> > > Message: 2 > Date: Mon, 29 Dec 2014 15:31:38 -0800 > From: Chuck Hast <[email protected]> > Subject: Re: [PLUG] O.T.VoIP and Satellite > To: "Portland Linux/Unix Group" <[email protected]> > Message-ID: > <CADNfBV-E-CAKZTq3tkYm5hxi4xCS7ZQ0a0635kt1j3KnD= > [email protected]> > Content-Type: text/plain; charset=UTF-8 > > MIke > Thank you for the observations. I did test the connection, since I am using > cellular, I found several phone numbers to test against, and all of them > provide > good inbound audio but my outbound audio is just all corrupted. >
Call quality or lack thereof, as it relates to the network, is mostly a function of packet loss, delay and jitter. Jitter is variance in delay. 150 ms 1-way delay is the standard measurement for "toll quality voice." That is, the voice quality is good enough to charge for. > > I do not think that the DHCP assignment caused the problem, but I am trying > to figure out if something else was changed on the network at the same > time. > i.e. different routing. > > I know that normally ip addresses are not geo based, but it was always of > note that in the past any search or application that took me to a map would > always take me to a map of the area I live in, now since I am one > HughesNet, I see > that I am now taken to sites that are no where near where I am, and I > figured that it was probably where the gateway to the uplink to the > satellites was > located. I know that they have several of them, so I thought that might be > the issue. > You're correct in that a new DHCP ip address assignment could change the gateway and the routing to and from the satellite. It's also possible that only the outbound route is problematic. You would want to run at least some extended basic ping tests to the default gateway. What would be really useful at this point is to get some relevant network connectivity data. Can you go here - http://voiptest.8x8.com/ and run a few tests and post the results? I would run the test for 69 secs and run it for both G.729 and G.711 codecs. The reason being is that your internet connection might support the lower quality G.729 codec and you might be able to set that in your microcell or in your smartphone voip app. When the test is complete please click on the advanced tab and copy and paste all the statistics reported. Also, if you go to the "summary tab" and click on "result analysis" of "voip test" that would be useful info too. > > > Typical internet is asymmetric - when somebody is watching a movie > on netflix or surfing the web, they are receiving a firehose of > bits and sending out a trickle of ACK packets. > > VOIP usage is symmetric, moderate bandwidth data streams in both > directions. > > Satellites are also asymmetric - they have a limited number of > transponders with limited bandwidth, which they will allocate to > maximize overall customer retention, which means catering to the > majority. Which isn't thee and me. > > The satellite provider probably recently reallocated a customer > uplink transponder as a customer downlink transponder, to better > serve the netflix users. There might be an FCC or ITU document > or ruling about this. Do you know which particular satellite > you are talking to? One of the ANIKs? > > This is really getting off into the weeds. What matters with VOIP call quality is consistency. Consistency of packet loss, delay and jitter. Jitter is variance in delay. 3 Mbs of bandwidth in each direction should be sufficient to provide okay call quality. Call distortion is caused mostly by packet loss, delay and jitter. At the simplest level, 150 ms one way delay is the standard measurement to provide what's called "toll quality voice." That is, it's good enough to charge for. A g.729 call requires 32 kbps. The average satellite link bandwidth is approx 400 kbps. If you're just making a voip call, there shouldn't be any call quality problems due to "asymmetric, moderate bandwidth streams in both directions." However, asymmetric routing in which the outbound and inbound calls take different routes with different packet loss, delay and jitter is a real problem. _______________________________________________ PLUG mailing list [email protected] http://lists.pdxlinux.org/mailman/listinfo/plug
