none
channel coefficient in umxsdrbrick

    Question

  • Hello,

    I have a quick question on the channel estimation. Specifically, the channel coefficient.
    We are working with the 802.11a and trying to get access to this information. 
    The only place we have found is in the vector variable "BB11aDemodCtx.CF_Channel_11a::ChannelCoeffs()".
    It is a vector containing 64 pairs of real and imaginary number.
    Below is the code segment we added to "mac.cpp" to print out each real and imaginary component in this vector.

    printf ( "===>channels<===\n" );
    for (int i = 0; i < 64; i++) 
    {
    printf ( "%d %d\n", 
    BB11aDemodCtx.CF_Channel_11a::ChannelCoeffs()[i/4][i%4].re, 
    BB11aDemodCtx.CF_Channel_11a::ChannelCoeffs()[i/4][i%4].im);
    }

    However, the result does not seem right. Please take a look at the following output.
    Numbers like 31936, 8000, -19200, -24000, -11200 and 12266 are not quite right in this case.

    So the question is:
    Did we do something wrong? and How should we get access to such information.

    Thanks
    Muye

    MACAddress: 02-50-F2-5A-95-09
    ===>channels<===
    -201 -40
    807 1230
    286 1669
    -933 1228
    -1329 907
    -2215 -393
    -1075 -1280
    -93 -1862
    1419 -1819
    1483 -680
    2095 495
    998 1347
    220 2740
    -1200 2162
    -2640 1140
    -2208 -131
    -1085 -2537
    -1066 -4133
    871 -2257
    2485 -2825
    4723 -3047
    16000 -5866
    5760 10560
    -355 8533
    669 3869
    -1200 3709
    -3105 1819
    31936 8000
    0 0
    0 0
    0 0
    0 0
    0 0
    0 0
    0 0
    0 0
    -19200 -24000
    9280 7360
    -816 178
    -819 -248
    -438 -731
    62 -841
    633 -666
    963 -257
    824 263
    600 790
    -2 971
    -683 837
    -852 406
    -878 -229
    -584 -826
    -105 -983
    589 -699
    981 -267
    1023 299
    620 880
    -104 1171
    -789 905
    -1501 -366
    -1099 -689
    -896 -972
    77 -1625
    654 -1050
    1186 -457
    ===>channels<===
    -211 17
    -409 -1143
    727 -1084
    1408 171
    1005 1046
    -71 1564
    -1059 735
    -1510 -419
    -1205 -1138
    79 -1806
    843 -1288
    1240 568
    1202 1422
    -1323 1466
    -1459 1038
    -1548 -800
    -100 -1714
    952 -2552
    2654 -636
    1951 2048
    525 3017
    -1645 1871
    -3062 114
    -2478 -580
    -2026 -2640
    81 -3335
    1714 -2209
    0 0
    0 0
    0 0
    0 0
    0 0
    0 0
    0 0
    0 0
    0 0
    -11200 12266
    -2947 -5136
    534 755
    -216 942
    -887 469
    -784 -408
    -335 -936
    409 -819
    811 -231
    934 423
    226 872
    -556 714
    -1022 145
    -722 -564
    33 -849
    669 -596
    929 117
    583 679
    -211 1012
    -755 682
    -888 -370
    -336 -1056
    298 -734
    900 -534
    884 429
    233 1182
    -535 894
    -1119 611
    Wednesday, December 18, 2013 12:24 AM

Answers

  • Hi,

    Thanks for the post. Would you please specify your question more clear? Why do you think the numbers are not right?

    I am not sure what exactly information you need. The channel coeffs here are actually the equalization coeffs. Say, taking the simple signal propagation model,

    y = h x. The coeffs you get through the brick is 1/h, instead of h itself.

    Hope that helps.

    Thanks,

    - Kun 

    • Marked as answer by MUYE20 Monday, December 23, 2013 8:43 PM
    Monday, December 23, 2013 5:46 AM
    Owner

All replies

  • Not sure what was the root cause. Do you need any extra information? Muye
    Friday, December 20, 2013 3:28 AM
  • Hi,

    Thanks for the post. Would you please specify your question more clear? Why do you think the numbers are not right?

    I am not sure what exactly information you need. The channel coeffs here are actually the equalization coeffs. Say, taking the simple signal propagation model,

    y = h x. The coeffs you get through the brick is 1/h, instead of h itself.

    Hope that helps.

    Thanks,

    - Kun 

    • Marked as answer by MUYE20 Monday, December 23, 2013 8:43 PM
    Monday, December 23, 2013 5:46 AM
    Owner
  • Hello Kun,

    Thanks for your information.

    In this case, we need h instead of 1/h. So, just wondering if there is any existing Sora code we can use to invert 1/h back to h?

    Thanks and Merry Christmas
    Muye 


    Monday, December 23, 2013 8:43 PM