On The Horizon
January 2002

Brough Turner Mobile Telephony -- Making High-Quality Voice A Reality

BY BROUGH TURNER


Mobile telephony is the telecom success story of the past decade and its spectacular growth has several decades yet to run. This success, however, has been based on “mobility,” not on “telephony.” Subscribers have been willing to pay extra for the ability to communicate on the fly, despite inconsistent service and poor voice quality. But in the next five years, this will improve. In fact, you will be able to get mobile phone service that sounds better than a landline phone call.

The significant difference between mobile and land-based telecom services is the vibrant competitive environment in mobile telephony. While residential landline customers have, in reality, only one viable provider of local telephone service, wireless customers in U.S. markets have four to six carriers to choose from -- Verizon, AT&T Wireless, Sprint, Cingular, Nextel, VoiceStream, etc. Likewise, all major European markets have multiple wireless service providers. Indeed, even third world countries with PTT monopolies have introduced wireless telecom competition. The result of this competition has been responsive service, interesting new services, more innovative packaging of standard services -- such as bundling of minutes and prepaid accounts -- and dramatic subscriber growth.

IMPROVING COVERAGE
Of course, subscriber growth has stretched carrier facilities to the limit, causing service problems in certain areas. Just getting enough capacity to handle the growing subscriber base has been a problem. Radio spectrum is limited and expensive and, until recently, the FCC restricted how much radio spectrum any one carrier could purchase. But access to new spectrum and ever improving radio technology is helping. And competitive pressure drives a constant effort to add capacity -- both additional cell sites and more efficient technology -- so as to guarantee service availability.

In some European markets, mobile service has penetrated 80 percent of the population and subscriber growth has begun to slow, giving carriers a chance to catch up with traffic growth. But as coverage and availability become a given, wireless carriers are looking for new competitive frontiers. Among the choices are diverse new services -- e-mail, Internet access, voice-activated dialing, voice portals, etc. – and improved voice quality.

IMPROVING VOICE QUALITY
The first goal in improving mobile voice quality is to at least match the quality of landline calls. Each successive generation of cellular technology has improved voice quality. The initial wireless offering, AMPS (Advanced Mobile Phone Service), was an analog system subject to increasing noise the further you are from the base station. Second-generation (2G) systems, such as GSM, CDMA (IS-95), and TDMA (IS-136), are digital. These systems use aggressive digital speech coding to limit the amount of information being sent over the air which, combined with improved modulation, increases the number of voice channels per MHz of available spectrum. These 2G systems eliminate earlier analog noise problems, but their voice coders (vocoders) limit voice quality. For example, the initial GSM vocoder uses only 13 Kbps (versus 64 Kbps for a landline phone call). The analog noise is gone, but voice quality is still much less than that of a landline telephone call.

As digital signal processors (DSPs) have improved, new vocoder algorithms have emerged. The GSM community developed both a half-rate vocoder that can pack twice as many calls into the same amount of spectrum with little additional degradation, and an enhanced full-rate (EFR) vocoder that uses additional DSP resources to deliver better voice quality over the original 13 Kbps channel. Mobile service providers who initially offered the better-sounding EFR service quickly gained a competitive advantage. As a result, EFR has become standard in competitive GSM markets and carriers are now aware that voice quality is a competitive issue.

IMPROVING THE USER EXPERIENCE
But whether it’s GSM, CDMA, or TDMA, none of the 2G vocoders match landline quality. So while we wait for next-generation vocoders to squeeze more quality into the available bandwidth, there are a number of other interesting technologies available today to enhance the perceived voice quality during a mobile call.

The first is improved echo cancellation. Because digital mobile telephony systems involve speech coding, they introduce significant latency into a call. Unlike landline systems, in which a local call can be made with no echo canceller, 2G wireless systems require echo cancellers on every call to a local landline subscriber. But these echo cancellers differ in the amount of cancellation they provide, the rate at which they adapt to changing conditions and the way in which they interact with other echo cancelling devices used by long distance and VoIP service providers.

Less well-known but more significant is a series of wireless-specific voice quality enhancements, which can be performed on the same platform as the echo cancellation. The first of these enhancements is background noise reduction. The amplitude and spectral content of the background noise is monitored, especially during pauses in the conversation, and that information is used to reduce the noise heard at the other end. Although it would be possible to totally suppress background noise this is not done -- a totally quiet line would make it seem that the mobile subscriber had hung up.

At the same time, by knowing the level of background noise a caller is experiencing, it’s possible to raise or lower the amplitude of the other party’s speech to compensate. The more sophisticated systems only provide gain for actual speech (not for noise) and dynamically adjust the target level to match the changing background noise level, allowing the subscriber to experience a comfortable listening level under all conditions.

Another voice quality issue is acoustic echo. Wireless standards require manufacturers to minimize the amount of coupling from the earpiece to the microphone in a mobile telephone. Unfortunately, many wireless phones do not meet standards, so acoustic echo contributes to speech quality problems. What’s more, traditional echo cancellers can make this problem worse. Traditional echo cancellers are designed to compensate for electrical echoes that are mostly stable for the duration of a call. Acoustic echoes are not stable, however, as sound bounces off of multiple nearby surfaces, and the movement of people or objects during a phone conversation creates different, rapidly changing, acoustic echo effects. In addition, the echo canceller sees this acoustic echo through the non-linearity of a speech compression algorithm. While many echo canceller equipment vendors claim to do acoustic echo control, the effectiveness of acoustic echo solutions varies widely from vendor to vendor.

A typical sequence of voice enhancement features includes Noise Reduction (NR), Acoustic Echo Control (AEC), Automatic Gain Control (AGC), and Intelligent Level Control (ILC) which integrates AGC with Noise Compensation (NC) (i.e. ILC = NC+AGC). The combination of these seemingly minor enhancements allows wireless carriers to dramatically improve the perceived quality of a mobile call, opening up a new area of competition. Already some wireless carriers are branding their sound quality and using it as a marketing tool to attract and retain subscribers. And as mobile subscriber densities increase, voice quality becomes an issue for mobile-to-mobile calls as well as mobile-to-landline calls. This means the voice quality enhancement technology must move to the radio side of the mobile switch to provide voice quality enhancements on every call.

REDUCING DELAY
Another factor degrading the quality of mobile calls is delay. Delay is introduced by digital modulation schemes and each time speech is coded. In 2G networks, the coded speech from the handset is converted to 64 Kbps G.711 so it can be switched with traditional switching equipment, even if the call is destined for another mobile subscriber. Beyond that, many service providers send speech through an inter-exchange network where, in order to save money, the speech is recompressed in a different format. Every time the speech is re-coded, additional delay is introduced, plus each sequential coding reduces quality.

To solve this, the industry has developed specifications for Tandem-Free Operation (TFO), at least for mobile-to-mobile calls. With TFO, intermediate transcoding is turned off on mobile-to-mobile calls between compatible handsets (GSM-to-GSM or CDMA-to-CDMA). The result is greatly reduced delay and less voice quality degradation. TFO-capable equipment is becoming available and beginning to be deployed. However, TFO only provides an advantage on mobile-to-mobile calls. And, with today’s equipment, you lose the voice quality enhancements (NR, AEC, AGC and ILC) when TFO is turned on as the voice quality equipment works on 64 kbps signals and, under TFO, speech is never converted back to G.711 -- it’s kept in the vocoded domain. Luckily, there’s technology in the lab that will allow the voice quality enhancements to be applied to vocoded speech, on a frame-by-frame basis. By the time TFO is widely needed, this technology should be commercially available.

GOING BEYOND LANDLINE QUALITY
Everything discussed so far makes mobile calls sound better, but still not as good as a landline call. And the voice quality of a landline call is not all that great -- it’s based on 120 year old technology. For example, the frequency response of a landline telephone call is limited to just over 3 kHz, even though the human voice has frequencies in it that go up easily to 7 kHz. That why it’s so difficult for people to understand you when you are spelling words that contain “s” or “f.”

The breakthrough will come with 3G mobile technology. The 3GPP (Third Generation Partnership Project), has specified vocoders for use in the 3G world. While they have defined an adaptive vocoder that is a superset of existing coders (AMR-NB: Adaptive Multi-Rate Codec for narrowband speech), they have also defined a new vocoder called AMR-WB (Adaptive Multi-Rate Codec for wideband speech). Here the speech coder is able to encode frequencies up to 7 kHz with the result that an AMR-WB telephone call actually sounds better than a landline telephone. While 7 kHz is still not enough for music, it’s enough for speech and will make a wireless voice call sound as though you were speaking face-to-face.

Actually getting AMR-WB service will have to wait for widespread deployment of 3G wireless and the availability of new wideband handsets. This will take a few years, but it will happen. The technology works in the lab and wireless service providers recognize the importance of voice quality. So in an extremely competitive market place, we can expect to be offered mobile service that sounds better than traditional telephony in the not too distant future.

Brough Turner is senior vice president of technology at NMS Communications, a leading provider of hardware and software technologies for developers of high-value telecommunications solutions. For more information, call NMS Communications at (508) 271-1000. E-mail to the author (addressed to brough_turner@nmss.com) is also welcome.

[ Return To The January 2002 Table Of Contents ]