Interactive Voice Response

This section of our technical library presents information and documentation relating to IVR and interactive voice response software as well as automatic call answering solutions. Business phone systems and toll free answering systems (generally 800 numbers and their equivalent) are very popular for service and sales organizations, allowing customers and prospects to call your organization anywhere in the country. Our PACER and Wizard IVR systems add another dimension to our call center phone system solutions. An Interactive Voice Response (IVR) processes inbound phone calls, plays recorded messages including information extracted from databases and the internet, and potentially routes calls to either in-house service agents or transfers the caller to an outside extension.

by Mel Beckman
HowStuffWorks.com

The Truth About High Speed Modems

Modem manufacturers like to advertise apparent speed, regardless of how fast their products actually pump data. When encountering these claims, keep in mind the following fact: for dial up modems, the fastest practical speed is about 28,000 bps. The 28,000 bps limit is a function of the public switched telephone network's (PSTN) signal-to-noise (S/N) ratio. (The laws of information theory -- first proven mathematically by Claude Shannon in 1948 -- determine how much information can pass through an analog channel with a given S/N ratio.) Most of the U.S. PSTN has an S/N ratio of about 1000:1 for voice-grade lines, which (according to Shannon's law) yields a maximum data rate of about 30,000 bps. How, then, does one explain the 38,400, 57,600, and even 112,000 bps claims made by vendors? The answer is Data compression.

The ITU recommendation for 28,000 bps modems, called V.34, specifies a signaling standard designed to work reliably on most PSTN voice-grade lines. The operative word in that standard is most: There are still many local phone companies in the U.S. where V.34 won't operate or where the overall quality of a long-distance connection is too poor for V.34. In such cases, V.32 can fall back to a slower speed -- 20,000, 14,400, 9600, 4800, or even 2400 bps.

A separate recommendation, V.42, defines an error detection and correction protocol for modems that lets the modems themselves ensure reliable, error-free data transport. A modem with both V.34 and V.42 capabilities is a handy thing because it removes from the attached computers the responsibility of routine error handling. Given that error-correcting modems provide guaranteed data transport, CCITT decided that the modem was also a good place to perform data compression and released the V.42bis (from the Latin bis for second) recommendation in 1990. The data compression algorithm used in V.42bis modems has the potential for achieving as much as a four-fold decrease in data volume. In real life, though, compression depends on the data; only in rare cases does it ever reach even 50 percent, and then only with plain-text data. The more forthright vendors report their 28,800 bps modems as running at 28,800 bps. Other vendors go for the gusto, multiplying 28,800 bps times two or four to get 56,000 or 112,000 bps.

There is another reason to look upon modem compression with jaundiced eye: It turns out that the modem wasn't such a good place to put compression, after all. It seemed like a good idea back when computer users ignored security and plain text was most often the data of choice. Now, however, users are turning to host-based encryption and compression to both protect their data and get better data reduction. Host-based compression algorithms have advanced beyond the original V.42bis recommendation and generally give higher compression ratios than modems. Also, to save space in online archives, users want to store files that are already compressed.

Thus, most file transfers today are encrypted or compressed (or both) by the host and cannot be compressed any further by the modem. In fact, modem compression actually increases the amount of data when you send a previously compressed file! For most modem users, onboard compression is becoming a nuisance they want to turn off. Finally, keep in mind that (when appropriate) ISDN TAs can perform compression, too. Many ISDN-capable routers provide compression because LAN data is often compressible. If you're getting confused by all the speed and compression variables in the modem world today, remember that it all goes away with ISDN.

How ISDN Does It

You might be asking: "If ISDN can squeeze 144 Kbps out of my phone line, why can't a modem do the same thing?" The answer lies in the evolution of the telephone network.

A given pair of wires connecting two parties for communication can carry electrical signals in one of two forms: analog or digital. An analog signal changes gradually through an infinite number of values, while a digital signal changes instantly (in theory) between just two values. The human voice and a musical instrument are examples of analog signals -- both produce complex variations in frequency and amplitude. A light switch typifies a digital signal -- it can be either on or off.

An analog signal's infinite number of variations makes it impossible to reproduce exactly. An analog signal will go only so far in copper wire; to go further the signal must be regenerated electronically with a device called a repeater. The repeater converts the weak input signal to a stronger output signal, unavoidably distorting it in the process. Each regeneration degrades the signal a bit more (in the same way that photocopies of photocopies get worse at each iteration). In a large telephone network, the "copy of a copy" problem becomes very expensive to solve, requiring sophisticated equipment and costly cabling.

Digital signals, on the other hand, are easy to regenerate precisely. Because there are only two possible states for the signal, even a heavily degraded signal can be regenerated into an exact copy of the original. What's more, the cost (and complexity) of equipment to regenerate digital signals is trivial compared with that for analog. Not surprisingly, telephone companies recognized this cost advantage a long time ago and have since converted all long distance transmission to digital signaling. When a subscriber makes a long-distance call, the central office (CO) converts the analog signal to digital using a technique called sampling (see "Analog-to-Digital Conversion Diagram"), in which the state of the analog signal is captured about 8,000 times per second. Each state is converted to an 8-bit binary number, and the resulting string of binary numbers becomes a digital data stream at 64 Kbps. This digital stream is routed along the long-distance network, being regenerated as needed. Each regeneration produces an exact copy of the original digital data stream, so no information is lost. When the destination CO receives the digital data stream, it reverses the sampling process and transmits the resulting analog signal to the receiving subscriber. (On a side note, many local telephone companies are converting to digital switching, even for local calls).

A modem can't get 64 Kbps out of an analog line because the CO's signal sampling, at 8,000 times per second, limits the bandwidth of the analog signal to about 3 kHz, which in turn invokes Shannon's Law (see "The Truth About High-Speed Modems") setting the practical speed limit for such a channel to about 30 Kbps. To get higher analog speeds requires higher fidelity in the audio signal, dictating faster sampling in the CO's analog-to-digital conversion, which would result in a digital data stream faster than 64 Kbps. Telco's aren't about to change out all their voice digitizers for faster versions or upgrade all their digital circuits to carry channels faster than 64 Kbps. So the probability of the phone system ever supporting faster analog signaling rates is zero.

This is where ISDN steps in. ISDN cuts out the middleman by eliminating the need for voice digitizers in the CO. ISDN carries through the 64 Kbps digital signal from the CO right to the subscriber, and the subscriber can use it for voice or data as required. Advances in electronics make it practical to do voice digitizing right in the subscriber's phone, and direct digital attachment of computers eliminates the need for modems.

The Last Mile

ISDN has its detractors, most of whom rally behind other methods for going digital directly to the subscriber. The alternatives include existing copper-wire digital services, such as T1 (at 1.544 Mbps), Frame Relay (FR, at 56 Kbps to 1.544 Mbps) Asynchronous Transfer Mode (ATM, at 25 Mbps to 100 Mbps), and Switched Multimegabit Data Service (SMDS, at 35 Mbps). Most of these services cost several hundred to several thousand dollars per month, pricing them out of reach for many small-scale users.

Another alternative is to replace copper wire with fiber optic cabling, which would reach out and touch every subscriber with essentially unlimited bandwidth. There is a single overriding problem with all these alternatives: They suffer from an inability to conquer a physical barrier that telephone companies euphemistically call "The Last Mile."

The Last Mile, also called the local loop, is telco talk for the twisted wire pair between the CO and the subscriber. This part of the telephone network is virtually unchanged since Alexander Graham Bell. Each telephone user requires a dedicated pair of copper wires. The length is usually more than a mile but fewer than 20 miles and averages about five miles in metropolitan areas. Faster digital services (such as T1, fractional T1, ATM and SMDS) require digital repeaters at least once per mile. But normal copper pairs -- buried perhaps 50 years ago -- don't have such repeaters. What's worse, they often have analog conditioning equipment that actually impedes digital signals! If you want high-speed digital service, your telco will cheerfully run conditioned lines to your office -- for a hefty fee.

What about just replacing all the copper with fiber optics, as the Fiber-to-the-Home (FTTH) proponents suggest? The 130 million phone lines in the U.S. use 650 million miles of copper pairs. Considering that planet Earth is only about 93 million miles from the sun, this is a hefty amount of wire in anybody's book. According to a 1987 Bellcore study, the cost to replace all the existing copper with fiber would be $250 billion (and several decades of labor). This is about ten times what it would cost to replace every telephone switch in the U.S. with digital equipment and lines!

What about replacing just the business lines, or metropolitan area lines, with fiber? Indeed, this is what will probably happen. But even such limited deployment won't be cheap or fast (about $10,000 per subscriber -- and still requiring decades to complete). All the while, information technology will continue to decentralize the workplace, increasing the demand for faster communications.

ISDN can serve users until FTTH (or some other technology) is ready for prime time. ISDN isn't as fast as everyone would like, but it's a heck of a lot faster and cheaper than what we've got. And it's being delivered today.

Wizard Simplifies Development