Have Phone, Will Travel?

International roaming solutions, at best, are half-baked, but carriers aren't inspired yet to make them better.

Housewives in a Dutch commuter town once noticed that a train rattled through the station toward the city at 10 a.m. and back again at 3 p.m. without stopping. This was the perfect time to get some shopping done, have lunch with a friend and get back in time for the children's return from school. They wrote to the railway company asking for the train to stop. A month later the railway company wrote back that it had instituted a trial, and the demand simply was not there. It had stopped the train in the station one day (in both directions) and no customers were waiting.

International roaming is something that nobody can object to, but that currently few consumers know is possible. Consequently, the demand does not appear to be very high. Plus, most can avoid the need if the price is too high or the inconvenience too great. To North American carriers, the domestic market is so large and demanding that international roaming rarely makes their radar screens. To the Europeans, it was the raison d'etre for GSM, which is better adapted to this purpose. But, for all technologies, international roaming has problems that remain unresolved. Solutions may be the victim of inertia, as it is cheaper to bumble along with half-baked solutions than to do it right. Customers need to be told that the train is at the platform, even if it isn't the best train that the company has. Once customers are used to a basic service, they soon will demand more.

SS7 Issues

One of the biggest problems facing international roaming is the use of the SS7 protocol. Even though the same name is in use around the world, the versions used in most countries are similar, but incompatible, with one another. Although there are railway tracks in most countries, they aren't always the same gauge. The basic address unit in SS7 is the point code, and it is assigned independently in each country. (Canada and the United States are the exceptions, sharing a common set of point codes.) Even if point codes were assigned globally, they differ in size (14, 16 or 24 bits), often making SS7 protocols incompatible at the very lowest, most fundamental, level.

SS7 does have another address format, known as global title, that does allow international communications, but more in theory than in practice. The global title is simply a telephony number phone number, mobile identification or calling-card number that can be used for routing. When a wireless phone registers, its identity (MIN: 10-digit mobile identification number or IMSI: 15-digit international mobile subscriber identity), for example, will be analyzed and mapped onto the point code of an international SS7 gateway. This complicated device converts the message into the format of the destination country, and eventually the global title is converted into a point code of the HLR in that country.

The problem with this is that global title information must be programmed independently in every STP (SS7 router). Furthermore, new global titles often are needed to support new services, multiplying the management headaches. GSM systems rely on a single global title based on phone numbers and a kludgy mapping from IMSI to phone numbers known as ITU-T E.214 that, unfortunately, does not work well in North America. Because of this, only a small portion of the IMSI resource can be used to maintain global title routing.

Wireless carriers using North American standards (analog, N-AMPS, TDMA and CDMA) have barely implemented global title at all. One of the hurdles they face is that even if this type of routing were implemented, compatible global titles would have to be implemented in all the countries they need to interconnect with, and these countries are even further behind than North America.

International roaming does work despite these problems, but only because clever ways have been found to use point codes for routing. The simplest way is to extend the ANSI (North American) SS7 network into countries, running in parallel with the native SS7 network, and the other is to use native SS7 protocols combined with simple point-code mapping from those in one country to another (resulting in multiple point codes being assigned to single systems). However, both solutions treat the ANSI point-code resource as global, something that it was not intended to be. If unused ANSI point codes become a rarity, illegitimate users of the resource will be the first to feel pressure to find another solution.

Mobile Identification

Another problem with international roaming is the need to identify every subscriber uniquely wherever he can roam. GSM has a simple solution with IMSI, a 15-digit number that identifies the home country in the first three digits, allowing each country to assign the rest of the resource any way it sees fit. In North America, GSM has the problem that only 1/10th of the resource can be used because of phone limitations (2-digit mobile network code instead of 3) and 1/24th because of SS7 limitations, resulting in only 1/240th of the available IMSI resource being available for GSM systems in North America.

Analog, TDMA and CDMA systems still rely on the venerable MIN, a 10-digit number that unfortunately does not identify the home country, not directly at least. A transition to IMSI has begun but has not reached the commercial implementation stage due to compatibility problems. Instead, the MIN has been converted to an international resource. The International Forum for ANSI-41 Standards Technology (IFAST) (www.ifast.org) assigns MIN codes beginning with 0 or 1 as international roaming MINs (IRMs). These codes cannot be used by North American systems that require the MIN and phone number to be the same.

The IRM allows the HLR to be identified by analysis of the first four digits of the MIN and ensures that every analog, TDMA and CDMA mobile can be identified anywhere in the world. Once the HLR is identified, it can be queried using the ANSI-41 protocol to obtain the subscriber's profile, validate his right to service and authenticate him.

Call Delivery

One more problem is the temporary local directory number (TLDN) used by North American systems for routing calls to roamers through the landline network. Although ANSI-41 allows internationally formatted TLDNs, most systems use 10-digit national numbers, causing difficulties with call delivery to roamers and sometimes misrouting of calls. The technical solution is to use international numbers, but, as with almost everything associated with international roaming, it stumbles because it is not compatible with older systems that only support national TLDNs. In the interim, complex translations can be used, but not generically. The risk is that, eventually, irresolvable ambiguities may occur, and current solutions are, at best, error-prone and a maintenance headache.

Will the Future Be Better?

It is lucky that wireless carriers still have a more entrepreneurial attitude than their landline cousins. The solutions to international roaming are far from pretty, but they make the train stop in the station every day. As international roaming revenues grow, carriers may turn their attention to making international roaming work better. More realistically, however (and more cynically), perhaps nothing will change until an elastic band or piece of duct-tape in the current system breaks, causing a major failure in international roaming and focusing embarrassing attention on its shortcomings.

---

Crowe (crowed@cnp-wireless.com) is a wireless-standards consultant and editor of Cellular Networking Perspectives, a wireless-standards and -technology bulletin.