Intermittent System Outage returns.

Matt_Is_My_Name,

I wonder if you are familiar with the ethernet concept that handles packet collision. This little discussion is not related to satellite communications per se, just the transmissions between multiple stations connected by a single ethernet cable which daisy chains from one station to the next, to the next, etc. It is a typical setup that might be used for the network in an company with a lot of offices, for example. Assume there are multiple stations on an ethernet circuit and two of them decide to send a data packet at essentially the same moment. There is collision detection circuitry used to detect that multiple stations are sending at the same time. It then puts a long pulse onto the ethernet cable that blocks both messages from getting through. The transmitting stations each recognize this and respond by each one transmitting its packet again after a random delay period. The assumption is that the two stations will choose different delay periods so their retransmissions do not collide again.

I suspect that the HughesNet satellite communications system works in a similar manner with regard to the many ground stations sending packets to the satellite. As long as each ground station's transmitted packet does not overlap that of another ground station, all goes well. If multiple ground stations try to transmit at the same moment, the satellite senses the packet collision and transmits a signal telling those stations their packets failed to get through and to retransmit them after some suitable delays.

I run an application program much of the time that keeps sending PING messages and it keeps track of how many of them succeed and how many fail to make the round trip to and from a selected server on the internet. Under typical conditions when weather is not an issue and we are not in the midst of one of these outages that we have been fighting, a usual success rate for the PING messages hovers around 50%. It does vary in a rather random way, a little higher or a little lower for a minute or two at a time but generally hangs near that 50% mark.

Wednesday evening, we had a lot of heavy rain cells passing through the area, replete with several tornado warnings. Not surprisingly, we had spells where no PING messages were succeeding at all. My app flagged those periods as outages but it was clear that such outages were caused by the local weather.

However, I noticed something else that happened on at least one occasion when there was no rain in my location. There was a period when the success rate for the PING messages rose considerably above the usual rate, well above the 75% success rate, and stayed there for a while. Later, it returned to its usual behavior.

I kept wondering what could have caused that. Now I can propose a scenario about what caused this that is related to my theory about the satellite handling those packet collisions that I mentioned above.

I suspect some of the heavy rain that was around the state but not in my own area was blocking transmissions from many of the other HughesNet ground stations. That gave my PING messages a much better chance of getting through.

Of course, with many people having their attention drawn to the weather, perhaps they were not so likely to be using the internet in the first place. Either way, it is a theory.

Matt_Is_My_Name, What do you think? Can you propose another explanation?