso, it becomes more clear for me.

Regard,
Kadir


19 Mayıs 2017 Cuma 22:23:38 UTC+1 tarihinde Tommaso Pecorella yazdı: