Friday, June 14, 2013

The Origins of VOIP

The other day someone asked me how VOIP (voice over internet protocol) worked and it got me thinking about the origins of digital telephony. I have a mild (hah) interest in telephony, loving every automatic switching system Ma Bell ever made. Too many days have been spent scouring the internet for descriptions of SxS, panel, crossbar and ESS switches. One of the things that has also piqued my interests was the various trunk lines that the Bell System used to carry long distance calls, including the very beautiful WECo microwave "funnels." Developments in both of these areas, first trunks then switches, will combine to form the basis for the digital subscriber line that many of us use in our homes and offices.

Our story starts perhaps a bit earlier than you might expect it to, back in the 1960s. In the Wikipedia article for VOIP the earliest recorded event in VOIP history is the development of a "Network Voice Protocol (NVP) developed by Danny Cohen and others to carry real time voice over Arpanet" in 1973. But we had the inklings of digital telephony as early as 1961 when Bell Labs introduced the T-Carrier, commonly known as a T-1 line. T-1 was a trunk multiplexer. It could carry 24 phone calls -digitally- over a pair of copper cables.

Before the introduction of T-1 the only way to carry multiple conversations over a single cable was to treat them almost like radio stations using an L carrier. Each conversation would be filtered and then modulated with a certain carrier to have it fit into a certain bandwidth. For example, you could take four signals, low-pass filter them to 4KHz and modulate them with frequencies that put them in different bands of the transmission line such as 20K, 25K, 30K and 35K (allowing for a little guard band). Then you could apply a similar process and get 16 lines down one pipe in full analog. But this was the 50s and these things were expensive, so pulse code modulation was considered as an alternative. In addition, the call quality improved, so they were all the more better for it.

So now we have all these digital trunk lines going everywhere, but the odd thing is that every time one of them passed through a class 4 switch (those that handle long distance linking) the trunk line was decoded back to analog, switched and then perhaps re-encoded to be sent off to a local office. This odd little practice caused a controversial change to be made in the way in which long distance calls were routed. In 1970 Bell began to develop the #4ESS, a completely digital class 4 Electronic Switching System. T-1 lines would come in, their data would we parsed and then routed through T-1 lines going out. Calls could now be completed long distance as if you were still calling within your own district (at least quality-wise). The first #4ESS was delivered in 1976 in Naperville, Illinois.

Now that we've gotten into the seventies, we need to take a step back to the sixties to start looking at the development of the internet. The idea first started as a government project, in the 1950s as the SAGE air defense network and developed into commercial and private computers being able to use standard phone lines and leased lines to send data to each other. Originally it was just a number of "dumb terminals" dialing in as users on a shared system or universities sharing programs by sending them serially (perhaps simply even tape-to-tape over a phone line). But towards the end of the decade, plagued by bad connections and the desire to have a constant network of shared information, a protocol was developed of sending packets of data so that if any one of them was somehow corrupted, the receiving computer could ask for that one little bit to be re-sent. Soon the military would get involved again and the rest is the history of the ARPANET.

Here's where we catch up with the developments made in 1973, because those individuals took the first development, pulse code modulation, and paired it with the new network that was starting to parallel the phone system. However, it appears that these were simply "voice" over the internet. As much as you could send someone an email or share a program or data, you could communicate with someone over this new internet protocol. However, you could only do that with someone who had access and the proper equipment and was online. You could not really call into the standard phone system. This was essentially a single phone channel out of the T-1 line using the internet as 'wires.' This is where the #4ESS comes back into play.

At the time the #4ESS was developed, the reason the developers decided to do switching completely digitally wasn't entirely because they saw the demodulation-remodulation as silly, but also because they saw the writing on the wall and realized that digital switching was the future. When you encoded things digitally and then sent them down a wire you could do so many things so much better than a standard analog switch. When higher level of T-carriers came out, the data that was coming in on slower T-1 lines from local offices could then be routed to faster T-2 or T-3 lines that could carry many more calls in between long distance offices. It was no longer a one-to-one relationship in transmission speed, which is kind of similar to the packet system, where bits could be sent anywhere in little chunks, or frames, to use T-carrier language.

As much as it was a similarity to packets, it was kind of a downside of T-1 that it was just a 24-channel transmission line. You could only send data down it in a specific way, in frames of bits that designated channels. But, theoretically once you got rid of the multiplexer and used it to interconnect class4 offices, it was just a 1.544Mbit/s data channel. You could send anything down it. Eventually this is what happened and some of these connections were used to form the backbone of the internet.

As a bit of an aside, I feel many people think that the internet was carried on phone lines originally because they used to use modems to dial into it. But I don't think that many people realize that the internet, up until somewhat recently, was completely separate. When you 'dialed in' to the internet, you were dialing into a serial data connection on a computer (similar to what we did in the 60s with terminals) that was connected to the internet- through which you could access data on it. This is part of why the next step almost perhaps seemed seamless when it came out in public use.

It wasn't until the mid 90s that someone realized that they were all using the same data cables to send phone data as computer data and that they already had the technology to send voice through packets and then created what are called "soft switches." Soft switches are essentially class4 and class5 (local office equipment) switches that can utilize the packet system of the internet to carry their voice data. So now for the first time you had a switched telephone network being carried over the internet and ironically carrying calls of people connecting to the internet. But you wouldn't really know if your phone calls were being carried over the internet at this point anyway, so the roundabout-ness of this was not always noticed.

In the present day, with the network of the internet being expanded as it is, where our computers connect to the internet like the computers we used to dial into, the switched aspect of the telephone service can take a completely different shape. Now class4 and 5 switches can be fully implemented as servers run by the phone companies existing as completely integrated bits of the internet. A simple box in your residence can do all of the pulse code modulation of the old T-1 line and send the data as packets out to be processed much like the #4ESS would. Perhaps the transition to this type of system came about in an attempt to not duplicate facilities as the internet was growing. What is clear though is that the idea that we have of encoding everything as data and sending it all down one big pipe to multiple destinations started out as a means of providing a more efficient pipe and a better switch. A phone system is a network. One that carries 'data' that can be encoded. And as demand for internet access grew and more people got better and faster access, enough room for a single 64Kbit/s voice channel was easy enough to provide.

-Always improving. That's the Bell System for ya!-

P.S.- There's one little offshoot in this story. In the mid 80s such a thing as digital telephones became popular in offices. The pulse code modulation was done in the phone and, rather than using an analog line to the switch, were connected via a serial RS-232 line. All switching was still done by a purpose-built private branch exchange, as it had for years prior. However, this is the direct ancestor of phones in offices today that connect via ethernet LAN to access the phone system. These appear to be still and only popular in offices. I have an early multiplexer for one of these systems that I'll be dusting off soon for the cameras - stay tuned!

No comments:

Post a Comment