DX Packet Cluster - analysis of spots
What is this?
Goal is to find a way to decide when to change bands in a contest based on DX cluster
activity. Determine where and when to move the antenna. Since this is based on the
DX cluster, there are a lot of underlying assumptions. (e.g. cluster activity doens't
necessarily mean a band is open - or does it?)
How does it work?
A program collects spots via a telnet session. The spots are logged in a mysql database, then parsed
every 10 minutes. The data is dumped to round robin DB using rrdtool. Graphs are made every 10 minutes
using the rrdtool. (rrdtool allows you to keep a data archive that never grows. The tools can create
graphs to display the values over time.)
What can we see?
- Spots per band - showing the distribution of data over the last 10 minutes,
30m, 1 hour, 2h and total. This works well for determining which band has the most activity.
- CQ Zone Distribution - shows the spots over time sorted by CQ zone. This locates the activity by zone.
- Azimuth graphs - create a query to show the distribution of activity by azimuth for a specific
region, time period and band. Use this to decide where to point the antenna.
- Current 2 day trend graphs - Graph spots for each band over time. Select a source region and mode
from the table. See the comment for a listing of eastern states. Use this to select a band.
- Current 2 day trend graphs - Graph spots for each band over time. Shows all regions on a single graph.
Interesting to see the differences between spots originating in various regions. Note the lag/lead times
between the Eastern region and the world.
1. Archived CQWW graphs and stats - see what happened in 2004
- Fix graph limits - I set the limit of the graphs to be identical so you can compare across graphs,
but this doesn't account for the HUGE difference between contest/non-contest periods. Look at
ssb graphs for an example
- Do a little more analysis of the graphs
- Write up a paper.
To get AC17 spot format to make it easier to parse:
Connect using "set/dx_arpc on"
Data is sampled every 10 minutes. Rates are calculated in graphs by mulitiplying sample by 6, which
gives 'spots per hour' instead of 'spots per 10 minutes'.
Program files: (mostly perl)
- getdata1.pl - connect to cluster and load database, run in foreground
- filldata.pl - run from cron, adds data to rrd database every 10 minutes
- makegraphs.pl - makes the graphs
- backfilldata.pl - create a new rrd file and fill it full of data
- index.html - main html index page
- rrdsetup.pl - set up rrdfile (don't need if backfilldata.pl is used.)
- mktable - initial mysql table create
- t1.pl - a simple time utility
- DX Line.txt - notes on the dx cluster output