Looking Assistance for Data Acquisition and Parallel Processing with USRP B200 in MATLAB

Hello dear MATLAB Community,

I am looking for a solution to receive and save data from the USRP B200 using the USRP Communicatios Toolbox and Parallel Computing Toolbox. When I try to save data using a parallel pool, I experience a delay of 2-3 seconds, during which I believe some samples are lost.

I am seeking a solution that allows my program to continuously receive data and, after a certain number of samples or a specific time period, save it using a parallel process.

Below is a part of my code. My system specifications include an 8-core CPU, 32 GB of RAM, and the latest version of MATLAB R2023a.

Thank you very much!

%% Create a parallel pool
poolobj = gcp('nocreate');  % If no pool, do not create new one.
if isempty(poolobj)
    poolobj = parpool(6); % Create a new pool
end

%% Data acquisition loop
data = complex(zeros(round(receiveSettings.samples_per_frame), round(calculatedParameters.corr_iterations)));
if hardware.radioFound
    % Capture loop
    for o_frame = 1 : calculatedParameters.capture_number
        while i_frame < calculatedParameters.corr_iterations
        [output, dataLen, overrun, timestamp] = radio();
    
        % Overrun check
        if overrun == 1
            overrun_counter = overrun_counter + 1;
        end
    
        % Validate received data
        if dataLen >= receiveSettings.samples_per_frame
            step(scope, output);            % Display frequency spectrum
            data(:, i_frame) = output;      % Save data from USRP
            i_frame = i_frame + 1;
        end
    end    
    tic
    f = parfeval(poolobj, @saveDataAsync, 0, data, calculatedParameters, recorderSettings, usrpSettings, o_frame); % call save fuction
    toc
end

%% Function that handles the saving of data to a WAV file
function saveDataAsync(data, calculatedParameters, recorderSettings, usrpSettings, o_frame)
    data = reshape(data, [], 1);                           % Reshape 'data' into a column vector
    % Normalize IQ data
    data_I = real(data) / max(abs(real(data)));
    data_Q = imag(data) / max(abs(imag(data)));
    data_cat = [data_I, data_Q];    
    % Prepare filename for WAV saving
    wavefile_IQ = sprintf('%s_%dHz_MCR_%dHz_Cfr_%dDf_%d', char(recorderSettings.wav_time), usrpSettings.master_clock_rate, usrpSettings.center_frequency, usrpSettings.descimationfactor, o_frame);
    folderPath = 'D:\test';
    filePath = fullfile(folderPath, [wavefile_IQ, '.wav']);
    wavwrite_personal_function(data_cat, calculatedParameters.sample_rate, recorderSettings.rcvr, recorderSettings.nbits, filePath);
    disp('Audio file saved successfully.');
end

[Mario Malic]

I don't know anything about your device, but are you sure parpool is necessary? This is what makes your execution slow.

[Nik Rocky]

Do you have a better suggestion? When I perform data reception and storage consecutively, it happens that after receiving about 500MB of data, the writing process into a WAV file causes a delay of about 5 seconds before data can be received again. Therefore, my idea was to continuously receive data and let the data storage run as a parallel process.

[Mario Malic]

Can you get me the link for documentation on radio function?

[Nik Rocky]

Yes, Sure: https://de.mathworks.com/help/supportpkg/usrpradio/ug/sdrureceiver.html#d126e7863