Re: ANI vs. Caller ID [Telecom]

_usage-based_billing_

You are saying 70-80% of landlines have local usage based billing? Where did you come up with that figure?

What is the percentage of subscribers still using rotary service? My guess it's about 10% for residential lines, if that much, and 1% for business lines. In terms of percentage of submitted calls, it's even less. Even if there were hardware devices as you describe, the traffic of rotary calls must be extremely low.

AFAIK, there is no such thing _today_ as "digit decoding gear". As mentoined before, all lines must be scanned often for on-off hook status. Dial pulsing is collected as part of that scanning. No extra hardware is required. The translation from dial pulses or Touch Tone signals to whatever the switch uses internally is done by software.

If there is a citation describing that modern switches have a hardware device for digit decoding, could you share it?

Software is not "tied up" on a computer when it is waiting for input. The CPU goes on to do something else. (Interrupt handling is beyond the scope of the discussion).

When you visit a website, the server is not "tied up" while it waits for you to think or type in stuff. It only acts when you're finished.

You confirmed my statement. "Parity of service" was so that newcomer phonecos would not be at a disadvantage in attracting subscribers. Likewise, so was number portability, which also adds unnecssary load and expense to swtiching--which we all had to pay for, just to benefit newcomers.

Uh, no. It's so that subscribers with numbers in overlay area codes wouldn't be at a disadvantge. Verizon hands out 646, 347, and 917 numbers, you know. Some of the pressure for area code splits was due to inefficient allocation to new entrants, but the number of actual numbers in use is increasing, too.

Good point. One black bakelite phone per house with service at regulated rates is all anyone really needs.

R's, John

I repeat, "FALSE TO FACT".

First, there is _NO_ repetitive 'scanning of all lines' for on-/off-hook status. That approach is TOO *DAMN* EXPENSIVE (in terms of resource consumption) to be practical.

I'm not going to go into all the gory details, but the basic outline of switch architecture is that it has some functions that *must* be done at specific time intervals. These are 'real-time' tasks. Each of these tasks has the exclusive, and *un-interruptible* use of the CPU until it finishes its processing. Obviously, these routines are written to use as little CPU as possible, and return control to the scheduler. "Bigger" real-time tasks are broken down into "whatever" number of smaller pieces as are needed to get the execution time of each individual piece under the size of the real-time scheduling 'slot'.

The 'real-time' subsystem has ABSOLUTE PRIORITY in call on resources. _Nothing_ can interrupt the real-time tasks -- because if anything was allowed to do so, one could not predict how long a given 'piece' (per above) would take to process, and one could not guarantee that it would finish within it's allocated time-slot (an absolute requirement, in order for the -other- real-time tasks to get _their_ slots in a timely manner). Note: There are an *ABSOLUTELY*FIXED* number of real-time time-slots in the system. There is no even theoretically possible way to get 10,001 time-slots of 100usec each in a system where each slot must run one time per second.

When the real-time subsystem has 'spare' time -- whether it is due to the presence of 'unused' scheduling slots, or of 'pieces' that completed their increment of processing before the end of their time-slot -- then the 'interruptible' subsystem runs.

The interruptible system consists of tasks with non-deterministic behavior, that do -not- have hard 'real-time' constraints on them. These tasks run under an _adaptive_ priority-based scheduler, but _only_ when the real-time system does not need the CPU. There are two basic variants of tasks in this system -- routine 'scheduled' tasks (initiated by a scheduling software directive), whose execution can be interrupted 'at will' for by any other task, for any reason; and 'interrupt service' tasks, which take precedence over any 'scheduled' task, and also take precedence over any lower-priority interrupt-service task. Interrupt-service priorities are hard-coded into the system architecture, while the priority of any 'scheduled' task can vary (within limits) from moment-to-moment.

Software-based digit-decoding -- of either DTMF or pulse digits -- is, of necessity, *REAL*TIME* task, because you must have a stable time-base for the samples for pulse-width or tone-frequency determination. This makes such decoders 'expensive to use', and '_relatively_ scarce', in the system architecture. Therefore such decoders are activated only for the _minimum_ necessary time periods.

To use on-/off-hook 'scanning' for pulse dial detection, the on-/off-hook scanning would have to run as a real-time task for *every* line on the switch. EVERY SECOND of every day. To reliably decode 20PPS dialing, one would have to sample each line a minimum of 40 times/sec. (Nyquist limit), and preferably more. Or a minimum of 3.45 _million_ times per line per 24 hour day.

What you do, _instead_, is hang a simple differentiator circuit (hardware) on each tail circuit, that generates an interrupt "when and only when" the line state (on-/off-hook) changes. This circuit is expressly designed to be _insensitive_ to changes as short as a dial-pulse. This circuit triggers twice per phone call (once going off-hook, and the 2nd time going on-hook). If one is making a call every 8 seconds, 24 hours/day, this interrupt happens less than 2,200 times/day -- more likely it will be only in the 10s to 100s of times per day. Worst-case: 1500 times less frequent than 'scanning', probably 15,000+ less load on the system.

WHEN, and *only* when, the line goes off-hook (10s to 100s of times a day) do you enable a REAL-TIME digit decoder task on that line. And it stays enabled _only_ for the duration of the actual dialing -- a few seconds for DTMF, maybe 20-25 seconds for pulse. Say it is 25 seconds times 100 calls, thats a total of 2500 seconds of real-time task per day for a _HEAVILY_USED_ line, vs 86,400 seconds for _every_ line under your scanning scenario. A switch-wide average on outgoing calls is probably more like 10-20 per line than 100. Making the 'decoder on demand' approach on the order of 100-200 times more efficient than full scanning'.

It is advisable to read one or two lines ahead, it can keep you from looking foolish.

See the above discussion of "real-time" subsystems, and why software-based digit-decoding must 'tie up' a real-time scheduling slot.

Irrelevant and immaterial to "real-time" processing.

WRONG. "Newcomer phonecos" _were_ part of the pressure that necessitated area-code splits and overlays, but *ONLY* part of it. The burgeoning use of fax machines and especially computers with modems put much _more_ pressure on the numbering system. In the 80s, I was working for a business with a grand total of 6 employees. We were datacomm intensive, had nearly 30 phone lines (12 for voice, 16 modems (some dial-in, a bunch for dial-out FAX) and 2 dedicated incoming fax line) , plus a T-1 internet connection, plus 8 dedicated data circuits. This kind of loading was not uncommon for 'technology using' small businesses in those days. The telecom model had changed RADICALLY -- from one trunk line per several employees to several lines per employee. THIS is what put the overall pressure on the numbering system. Full prefix allocations to CLECs aggravated the situation, but it wasn't the entire cause, nor even the primary one.

"Parity of service" was so that _existing_ USERS of the phone system,

*whether*or*not* they switched to a 'newcomer phoneco' would not be disadvantaged in retaining existing customers. _Every_time_ a "split" happened, _3/4_ of the pre-split universe was affected. (everybody in the new areacode, plus the circa 1/2 the calls from the old areacode to somebody that was now in the new one.) Two splits, and the numbs of 'inconvenienced' customers is greater than the total number of people in the area -- some have been inconvenienced more than once. OTOH. With an 'overlay', and mandatory 10digit dialing for all numbers, _everybody_ is inconvenienced (and to a lesser degree than 1/2 the people in a 'split'!) once, and once _only_. Additional overlays can be brought in, if/when needed, with no inconvenience to anyone.

Yes, it's correct that subscribers in overlay area codes wouldn't be at a disadvantage.

BUT the reason we need those new area codes, IMHO, was _largely_ due to new entrants, not actual growth. For example, my town didn't need to grow from three to twenty-five exchanges in one year because 'growth". The new addition to the town's C.O. building was not for growth, but to put in termination equipment for the the newcomer carriers.

Some years ago it was proposed to put faxes and wireless devices in the new area codes where a 10 digit number is not a disadvantage, but apparently this wasn't done very much.

I have no idea of how your response relates to number portability.

Let's be clear that number portability represented a significant cost-- which is a line item right on my bill. It benefits the newcomers to make it easier for customers to switch to them. As such, IMHO the newcomers ought to be the ones to pay for it, not the rest of us who aren't changing. If the newcomers were truly offering a superior service or better price, they could well afford the cost of number portability. However, the books I referred to in an earlier discussion said the newcomers offered nothing better.

As to "one black phone", that sounds like a strawman. The old Bell System offered a large variety of services and equipment well beyond "one black phone" for decades. Of course, the difference to today is that all of us must subsidize those with premium services --such as the portability fee and mandated 10 digit dialing--whereas in the past those with premium services paid their own way.

Like I said, there was a burst of poor allocation giving an entire NXX to each entrant. That's over. Now the allocations are quite efficient, and when they add overlays, it's due to real growth.

That was an experiment in area code 917. The experiment failed, since

917 filled up, too.

Perhaps.

Which is a gift from the regulators, unrelated to the actual cost.

Aw, come on. If the ILEC wasn't required to provide portability, the costs of switching would be impossible, since it would include changing one's phone number.

Putting nonsense in a book doesn't make it true. Early on the CLECs prevented a disaster by providing the inbound modem lines for BBS and dialup ISPs that the Bells were too myopic to do. (Their response was to lobby for the "modem tax" and hope they went away.) These days, compare the ILEC offerings to those from cablecos and wireless, and you can see why the number of ILEC lines is shrinking every year.

Oh, man, I don't know where to start. Back in the old regulated days, everything was priced by its "value" rather than its cost, which is why rural service was priced lower than urban due to fewer people in the local calling area, even though it cost and still costs far more to provide. Residential POTS was priced below cost, long distance was overpriced to subsidize it, and the maze of cross subsidies was impenetrable. Fred Goldstein knows this history way better than I do.

R's, John

Your experience apparently extends only to residential service.

Believe me, business service _is_ different.

And guess who places the vast majority of outgoing calls.

Could you name some other published books with a different point of view?

_All_ businesses try to charge for _value_ rather than cost, if they can get more money for an item. For instance, new medicationss under patent are priced very high until a generic or alternative comes out that undercuts them. They say generic brand aspirin is the same as national brand aspirin but national brand costs more. (AFAIK, aspirin is very cheap to make.)

With phones, they say text messaging is cheaper to handle than voice traffic, yet most cellphones give a better deal to voice users than text users, charging for texting unless one upgrades to an expensive plan. Thus, they're charging for _perceived value_. If teenagers ever get bored with texting, watch the price drop.

The business services in many cities and suburbs that I know of, pay only one message unit, untimed, for local calls. In a city like Philadephia, that would include the whole city. Obviously there may be other cities that time even local calls.

In any event, I doubt many businesses today have rotary phones, and if they do, they're not used very often. (eg an old store open irregularly, or an old dial phone in a back room.)

Not offhand. On the other hand, I don't see why one would prefer claims in some unnamed book to the observable fact that all over the counrtry the ILEC's share of phone lines is shrinking.

Hmmn. Two days ago you said that in the old days phone users paid for the cost of the services they use. Now you're agreeing with me that they didn't.

If we review our Economics 101, in a competitive market, marginal price meets marginal cost, with the difference between the price and the "value" kept by buyers as consumer surplus. (This is one of the good things about markets.) Sellers can only keep the price way above cost if there are market distortions, e.g., price regulation or monopoly power. Deregulation has had the predictable effect of increasing the prices of stuff that used to be deliberately underpriced, e.g. residential POTS, and decreasing the prices of stuff that used to be deliberately overpriced, e.g. long distance.

Right. This leads us to the not very surprising conclusion that US mobile carriers don't compete very hard on price. Here in the UK they do compete on price, there are all sorts of really cheap text bundles, and the overall cost of mobile service is (to my surprise) less than in the US with the exception of the monopoly-priced calls from landline to mobile.

R's, John