This list is a compilation of additional plots which can be included
with NoPi Display to make it functionally similar to LoadPi:

  o Polar Plots of satellites tracked and tracking C/No on a per combo
  basis
  o Base and Rover common SV tracking plots on a per combo basis along
  with an indicator showing the reference satellite used at each epoch
  o Base minus Rover C/No variations on a per combo basis 
  o Absolute C/No versus Time and Absolute C/No versus elevation on a
    per combo basis.

This list is a compilation of analysis and validation functions which
would be nice to have within the NoPi Data Checker tool. This list is
based on suggestions from the Stinger group members and Nick Talbot.

  o Formulating a performance metric to characterize the reacquisition
  performance of a receiver. This is particularly useful for data from
  the Antenna switch receiver.
    * Perform statistical analysis of the first epoch and the first
      second after reacquiring a signal. Compare with a one second
      period in steady state, 5 seconds after reacquisition. Similarly
      do 1 second 5 seconds before losing lock, the last second and
      last epoch before losing lock.
    * Consider putting an event marker in the T02 file when the
      antenna switch disconnects and reconnects the antenna. This will
      allow better time alignment of the statistics.
    * Quantify the distribution of the reacquisition data into three
      separate histograms: Data which are modulo 200 ms (full
      measurements); All other data points (demi measurements) and a
      combination of the two (all measurements). 

  o Single Receiver Analysis:
    * Ionospheric Phase Combination: useful to detect "undetected"
      cycle slips in Stinger (phase1*wavelength1 - phase2*wavelength2)
    * Ionospheric Code Combination 
    * Iono-free Phase and Code Combination
    * Stationary Range Single Frequency = pseudorange + phase*wavelength
      (code minus carrier)
    * Widelane Stationary Range: <verify this with Nick/Gang, and
      figure out what it means>
        phi_w = phi_L1 - phi_L2
        range_N = ( range_L1 + range_L2 )/2
        metric = phi_w * wavelength + range_N
      Nick indicated that this combo doesn't have iono. 
    * Use a high-pass filter to remove long term trends from data
      (more sophisticated detrending). Then look for jumps in the
      filtered data. Jumps in phase data could indicate undetected
      cycle slips. Remember to reset the filter for phase data on
      flagged cycle slips. 
    * Generate absolute code residuals after detrending/highpass
      filtering the data for the required signal frequency bands

  o Single Difference Analysis:
    * All the Single Receiver measurements can also be single
      differenced across two receivers. Don't use high-pass filtering
      if you are single differencing the data. The single difference
      process should remove any common trends between the data sets
      (that would otherwise have been removed by the highpass
      filtering), and any remaining trends are interesting
      information.

  o Plots:
    * Plot measurements in the time domain. 
    * Generate and plot a histogram of the measurements. 
    * Plot a best-fit distribution curve on top the histogram. Best
      fit normal distribution should be estimated using techniques
      described in "Robust Statistics" by Ricardo Maronna.
    * Determine goodness-of-fit using appropriate measures.

  o Double Difference Analysis:
    It was proposed that the reference SV for a Double Difference
    analysis can be the sv corresponding to the median single
    difference. Currently, the sv with the highest elevation is chosen
    as the default SV.

  o Double Difference Analysis
    * Robust Estimation of the median and dispersion of the double
      difference residuals. Use these values to generate a best-fit
      normal distribution.
    * Pearson's chi-square test or some appropriate test to establish
      a goodness of fit of the observed robustly estimated
      distribution and compare it to a theoretical Standard Normal
      Distribution. This would help establish if any resulting double
      difference distributions have any inherent biases indicating a
      receiver malfunction.


Related TODO Items: 
  o Look at GPSEdit by Joachim Tappe in Terrasat. 
  o Figure out data structures to use in NoPi and Data Checker,
    NoPi_Display.
  o Figure out how to divide tasks amongst the three tools. 
  o Figure out cause of half cycle errors on antenna switch data. Does
    turning off fast lockpoint techniques reduce the half cycle
    errors? Are the half cycle errors real i.e. are we seeing errors
    *after* lock point is known?