Look Before You Leap

The 3G/IMT-2000 initiative is an attempt by standards committees to define the future of cellular/PCS radio interfaces. It is doomed to fail, like many other premature standardization efforts. Ironically, international standardization is needed for network protocols, not for radio interfaces.

Timing Is Everything

Computer programming languages have been around much longer than cellular systems, and standardization (whether formal or de facto) has been a mixed blessing. Many languages, such as academia's Algol-68, IBM's PL/1 and the U.S. military's ADA, have come with imposing specifications developed by a committee or by a large company, but all eventually fell by the wayside. (If you have never heard of these programming languages, then my point is proved.) The most important programming language of the 1990s is C (and its derivatives C++ and Java), developed by a couple of smart guys at Bell Labs.

The specification was published in a thin book (of which the actual specification occupied only a few pages). However, the language was not standardized until it first had become a significant programming language in academia and then in business and engineering. Standardization of this language in the mid-1980s was important, both to resolve deficiencies and ambiguities in the language specification in a single way and to standardize the library of functions that supports it. Although it was important that standardization occurred, it also was important that it did not occur too early.

Radio interfaces still are evolving too rapidly to be standardized. The question of whether CDMA is better than TDMA cannot yet be answered -- nor even the more fundamental question of whether digital is always superior to analog. Will it be Beta vs. VHS (victory of the inferior), C vs. Pascal (victory of the superior) or AM vs. FM (peaceful co-existence)?

Standards committees have to operate by consensus, and there can be no consensus until all technologies have had a chance to be proved in the marketplace. Only then can you understand the fundamental limitations of the technology, as well as the consumer needs that require a next-generation system. The worst outcome would be if the ITU were able to cajole and bully companies into marching single file, resulting in a compromise system.

Economic Cents

The ITU's major error is its emphasis on high-speed data. Wireless data always has been tomorrow's technology. This is probably due to the emphasis on high-speed, real-time consumer applications at the expense of more mundane, low-speed, non-real-time commercial applications. The ITU has taken this tendency to a utopian extreme. High-speed wireless data certainly is technically feasible, but it is not economically feasible due to the disparity between the bandwidth required for voice (8kb/s to 13kb/s) and that required for high-speed data (64kb/s to 2Mb/s).

The ITU also has made a point of defining a long list of exotic features that most subscribers never will use. These features largely are used by one camp (GSM vs. D-AMPS vs. CDMA) to attack the capabilities of the others. Those with the longest list of under-used features win, according to some. This is as productive as building a wall around your house because you cannot get along with your neighbors.

The industry needs to place more emphasis on integrating the world's two mobile application network protocols: GSM MAP and TIA/EIA-41 (formerly known as IS-41). Phones could operate on different frequency bands seamlessly using a variety of radio interfaces if better integration between these two protocols were available. Currently, GSM and TIA/EIA-41 differ in many ways, including feature control, short-message delivery, mobile handoff and authentication. Moving from one backbone network to another means that you lose most services beyond the ability to make and receive calls. It is possible that integration of these two protocols never will be feasible, but a second-generation network protocol that incorporates the best of GSM and TIA/EIA-41 and that provides backward compatibility would be an achievable goal. (Perhaps the protocol could be called Inter-MAP, in honor of the success of the Internet in joining disparate LAN protocols through a common intermediary.) Existing air interfaces could be adapted to this new network protocol with modest modifications.

Using authentication as an example, the TIA already is developing a second generation of authentication algorithms. Why not use this as an opportunity to develop algorithms that would be applicable to use in a GSM smart card as well?

Weeding Out Weakness

Feature control is another example of misplaced emphasis. GSM provides special air interface messages to control each feature ("call forward unconditional"), making it slow to adopt new features. TIA/EIA-41 uses * feature code strings (*73 often disables call forwarding). This is more adaptable but less efficient in the use of radio resources because you have to allocate a voice channel. Inter-MAP could allow both methods to co-exist, with the phone using the GSM method when it is available for a particular feature and the TIA/EIA-41 method at other times such as when a feature requires use of a voice channel. For example, "cancel call waiting" operates only for a single call.

Another weakness of wireless telecommunications systems is their reliance on SS7 message transport services. Although SS7 is robust and works well for national networks, it is awkward to use for international networking. Furthermore, there is little integration between the wireless SS7 users (GSM MAP and IS-41) and wireline (ISUP). Seamless international wireless/wireline signaling could allow least-cost routing of calls from a wireline phone in one country to an international roamer across the street in the same country. Currently, that call requires two international long-distance legs. However, the canyon between wireless and wireline appears to be even wider and deeper than between cdma2000 and W-CDMA proponents.

If you consider that a decade is the time required for a major new technology to mature, then network standards are ready for a new standardization initiative. (IS-41 was first published in 1987.) The development of digital radio interface standards is much more recent, and they need several more years of breeding before the weakest can be culled from the herd. The ITU is earnest and well meaning, but just because next year ends with three zeroes does not mean that it is an especially auspicious year for standardization. Sure, a new acronym will be required (IMT-2004 just will not do), but that is a small price to pay.