icebear.processing

icebear.processing.process

icebear.processing.process.append_level1_hdf5(filename, hour, minute, second, data_flag, doppler_shift, rf_distance, snr_db, noise, spectra, spectra_variance, spectra_noise, spectra_clutter_corr, xspectra, xspectra_variance, xspectra_noise, xspectra_clutter_corr)

Append spectra and cross-spectra ICEBEAR data to previously created HDF5 file

Parameters

• filename –

• hour –

• minute –

• second –

• data_flag –

• doppler_shift –

• rf_distance –

• snr_db –

• noise –

• spectra –

• spectra_variance –

• spectra_noise –

• spectra_clutter_corr –

• xspectra –

• xspectra_variance –

• xspectra_noise –

• xspectra_clutter_corr –

Returns

Todo – Make sure everything is put into numpy arrays

icebear.processing.process.create_level1_hdf5(config, filename, year, month, day)

Create a level1 HDF5 formatted file for storage of ICEBEAR spectra and cross-spectra

Parameters

• config (Class Object) – Config class instantiation.

• filename –

• year –

• month –

• day –

Returns

None

Todo

Make sure everything is put into numpy arrays

icebear.processing.process.decx(config, time, data, bcode, channel1, channel2, correction1, correction2)

Performs cross-correlation and decimation for inputed baseline from the radar data

Notes

• ssmfx CUDA can only handle number_ranges = 2000 exactly. For farther ranges we loop at step size 2000.

• Currently the rea_vector command is resulting in an error at the end of execution. This oes not appear to affect the output of the script. Issue may be in h5py or digital_rf. This error only appears when using python3

icebear.processing.process.func()

Python wrapper for GPU CUDA code in the libssmf.so file.

Parameters

* –

Returns

func (wrapper) – Python function wrapper with inputs identical to wrapped CUDA function in the ssmf.cu file.

Noes:

• Wrapped function is ssmf.cu is denoted by extern “C” {} tag

• Inputs are (cufftComplex *meas1, cufftComplex *meas2, cufftComplex *code, cufftComplex *result, size_t measlen, size_t codelen, size_t size, ing avg, ing check)

icebear.processing.process.generate_bcode(filepath)

Uses the pseudo-random code file to generate the binary code for signal matching

todo

Make function able to take any code length and resample at any rate

icebear.processing.process.generate_level1(config)

Parameters

config –

icebear.processing.process.ssmf(meas, code, averages, nrang, fdec, codelen)

Formats measured data and CUDA function inputs and calls wrapped function to determine single antenna spectra.

Parameters

• meas (complex64 np.array) – Antenna voltages loaded from HDF5 with phase and magnitude corrections.

• code (float32 np.array) – Transmitted psuedo-random code sequence.

• averages (int) – The number of 0.1 second averages to be performed on the GPU.

• nrang (int) – Number of range gates being processed. Nominally 2000.

• fdec (int) – Decimation rate to be used by GPU processing, effects Doppler resolution. Nominally 200.

• codelen (int) – Length of the transmitted psuedo-random code sequence.

Returns

S (complex64 np.array) – 2D Spectrum output for antenna (Doppler shift x Range).

Notes

• ssmf.cu could be modified to allow code to be a float32 input. This would reduce memory requirements on the GPU.

• ‘check’ input of __fmed is 0 which indicates a single antenna is being processed.

icebear.processing.process.ssmfx(meas0, meas1, code, averages, nrang, fdec, codelen)

Formats measured data and CUDA function inputs and calls wrapped function for determining the cross-correlation spectra of selected antenna pair.

Parameters

• meas0 (complex64 np.array) – First antenna voltages loaded from HDF5 with phase and magnitude corrections.

• meas1 (complex64 np.array) – Second antenna voltages loaded from HDF5 with phase and magnitude corrections.

• code (float32 np.array) – Transmitted psuedo-random code sequence.

• averages (int) – The number of 0.1 second averages to be performed on the GPU.

• nrang (int) – Number of range gates being processed. Nominally 2000.

• fdec (int) – Decimation rate to be used by GPU processing, effects Doppler resolution. Nominally 200.

• codelen (int) – Length of the transmitted psuedo-random code sequence.

Returns

S (complex64 np.array) – 2D Spectrum output for antenna pair (Doppler shift x Range).

Notes

• ssmf.cu could be modified to allow code to be a float32 input. This would reduce memory requirements on the GPU.

• ‘check’ input of __fmed is 1 which indicates a pair of antennas is being processed.