| Attachment | Size |
|---|---|
 thermal noise.jpg | 76.47 KB |
| return.jpg | 76.53 KB |
Late last week I started monitoring aggregate modem stats on my networks. Every 20 minutes I poll the plant and pull in modem reported rx and tx power levels. From these I calc and report the mean, median, the quartiles and the first few moments. I've been doing this sort of thing on plant turn up as it shows fairly well how balanced the plant is.
So I come in this morning and just check in and see this interesting trend.
This is a plot for a 4 day period. The red is the entire range of modem RX powers (at the modem). The Yellow represents the inner mean quartile (divide the population into two halves at the mean value, then again athe mean of the two halves), the blue is the mean and the black is the median.
There is a very distinctive wave pattern to the data which appears to be diurnal in nature. This pattern appears in all of my sites and appears to be more prominent the further you go from the mean (one site has one or two modems around -15, and the pattern is more pronounced compared to the inner quartile).
It is interesting that the return path shows no such behavior at all. Does this mainly affect the higher frequencies?
I'm not concerned about it being a problem, but it is always interesting to see periodic patterns appear in data, so I thought I'd share.
I started Cable tv back in 82' when there was no fiber. I would be interested to see what the return looks like over a period of time
~Carl
I attached the same style plot for the return path for roughly the same time period. The powers are the modem reported transmit powers.
This plant is 100% coax with 50 modems and 4 or 5 line amps in the field. It would be interesting ot break this down by segment and see which amp is contributing most of the resistance.