Ways for telco to signal answer/hangup? [telecom]

Hello

I'm trying to figure out how Asterisk and an FXO module can detect that the remote end has answered and then hung up.

Google and a check in the configuration of the Linksys 3102 POTS gateway seem to say that there are four ways for a telco to signal a hangup:

Detect CPC (Calling Party Control): = open loop disconnect?

formatting link

Detect Polarity Reversal:

Detect Disconnect Tone:

Detect PSTN Long Silence:

Is this correct? As for signaling an answer, I guess two methods are available:

Loop disconnect Polarity Reversal

In addition, is there a tone-based way to tell that the remote end has answered, eg. listen for the ringback tone to stop?

Thank you.

***** Moderator's Note *****

Giles,

This is one of the oldest problems in telephone signalling. In no particular order, here are some of the questions involved:

Did the called party hang up, or did they just fumble or drop the phone, or did customers at two called-party extensions think the other person had the call?

Is the called party's line or instrument defective or has it been tampered with? "Ring trip silence" faults are common after rainstorms with copper cables, or when some kid finds an old copy of Ramparts magazine and tries to reinvent the black box.
Is the answering machine out of tape? Are its batteries failing?

Well, you get the idea: the first problem is determining _IF_ the called party hanged up. We must be sure that they did before we signal the calling party. Bill Horne Moderator

authoritative answer: "It depends."

On what the far end of _your_ circuit does when the switched circuit has been torn down.

If you are connected to a PBX, 'anything goes' -- you'll have to talk to the PBX admin to find out (a) what the PBX is capable of doing, and (b) how it is programmed to react for _your_ line..

For telco C.O. gear, the situation is _somewhat_ more predictable:

With a -basic- 'loopstart' circuit, you pretty much have to wait for the C.O. to (a) re-assert dial-tone, (b) give you a 'busy' or 'fast busy', or (c) give you a 'crybaby'.

With 'kewlstart' (an enhanced/extended loopstart) the C.O. will -- it is virtually certain, at least in the U.S. -- drop battery to you, to signal the 'disconnect'.

Pretty much anything you do in this area is simply a _guess_. You can have situations where the far end answers _before_ you get any ring indication. And you can have situations where ringback is interrupted/ delayed, =without= the called party answering.

You can -try- looking for continuous 'tone' -- e.g. a modem 'answer' signal; or 'voice' audio (will false-positive on far end 'intercept' message, unless you have SIT checks in place, and they were not triggered); or (maybe) an increase in 'background noise'.

It's _possible_ that you get a battery polarity reversal on connect, but unlikely (except maybe on a pay-phone circuit). Early "touch-tone" pads =were= polarity sensitive, and customers bitched like h*ll about not being able to navigate far end IVR systems, or enter their answering machine password. Newer pads are -- almost universally -- not polarity sensitive for exactly that reason, _and_ C.O.s do not reverse polarity so as to not provoke user complaints.