 |
|
|
The “A” interface don’t get no respect. Everyone can appreciate the features that the radio interface provides. In fact, many consumers believe that this is the only interface in a wireless system. Those who are more tech-savvy have figured out that there must be a network to provide roaming, especially people who have fallen behind on payments and realized that, no matter where they turn, they get the same answer when they try to make a call – “No Service” (until you pay your bill, at least). But, an interface between a Base Station and a Switch. How and why could that possibly be exciting?
The concept of an open interface to connect base stations to wireless switches (MSCs) was first established by GSM. The desire to open this interface was not academic – the base station and MSC are some of the most expensive components of a wireless system. A proprietary interface forces carriers to buy both from the same vendor. An open interface allows carriers a choice. Even if they do not exercise it, their suppliers know that it would not be impossible to replace them. And, every time a network expansion is planned, the carriers can use the competition for their business to keep prices down.
GSM has a highly developed series of specifications for the “A” interface. Not only does the series define the interface between a GSM Base Station Subsystem (BSS) and the switch (MSC) in TS 48.001 through 48.008, but also between the BSS and the GPRS SGSN (router for GPRS packets) in TS 48.018. A newer interface between the BSS and the BTS (Base Transceiver Subsytem; the radio equipment) is defined in TS 48.051 through 48.058.
AMPS (analog) systems in North America did not put a premium on the “A” interface. An attempt to standardize the “A” interface for analog was launched in the late 1980’s but eventually abandoned. Nor did the carriers who started TDMA service in the early 1990’s consider this important. CDMA carriers, particularly Sprint, were different. An open “A” interface became a priority for them (and still is). Asian CDMA carriers are also active participants in standards development, including KDDI in Japan and SK Telecom and KT Freetel in Korea.
The first CDMA “A” interface standard (IS-634, published in 1995 by TIA TR-45.4) was closely modeled on GSM, being based on a circuit-switched voice with SS7 signaling. It tried to be technology agnostic but, with only CDMA carriers showing an interest, became more and more focused on CDMA. The interface assumed that base stations would be connected to MSCs over T1s, as is common even with proprietary interfaces. Each of the timeslots within the T1s (DS0s) would carry either voice or SS7 with IS-634 as the application layer carrying information regarding important functions such as call setup and teardown, handoff and registration.
The “A” interface is heavily affected by changes in radio interfaces, and IS-634 has had to be revised in parallel with several of the revisions to the CDMA standards (IS-95, TIA/EIA-95 and IS-2000). The identity of the standard changed to IS-2001 in June, 2000, and revisions have so far paralleled the IS-2000 CDMA standard that it supports.
As more changes introduced more options it became clear that the goal of interoperability was fading away. Sprint and the CDG then developed their own specification, known as IOS (Inter-Operability Standard). They then applied some not-so-subtle pressure on vendors to make sure that the standard became more focussed.
Data has become more of an issue for the “A” interface as it has developed. The goal has been to route data to the internet or other data networks at the base station, rather than through the voice/circuit-oriented MSC. 1XEV-DO is a new high-speed data-only interface which is supported by a November 2001 standard published as TIA-878. GSM, similarly, has had to adapt their “A” interface to the internet-oriented GPRS protocol.
Many people believe that all communications, including voice, from the base station will eventually become packet-oriented. All interfaces will then be IP-based, running a variety of other IETF protocols at higher layers. Voice-over-IP matches wireless quite well because voice is already digitized and compressed and, with variable rate voice coders, is well suited to transmission over a packet interface.
Transmitting coded voice is a very attractive idea even before the All-IP concept is a practical reality. Radio interfaces generally run voice coders at 8-13 kbps (versus 32-64 kbps in landline networks), and speeds may even go lower in the future. The fewer bits used by voice coders, the more subscribers who can be supported by the same amount of spectrum. There is a fly in the ointment, however - ensuring that there is a compatible voice coder at the destination. Most radio interfaces use different voice coders from each other and each support multiple voice coders of their own. There is no guarantee that two mobiles will use the same voice coder, even if they are using the same radio interface (and, it is also true that the same voice coder could be in use by mobiles using two different radio interfaces).
Using 32-64 kbps voice not only uses more bandwidth, but paradoxically reduces voice quality because it implies four encoding or decoding stages. In fact, if speech is recorded by a voice mail system, even more coding steps are applied. Each coding or decoding loses information, and reduces the quality of the transmission. Tandem Free Operation (as it is known) requires only encoding by the transmitter and decoding by the receiver and results in better voice quality. The ideal solution would be to have end-to-end signaling to negotiate a voice coder. However, that is impractical when calls are transmitted through the PSTN (as most are).
A surprising solution has been found (in TIA/EIA-895, published in September 2001) to support tandem free operations that only requires the participation of base stations. Through the technique known as ‘bit robbing,’ the digital DS0 facilities can transmit both the coded voice and uncoded voice. The bits that are ‘robbed’ are the least significant bits of the 56 kbps channel, and make little difference to the voice quality. By including the voice both coded and encoded it is possible for the terminating base station to decide which stream to use. Moreover, it is possible to switch between coded and decoded voice, which is sometimes necessary when a mobile participates in a feature such as 3-way calling that may disrupt the flow of the robbed bits.
An “A” interface is going to be in everybody’s wireless future. Perhaps it is only a coincidence that the wireless protocols with the most promise for the future (cdma2000 and the GSM family, including UMTS) are the only ones that have opened up this important interface. Or perhaps, as with roaming, what was once seen as this year’s luxury will become next year’s necessity.
© – Copyright