Growing Pains

By Richard "Zippy" Grigonis          

IMS is unusual in that it was designed in a top-down manner — a huge, architectonic, perfectly logical schematic diagram.

Old circuit-switched networks such as the PSTN, evolved in a “bottom-up” manner, built mostly through trial-and-error on existing technology. In the 19th century, for example, the “pair” of wires connecting two telephones was actually just one wire (and that wire was iron or steel instead of copper, as was the custom with the earlier telegraph technology). The oldest telephone circuits could get away with one wire because they were “earth return” circuits, which consisted of a single transmission wire from the phone company to each phone, with a metal rod stuck into the ground each end — the electrical circuit was thus “completed” through the earth between the rods. This saved on wire, and the telegraph industry had used this design for decades. However, recall that telegraphy is really a robust “digital” medium and is more immune to electromagnetic noise than a subtle analog system such as the telephone. If the soil was too dry or too wet the circuit would lose the earthed ground. Moreover, phones could pick up strange noises (“telluric” eddy currents in the ground, movement of the earth’s magnetic field by the solar wind, etc.). On July 19, 1881 Alexander Graham Bell received a patent for using a second wire to achieve a stable ground by providing an insulated return path (back to the phone company) for the signal current, an arrangement now called a voice circuit. Over a 20-year period the one-wire “grounded” system was replaced with the noise-eliminating, two-wire “metallic” system.

Just as 19th century technicians attempted to build the fledgling telephone system with the base materials and components of the earlier telegraph system, so too did 20th century experts attempt to build early IP networks as extensions and adaptations to the older telephony network and circuit switching legacy. A special signaling protocol such as BICC (Bearer Independent Call Control) was just a method for transporting SS7 information over a packet switched network. Softswitching and gateway control protocols such as MGCP and Megaco/H.248 also simply “decomposed” a circuit switch into various interacting though independent components.

IMS, however, is based totally on packet-switched IETF standards and Internet-friendly signaling and call control protocols such as SIP ( Session Initiation Protocol (News - Alert) ). Routing, switching, directories and general “intelligence” itself are distributed throughout the network instead of being totally hierarchical and centralized.

Its a priori design philosophy and reliance on IETF standards leads us to the irony that IMS is a mature concept with as yet immature or nonexistent applications and components. Since IMS has a theoretically perfect, top-down design, not a whole lot “at the bottom” that’s physically real is equally perfect and will work perfectly with it! Real network element stuff needs to be tweaked and tested for interoperability at various “bake-off” interoperability events. Moreover, every vendor tends to sneak in a “special sauce” extension of some sort, so thorough interoperability testing among various vendors’ wares has the potential to succumb to combinatorial explosions of test-revise-test cycles — though nothing that serious has happened yet.

Some vendors appear more “prepared” for IMS than others. Sonus Networks (News - Alert) (www.sonusnet.com) for example, claims that their switching equipment has always worked in accordance with IMS principles, so they’ll simply have to “divide the functionality currently supported by one Sonus product into two separate products to better adapt to the IMS architecture”, and then merely do some software upgrades. Sonus customers will be able to upgrade to full IMS compliance by updating the software while it’s still running, processing data and calls.

Moving to an all-IMS network will doubtless be a more thrilling experience for some vendors and network operators than others. We’ll see what happens.

Richard Grigonis is Executive Editor of TMC’s IP Communications Group.