Skip to content

CHANGELOG.md

Latest commit

 

History

History
184 lines (142 loc) · 6.31 KB

CHANGELOG.md

File metadata and controls

184 lines (142 loc) · 6.31 KB
Raw

0.8 -> 0.9

Ugh...forgot to update one of the version numbers

0.7 -> 0.8

Bug fixes in the solver and add a Makefile so I don't have to keep looking up the commands.

0.6 -> 0.7

Windows support was added. It actually took shockingly little effort. Tests pass, LaTeX and image generation works. I also added expanded human_b support as well as an actual unit-test for human_b.

0.5 -> 0.6

The HyperSpectral Imaging code was integrated here. I know that it doesn't really belong, and I should probably separate everything out into 3 packages (DAG, link budgets, and hyperspectral)...but until someone cares or there's a reason other than principle I'll wait.

0.4 -> 0.5

The package was published on PyPi here. The main change was that it was published under the title pylink-satcom since there was a name collision in the PyPi index.

0.3 -> 0.4

There were a number of minor changes, but mostly it was updated to work properly with Python3 and Anaconda. I've switched to using Python 3.7 in Anaconda as my primary dev environment.

0.2 -> 0.3

  1. LinkBudget is now a Tributary, just like channel, or geometry

  2. The logic to combine all of the tributaries now resides inside of DAGModel

  3. The report generation logic was separated out into the Report class

  4. Modulations have been implemented for realz this time

  5. Some of the nodes have been added/removed/renamed/...

Points 1 & 2

Points 1 and 2 are pretty easy to deal with. In older versions, you'd build a LinkBudget object, which would both combine other tributaries, add in your custom functions, and add its own nodes as well:

budget = pylink.LinkBudget([pylink.Geometry(),
                            pylink.Antenna(is_rx=True),
                            pylink.Interconnect(is_rx=True),
                            pylink.Receiver(),
                            pylink.Transmitter(),
                            pylink.Interconnect(is_rx=False),
                            pylink.Antenna(is_rx=False),
                            pylink.Channel(),
                            pylink.Modulation(name='QPSK', perf=perf)],
                           name="My Really Cool Budget",
                           rx_antenna_noise_temperature_k=350,
                           **extra_stuff)

In the newer rev, it's a bit cleaner, and more true to the DAG nature of the system:

model = pylink.DAGModel([pylink.Geometry(),
                         pylink.Antenna(is_rx=True),
                         pylink.Interconnect(is_rx=True),
                         pylink.Receiver(),
                         pylink.Transmitter(),
                         pylink.Interconnect(is_rx=False),
                         pylink.Antenna(is_rx=False),
                         pylink.Channel(),
                         pylink.Modulation(name='HomeGrown', perf=perf),
                         pylink.LinkBudget(name="My Really Cool Budget")],
                         **extra_stuff)

As you can see, the LinkBudget just becomes a regular tributary which makes it easier to review, instantiate, update, etc etc etc. There's also a hint, there of point 4, but we'll get to that.

Point 3

Point 3 is also pretty simple. Previous versions included the export logic within LinkBudget, so you'd do things like this:

my_awesome_budget.to_latex('awesome.tex',
                           author='The Incredible Hulk',
                           intro='And now, for a budget that needs no intro',
                           watermark_text="Ruh Roh")

Now that the logic has been separated out, you would do this instead:

report = pylink.Report(my_awesome_budget)
report.to_latex('awesome.tex',
                author='The Incredible Hulk',
                intro='And now, for a budget that needs no intro',
                watermark_text="Ruh Roh")

Point 4

Point 4 is, perhaps, the most complicated. It introduces a cycle into the model (as you can see above in Cycles), and also necessitates a bit of rethinking.

Previously, you'd do something like this:

modulation = pylink.Modulation(name='MSK (Magical Shift Keying)',
                               required_ebn0_db=5,
                               spectral_efficiency_bps_per_hz=1)

If, however, you have multiple code-options (cough DVB-S2X cough) and operate an adaptive bitrate loop, that approach won't work. You also don't differentiate between your transmit spectral efficiency, and your receive spectral efficiency. If, for example, you have terrible filtering and need 20% of your allocation for rolloff, then those two numbers will be different. The rx (rx efficiency) value affects the noise bandwidth so that you can properly make computations such as required_demod_ebn0_db and excess_noise_bandwidth_loss_db. The tx value is useful for things like your channel capacity.

To address these shortcomings, the Code object was introduced. This allows you to specify a series of these Code objects, which will be automagically-selected based upon your allocation size, and C/N0:

perf = [
    #           Name, tx, rx, Es/N0
    pylink.Code("BPSK", .5, .5, 4),
    pylink.Code("QPSK", 1, 1, 8),
    pylink.Code("8PSK", 2, 2, 13),
    ]
modelling_clay = pylink.DAGModel([pylink.Geometry(),
                                  pylink.Antenna(is_rx=True),
                                  pylink.Interconnect(is_rx=True),
                                  pylink.Receiver(),
                                  pylink.Transmitter(),
                                  pylink.Interconnect(is_rx=False),
                                  pylink.Antenna(is_rx=False),
                                  pylink.Channel(),
                                  pylink.Modulation(name='QPSK', perf=perf),
                                  pylink.LinkBudget()])
modulation = pylink.Modulation()

The Modulation tributary defaults to DVB-S2X, with 80% efficiency of the transmit spectral efficiency.

Point 5

Point 5 should be pretty obvious as you go. For example, spectral_efficiency_bps_per_hz doesn't exist anymore since we consider both the tx and rx sides, so now you have tx_spectral_efficiency_bps_per_hz and rx_spectral_efficiency_bps_per_hz.