Data Segments and Channels

Time-series segmentation

In order to make the algorithm faster, the time-series is divided into smaller segments with a length in seconds set by the segment_length parameter. The segmentation can also be done such that consecutive segments can overlap, thereby removing filtering artifacts and edge effects during the construction of the power spectral density and the excess power search. The length in seconds by which 2 consecutive segments are separated is set by the segment_stride parameter. In some part of the algorithm, the parameters will be used in sample units (not seconds). The conversion can be done straightforwardly using the sample_rate metadata from the ts_data time-series object as follows:

# Define segment length for PSD estimation in sample unit
seg_len = int(segment_length * ts_data.sample_rate)
# Define separation between consecutive segments in sample unit
seg_stride = int(segment_stride * ts_data.sample_rate)

Narrow band channels

We then check if the maximum frequency in the filter bank is defined, that is the frequency of the last filter from the filter bank. If the value is not defined, we set it to be equal to the Nyquist frequency, that is half the sampling rate, which makes sense as any larger signal would not be identifiable:

# Check if tile maximum frequency is not defined
if fmax is None or fmax>sample_rate/2.:
    # Set the tile maximum frequency equal to the Nyquist frequency
    fmax = sample_rate / 2.0

Depending whether the bandwidth of the finest filter, band , or the total number of channels, nchans , is defined, one can then determined the filter settings easily as the bandwidth of the finest filter effectively corresponds to the bandwidth of one individual channel. Therefore, if the bandwidth of the finest filter is defined, the total number of channels is set to be equal to the total frequency band divived by a single channel’s bandwidth. Conversely, if the channel’s bandwidth is not defined but the total number of channels is, the smallest bandwidth will therefore correspond to the frequency band divided by the number of channels:

# Check whether or not tile bandwidth and channel are defined
if band is None and nchans is None:
    # Exit program with error message
    exit("Either bandwidth or number of channels must be specified...")
else:
    # Check if tile maximum frequency larger than its minimum frequency
    assert fmax >= fmin
    # Define spectral band of data
    data_band = fmax - fmin
    # Check whether tile bandwidth or channel is defined
    if band is not None:
        # Define number of possible filter bands
        nchans = int(data_band / band)
    elif nchans is not None:
        # Define filter bandwidth
        band = data_band / nchans
    # Check if number of channels is superior than unity
    assert nchans > 1
    nchans -= 1