Пример #1
0
  def __init__(self, params):
    """
    Initialize the parameters, which are common across different
    types of tracking channels.

    Parameters
    ----------
    params : dictionary
      The subset of tracking channel parameters that are deemed
      to be common across different types of tracking channels.


    """
    for (key, value) in params.iteritems():
      setattr(self, key, value)

    self.prn = params['acq'].prn
    self.signal = params['acq'].signal

    self.results_num = 500
    self.stage1 = True

    self.lock_detect = LockDetector(
        k1=self.lock_detect_params["k1"],
        k2=self.lock_detect_params["k2"],
        lp=self.lock_detect_params["lp"],
        lo=self.lock_detect_params["lo"])

    self.alias_detect = AliasDetector(
        acc_len=defaults.alias_detect_interval_ms / self.coherent_ms,
        time_diff=1)

    self.cn0_est = CN0Estimator(
        bw=1e3 / self.coherent_ms,
        cn0_0=self.cn0_0,
        cutoff_freq=0.1,
        loop_freq=self.loop_filter_params["loop_freq"]
    )

    self.loop_filter = self.loop_filter_class(
        loop_freq=self.loop_filter_params['loop_freq'],
        code_freq=self.code_freq_init,
        code_bw=self.loop_filter_params['code_bw'],
        code_zeta=self.loop_filter_params['code_zeta'],
        code_k=self.loop_filter_params['code_k'],
        carr_to_code=self.loop_filter_params['carr_to_code'],
        carr_freq=self.acq.doppler,
        carr_bw=self.loop_filter_params['carr_bw'],
        carr_zeta=self.loop_filter_params['carr_zeta'],
        carr_k=self.loop_filter_params['carr_k'],
        carr_freq_b1=self.loop_filter_params['carr_freq_b1'],
    )

    self.next_code_freq = self.loop_filter.to_dict()['code_freq']
    self.next_carr_freq = self.loop_filter.to_dict()['carr_freq']

    self.track_result = TrackResults(self.results_num,
                                     self.acq.prn,
                                     self.acq.signal)
    self.alias_detect_init = 1
    self.code_phase = 0.0
    self.carr_phase = 0.0
    self.samples_per_chip = int(round(self.sampling_freq / self.chipping_rate))
    self.sample_index = params['samples']['sample_index']
    self.sample_index += self.acq.sample_index
    self.sample_index += self.acq.code_phase * self.samples_per_chip
    self.sample_index = int(math.floor(self.sample_index))
    self.carr_phase_acc = 0.0
    self.code_phase_acc = 0.0
    self.samples_tracked = 0
    self.i = 0

    self.pipelining = False    # Flag if pipelining is used
    self.pipelining_k = 0.     # Error prediction coefficient for pipelining
    self.short_n_long = False  # Short/Long cycle simulation
    self.short_step = True    # Short cycle
    if self.tracker_options:
      mode = self.tracker_options['mode']
      if mode == 'pipelining':
        self.pipelining = True
        self.pipelining_k = self.tracker_options['k']
      elif mode == 'short-long-cycles':
        self.short_n_long = True
        self.pipelining = True
        self.pipelining_k = self.tracker_options['k']
      else:
        raise ValueError("Invalid tracker mode %s" % str(mode))
Пример #2
0
class TrackingChannel(object):
  """
  Tracking channel base class.
  Specialized signal tracking channel classes are subclassed from
  this class. See TrackingChannelL1CA or TrackingChannelL2C as
  examples.

  Sub-classes can optionally implement :meth:'_run_preprocess',
  :meth:'_run_postprocess' and :meth:'_get_result' methods.

  The class is designed to support batch processing of sample data.
  This is to help processing of large data sample files without the need
  of loading the whole file into a memory.
  The class instance keeps track of the next sample to be processed
  in the form of an index within the original data file.
  Each sample data batch comes with its starting index within the original
  data file. Given the starting index of the batch and its own index
  of the next sample to be processed, the code computes the offset
  within the batch and starts/continues the tracking procedure from there.

  """

  def __init__(self, params):
    """
    Initialize the parameters, which are common across different
    types of tracking channels.

    Parameters
    ----------
    params : dictionary
      The subset of tracking channel parameters that are deemed
      to be common across different types of tracking channels.


    """
    for (key, value) in params.iteritems():
      setattr(self, key, value)

    self.prn = params['acq'].prn
    self.signal = params['acq'].signal

    self.results_num = 500
    self.stage1 = True

    self.lock_detect = LockDetector(
        k1=self.lock_detect_params["k1"],
        k2=self.lock_detect_params["k2"],
        lp=self.lock_detect_params["lp"],
        lo=self.lock_detect_params["lo"])

    self.alias_detect = AliasDetector(
        acc_len=defaults.alias_detect_interval_ms / self.coherent_ms,
        time_diff=1)

    self.cn0_est = CN0Estimator(
        bw=1e3 / self.coherent_ms,
        cn0_0=self.cn0_0,
        cutoff_freq=0.1,
        loop_freq=self.loop_filter_params["loop_freq"]
    )

    self.loop_filter = self.loop_filter_class(
        loop_freq=self.loop_filter_params['loop_freq'],
        code_freq=self.code_freq_init,
        code_bw=self.loop_filter_params['code_bw'],
        code_zeta=self.loop_filter_params['code_zeta'],
        code_k=self.loop_filter_params['code_k'],
        carr_to_code=self.loop_filter_params['carr_to_code'],
        carr_freq=self.acq.doppler,
        carr_bw=self.loop_filter_params['carr_bw'],
        carr_zeta=self.loop_filter_params['carr_zeta'],
        carr_k=self.loop_filter_params['carr_k'],
        carr_freq_b1=self.loop_filter_params['carr_freq_b1'],
    )

    self.next_code_freq = self.loop_filter.to_dict()['code_freq']
    self.next_carr_freq = self.loop_filter.to_dict()['carr_freq']

    self.track_result = TrackResults(self.results_num,
                                     self.acq.prn,
                                     self.acq.signal)
    self.alias_detect_init = 1
    self.code_phase = 0.0
    self.carr_phase = 0.0
    self.samples_per_chip = int(round(self.sampling_freq / self.chipping_rate))
    self.sample_index = params['samples']['sample_index']
    self.sample_index += self.acq.sample_index
    self.sample_index += self.acq.code_phase * self.samples_per_chip
    self.sample_index = int(math.floor(self.sample_index))
    self.carr_phase_acc = 0.0
    self.code_phase_acc = 0.0
    self.samples_tracked = 0
    self.i = 0

    self.pipelining = False    # Flag if pipelining is used
    self.pipelining_k = 0.     # Error prediction coefficient for pipelining
    self.short_n_long = False  # Short/Long cycle simulation
    self.short_step = True    # Short cycle
    if self.tracker_options:
      mode = self.tracker_options['mode']
      if mode == 'pipelining':
        self.pipelining = True
        self.pipelining_k = self.tracker_options['k']
      elif mode == 'short-long-cycles':
        self.short_n_long = True
        self.pipelining = True
        self.pipelining_k = self.tracker_options['k']
      else:
        raise ValueError("Invalid tracker mode %s" % str(mode))

  def dump(self):
    """
    Append intermediate tracking results to a file.

    """
    fn_analysis, fn_results = self.track_result.dump(self.output_file, self.i)
    self.i = 0
    return fn_analysis, fn_results

  def start(self):
    """
    Start tracking channel.
    For the time being only prints an informative log message about
    the initial parameters of the tracking channel.

    """

    logger.info("[PRN: %d (%s)] Tracking is started. "
                "IF: %.1f, Doppler: %.1f, code phase: %.1f, "
                "sample index: %d" %
                (self.prn + 1,
                 self.signal,
                 self.IF,
                 self.acq.doppler,
                 self.acq.code_phase,
                 self.acq.sample_index))

  def get_index(self):
    """
    Return index of next sample to be processed by the tracking channel.
    The tracking channel is designed to process the input data samples
    in batches. A single batch is fed to multiple tracking channels.
    To keep track of the order of samples within one tracking channel,
    each channel maintains an index of the next sample to be processed.
    This method is a getter method for the index.

    Returns
    -------
    sample_index: integer
      The next data sample to be processed.

    """
    return self.sample_index

  def _run_preprocess(self):
    """
    Customize the tracking run procedure in a subclass.
    The method can be optionally redefined in a subclass to perform
    a subclass specific actions to happen before correlator runs
    next integration round.

    """
    pass

  def _run_postprocess(self):
    """
    Customize the tracking run procedure in a subclass.
    The method can be optionally redefined in a subclass to perform
    a subclass specific actions to happen after correlator runs
    next integration round.

    """
    pass

  def _get_result(self):
    """
    Customize the tracking run procedure outcome in a subclass.
    The method can be optionally redefined in a subclass to return
    a subclass specific data as a result of the tracking procedure.

    Returns
    -------
    out :
      None is returned by default.

    """
    return None

  def _short_n_long_preprocess(self):
    pass

  def _short_n_long_postprocess(self):
    pass

  def is_pickleable(self):
    """
    Check if object is pickleable.
    The base class instance is always pickleable.
    If a subclass is not pickleable, then it should redefine the method
    and return False.
    The need to know if an object is pickleable or not arises from the fact
    that we try to run the tracking procedure for multiple tracking channels
    on multiple CPU cores, if more than one core is available.
    This is done to speed up the overall processing time. When a tracking
    channel runs on a separate CPU core, it also runs on a separate
    process. When the tracking of the given batch of data is over, the process
    exits and the tracking channel state is returned to the parent process.
    This requires serialization (pickling) of the tracking object state,
    which might not be always trivial. This method essentially defines
    if the tracking channels can be run in a separate processs.
    If the object is not pickleable, then the tracking for the channel is
    done on the same CPU, which runs the parent process. Therefore all
    non-pickleable tracking channels are processed sequentially.

    Returns
    -------
    out : bool
      True if the object is pickleable, False - if not.

    """
    return True

  def run(self, samples):
    """
    Run tracking channel for the given batch of data.
    This method is an entry point for the tracking procedure.
    Subclasses normally will not redefine the method, but instead
    redefine the customization methods '_run_preprocess', '_run_postprocess'
    and '_get_result' to run signal specific tracking operations.

    Parameters
    ----------
    sample : dictionary
      Sample data. Sample data are provided in batches

    Return
    ------
      The return value is determined by '_get_result' customization method,
      which can be redefined in subclasses

    """

    self.samples = samples

    if self.sample_index < samples['sample_index']:
      raise ValueError("Incorrect samples offset")

    sample_index = self.sample_index - samples['sample_index']
    samples_processed = 0
    samples_total = len(samples[self.signal]['samples'])

    estimated_blksize = self.coherent_ms * self.sampling_freq / 1e3

    self.track_result.status = 'T'

    while self.samples_tracked < self.samples_to_track and \
            (sample_index + 2 * estimated_blksize) < samples_total:

      self._run_preprocess()

      if self.pipelining:
        # Pipelining and prediction
        corr_code_freq = self.next_code_freq
        corr_carr_freq = self.next_carr_freq

        self.next_code_freq = self.loop_filter.to_dict()['code_freq']
        self.next_carr_freq = self.loop_filter.to_dict()['carr_freq']

        if self.short_n_long and not self.stage1 and not self.short_step:
          # In case of short/long cycles, the correction applicable for the
          # long cycle is smaller proportionally to the actual cycle size
          pipelining_k = self.pipelining_k / (self.coherent_ms - 1)
        else:
          pipelining_k = self.pipelining_k

        # There is an error between target frequency and actual one. Affect
        # the target frequency according to the computed error
        carr_freq_error = self.next_carr_freq - corr_carr_freq
        self.next_carr_freq += carr_freq_error * pipelining_k

        code_freq_error = self.next_code_freq - corr_code_freq
        self.next_code_freq += code_freq_error * pipelining_k

      else:
        # Immediate correction simulation
        self.next_code_freq = self.loop_filter.to_dict()['code_freq']
        self.next_carr_freq = self.loop_filter.to_dict()['carr_freq']

        corr_code_freq = self.next_code_freq
        corr_carr_freq = self.next_carr_freq

      coherent_iter, code_chips_to_integrate = self._short_n_long_preprocess()

      for _ in range(self.coherent_iter):

        if (sample_index + 2 * estimated_blksize) >= samples_total:
          break

        samples_ = samples[self.signal]['samples'][sample_index:]

        E_, P_, L_, blksize, self.code_phase, self.carr_phase = self.correlator(
            samples_,
            code_chips_to_integrate,
            corr_code_freq + self.chipping_rate, self.code_phase,
            corr_carr_freq + self.IF, self.carr_phase,
            self.prn_code,
            self.sampling_freq,
            self.signal
        )

        if blksize > estimated_blksize:
          estimated_blksize = blksize

        sample_index += blksize
        samples_processed += blksize
        self.carr_phase_acc += corr_carr_freq * blksize / self.sampling_freq
        self.code_phase_acc += corr_code_freq * blksize / self.sampling_freq

        self.E += E_
        self.P += P_
        self.L += L_

      more_integration_needed = self._short_n_long_postprocess()
      if more_integration_needed:
        continue

      # Update PLL lock detector
      lock_detect_outo, \
          lock_detect_outp, \
          lock_detect_pcount1, \
          lock_detect_pcount2, \
          lock_detect_lpfi, \
          lock_detect_lpfq = self.lock_detect.update(self.P.real,
                                                     self.P.imag,
                                                     coherent_iter)

      if lock_detect_outo:
        if self.alias_detect_init:
          self.alias_detect_init = 0
          self.alias_detect.reinit(defaults.alias_detect_interval_ms /
                                   self.coherent_iter,
                                   time_diff=1)
          self.alias_detect.first(self.P.real, self.P.imag)
        alias_detect_err_hz = \
            self.alias_detect.second(self.P.real, self.P.imag) * np.pi * \
            (1e3 / defaults.alias_detect_interval_ms)
        self.alias_detect.first(self.P.real, self.P.imag)
      else:
        self.alias_detect_init = 1
        alias_detect_err_hz = 0

      self.loop_filter.update(self.E, self.P, self.L)
      self.track_result.coherent_ms[self.i] = self.coherent_ms

      self.track_result.IF = self.IF
      self.track_result.carr_phase[self.i] = self.carr_phase
      self.track_result.carr_phase_acc[self.i] = self.carr_phase_acc
      self.track_result.carr_freq[self.i] = \
          self.loop_filter.to_dict()['carr_freq'] + self.IF

      self.track_result.code_phase[self.i] = self.code_phase
      self.track_result.code_phase_acc[self.i] = self.code_phase_acc
      self.track_result.code_freq[self.i] = \
          self.loop_filter.to_dict()['code_freq'] + self.chipping_rate

      # Record stuff for postprocessing
      self.track_result.absolute_sample[self.i] = self.sample_index + \
          samples_processed

      self.track_result.E[self.i] = self.E
      self.track_result.P[self.i] = self.P
      self.track_result.L[self.i] = self.L

      self.track_result.cn0[self.i] = self.cn0_est.update(
          self.P.real, self.P.imag)

      self.track_result.lock_detect_outo[self.i] = lock_detect_outo
      self.track_result.lock_detect_outp[self.i] = lock_detect_outp
      self.track_result.lock_detect_pcount1[self.i] = lock_detect_pcount1
      self.track_result.lock_detect_pcount2[self.i] = lock_detect_pcount2
      self.track_result.lock_detect_lpfi[self.i] = lock_detect_lpfi
      self.track_result.lock_detect_lpfq[self.i] = lock_detect_lpfq

      self.track_result.alias_detect_err_hz[self.i] = alias_detect_err_hz

      self._run_postprocess()

      self.samples_tracked = self.sample_index + samples_processed
      self.track_result.ms_tracked[self.i] = self.samples_tracked * 1e3 / \
          self.sampling_freq

      self.i += 1
      if self.i >= self.results_num:
        self.dump()

    if self.i > 0:
      self.dump()

    self.sample_index += samples_processed

    return self._get_result()