calculation mode of wideband_SINR of UE

Viewing 8 posts - 1 through 8 (of 8 total)
  • Author
    Posts
  • #4135
    xingtong
    Participant

    Hi,all
    When I run the same LTE_sim_main_luancher_*.m with different trace_version(v1 & v2),
    I found the wideband_SINR value of the UE are quite different at same TTI.(PS:SNR_avg_preequal values are same)
    v1: the wideband_SINR value ranges from 0dB ~ 20dB
    calculation mode:
    obj.wideband_SINR = 10*log10(sum(RX_total(:))/(sum(interf_power_all_RB(:))+thermal_noise_watts_per_half_RB*nSC));
    v2(runtime precoding): the wideband_SINR value ranges from -20dB ~ 0dB
    calculation mode:
    obj.wideband_SINR = 10*log10(sum(reshape(RX_total_half_RB_layers_UE/avg_pathloss_linear_0/avg_shadow_fading_linear_0,1,[])) /…
    (sum(reshape(RX_W_half_RB_i_UE./avg_pathloss_lin_i_calc(pathloss_idxs)./avg_pathloss_shadow_i_calc(pathloss_idxs),1,[]))…
    +noise_W_half_RB*numel(RX_total_half_RB_layers_UE)));
    Why are the values so different?Could you help me understand it?
    PS: You can find the above code in +network_elements/UE.m

    Best regards,
    XingTong

    #4165

    Dear XingTong,

    we also experienced this behaviour. As far as I remember, this was a bug with the scale [linear or logarithmic] of the path loss and the shadowing. Please set breakpoints at the calculation of RX_total and interf_power_all_RB in v1, and avg_pathloss_linear_0, avg_shadow_fading_linear_0, avg_pathloss_lin_i_calc, avg_pathloss_shadow_i_calc in v2, and check whether they are in linear or dB scale.

    Best regards,
    Martin.

    #4172
    xingtong
    Participant

    Hi,Dear Martin Taranetz
    Thank you for your attention.I did as you told me.I find these variables are in linear scale which seems to be right.But with your kind guidance,I found avg_pathloss_lin_i_calc(pathloss_idxs) in v2 was wrong because pathloss_idxs were all the same.That is to say, different interference suffers the same passloss.Then I paid attention to pathloss_idxs.Its calculation is shown as follows.
    ——————–before———————–

    % With this calculation, size(RX_W_half_RB_i_UE,3) is 1 which must be unreasonable.
    pathloss_idxs = zeros(size(RX_W_half_RB_i_UE));
    for i_=1:size(RX_W_half_RB_i_UE,3)
        pathloss_idxs(:,:, i_ ,<strong>:</strong>) = i_;
    end

    ——————–later————————

    pathloss_idxs = zeros(size(RX_W_half_RB_i_UE));
    for i_=1:size(RX_W_half_RB_i_UE,4)
        pathloss_idxs(:,:,:,i_) = i_;
    end

    ————————————————-
    With later calculation,the wideband_SINR value of the UE becomes normal.

    Best regards,
    Xing.

    • This reply was modified 5 years, 5 months ago by xingtong.
    • This reply was modified 5 years, 5 months ago by xingtong.
    • This reply was modified 5 years, 5 months ago by xingtong.
    #4426
    ramdharma
    Participant

    Hi, I want to work with BCQI algorithm, where I would like to use Highest CQI is 8, that is instead of that 1-15, i want to run the values between 1-8, so, is it possible, if yes, tell me the ways

    #4452

    Dear XingTong,

    For calculation of the wideband_SINR in v2 trace there was a bug in the code. The right expression is:

    obj.wideband_SINR = 10*log10(sum(reshape(RX_total_half_RB_layers_UE./avg_pathloss_linear_0./avg_shadow_fading_linear_0,1,[])) /…
    (sum(reshape(RX_W_half_RB_i_UE./10^(-2)./avg_pathloss_lin_i_calc(pathloss_idxs)./avg_pathloss_shadow_i_calc(pathloss_idxs),1,[]))…
    +noise_W_half_RB*numel(RX_total_half_RB_layers_UE)));

    Regarding the pathloss indexing you should use the ‘later’ calculation as:

    % Indexing matrix for the pathlosses for wideband SINR calculation
    pathloss_idxs = zeros(size(RX_W_half_RB_i_UE));
    for i_=1:size(RX_W_half_RB_i_UE,4)
    pathloss_idxs(:,:,:,i_, 🙂 = i_;
    end

    Note that the pathloss maps are generated per site and not per sector, this leads to the same pathloss value for three sectors, that is why interfering eNodeB sectors of the same site have the same pathloss.

    Best regards,
    Fjolla.

    #4476
    xingtong
    Participant

    Dear Fjolla Ademaj,
    I wonder why RX_W_half_RB_i_UE should divide 10^(-2),can you explain it?

    Best regards,
    Xing.

    #4477
    xingtong
    Participant

    Dear Fjolla Ademaj,
    I wonder why RX_W_half_RB_i_UE should divide 10^(-2),can you explain it?

    Best regards,
    Xing.

    #4497
    ramdharma
    Participant

    Hi.. I would like to know how bcqi assigns the RB to the user. What are all the requirements the user should posses?
    is enb chooses the received feedback or CQI sent of users. I have confusion with this. So, Kindly help me to get over this.
    even in Prop fair also i have the same problem…..

Viewing 8 posts - 1 through 8 (of 8 total)
  • You must be logged in to reply to this topic.