[Ground-station] Operation PFB

[Michelle Thompson]

We've had some recent successes with some very difficult pieces of the
project.

There is yet another one that needs some attention! :+D

I'm talking about the polyphase filter bank.

We have enough interest, enthusiasm, and competence to get this rolling and
succeed. Like LDPC, this technology has much wider application and impact
than just our project.

This filter bank is in the payload. We call it a channelizer. The receive
bandwidth is organized into channels by this filter bank. From there, the
communications are multiplexed into the time division downlink DVB frames.

Polyphase filter bank is the exact right method to use, but it is going to
require a real team effort to get it working and demonstrated.

We have excellent MATLAB models for several different variants of polyphase
filter banks directly from fred harris. He wrote the book about multirate
processing and wants us to succeed. His contribution can be found in the
Polyphase Filter Bank repo (
https://github.com/phase4ground/polyphase-filter-bank).

We have a number of volunteers that want to help with this.

There is a polyphase filter bank in GNU Radio. It's used on our ground
station emulator flowgraphs. It's based on work by fred harris and Chris
Dick.

There was a polyphase filter bank in RFNoC, described at GRCon17.

There was an update and revision to the RFNoC work discussed at GRCon18.
The work appears to be ongoing. RFNoC ties you to USRPs, but the increase
in efficiency and speed put it into the Phase 4 Ground category.

Recently, we asked for a particular working repository from a particular
source to be released as open source, so we could use it as a codebase and
adapt it for Phase 4 Ground. That request was unfortunately denied. This is
not the only polyphase codebase that I've campaigned to get donated, but it
did feel like the most promising.

I know and appreciate how difficult it can be to dive in and start working
on a challenging and mathematically intimidating function like multirate or
polyphase.

If we want to succeed, then we need to either get a fast open source
implementation identified (or a proprietary one donated) that we can adapt,
or write one from scratch.

Here's the plan. We need eyeballs, and we need some risk takers.

1) Go find all the open source polyphase implementations that are currently
out there. Report them here: https://goo.gl/forms/BTIe81jb8746PqJ23

2) Make and share a GNU Radio flowgraph that uses the polyphase filter bank
to receive several channels. Pick 2, or 4. Pick more if you're ambitious.
Use your RTL-SDR, your Lime, your HackRF, whatever you have. How far did
you get? What problems did you have? What's not clear in how to use it?
This is a pragmatic, operator-focused experiment in using a very powerful
technique. We need to know where it falls down and how far we can push it
with today's code.

3) Write me if you want to participate but aren't currently one of the
members of our polyphase filter bank repository and I will add you.

It goes without saying but it's always better with. You do not have to be
an expert to participate, contribute, and learn. You just have to be
willing to accidentally become one while digging in.

I'm here to help make it fun and worthwhile. Want to get competent in
multirate and polyphase filter techniques? This is a wonderful opportunity
for that. It's cutting edge and highly marketable knowledge, and is also a
critical component for our radio system to work. Ask questions, pick
something, and publish it. The Polyphase Filter Bank repository is where
our work lives. Let's light it up.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openresearch.institute/pipermail/ground-station-openresearch.institute/attachments/20181120/3de6da29/attachment.html>