Wireless quality: A ticking time bomb? Wireless networks are becoming increasingly ubiquitous-and carriers are eyeing new fixed applications for the technology. But can digital wireless support the full gamut of telephony offerings? >BY C.R. BAUGH

The desire to use spectrum more efficiently has spurred the worldwide creation and deployment of digital wireless technologies. The success of analog cellular service has saturated the available spectrum, and the new digital techniques provide more capacity, whether operating at traditional analog frequencies or in the new personal communication services range.

Unfortunately, compared with the previous generation of analog cellular technologies, digital technologies increase capacity by sacrificing some service quality. Tradeoffs include lower and marginal voice quality, extended delay in the voice path and the inability to support voiceband modem calls.

If digital wireless has the success that many marketing pundits are predicting, then 30% to 40% of the public in many countries soon will use these technologies. As the population of subscribers grows, the number of calls between two digital wireless telephones will be much more frequent. Some service quality penalties are already strikingly apparent in cellular-to-cellular calls. Over time, they may become major impediments to wireless usage and present barriers to subscriber growth.

Every user has a threshold. Subscribers will decide for themselves what's acceptable quality for outdoor mobility, wireless office phones, home cordless phones and wireless local loop service. Quality compromises could turn into time bombs for digital wireless.

Toll-quality voice The low bit-rate codecs used in digital wireless phones-13 kb/s or less-are designed to make optimum use of the spectrum, not to provide robust toll-quality voice. Although the codecs can provide reasonable voice quality for vehicular use, they do not meet the requirements for toll quality that we have in our offices and homes-and they are not meant to work in tandem with other equipment in the network, including other wireless technologies.

Voice mail systems also make compromises on voice quality. Many use low bit-rate encoding to conserve disk space for storing voice messages.

Higher quality systems use 32 kb/s adaptive differential pulse code modulation (ADPCM). Home answering machines use much lower bit rates.

Digital wireless quality deteriorates significantly when used with typical voice mail systems. A simple way to test this is to use a digital wireless phone to leave a voice mail message. Then use the same phone to call the voice mail system and retrieve the message.

With many voice mail systems, the retrieved message will be of very poor quality. If unfamiliar names are included in the message, they may be difficult to understand.

To see the compromise of digital wireless, retrieve the same voice mail again, this time using an ordinary wireline phone or moderate quality cordless phone. You will notice a significant difference in the quality of the voice message.

The quality deterioration is caused by tandem encodings of the voice signal. The original voice signal is encoded into a low-rate digital signal by the wireless system to transmit over the air. The digital wireless system converts that into a 64 kb/s pulse code modulation signal when it passes the call to the public network. The telephone network converts the digital signal to an analog signal and delivers it to the voice mail system, then converts it to a 32 kb/s ADPCM signal and stores it on disk.

When the message is retrieved, the entire sequence of encodings is reversed (Figure 1). All this transcoding causes voice quality to deteriorate.

In addition, the reduced voice quality dramatically distorts music on hold. To experience this, simply call a location that has music on hold and listen to the dramatic distortion.

If wireless service becomes as cost-effective and as popular as many people believe, then it will not be long before there are many wireless-to-wireless calls. In such calls delay becomes a problem in both directions.

Each of the digital wireless technologies-time division multiple access, code division multiple access and GSM, the pan-European wireless standard that is also being adopted in North America-has different one-way delay timing, all in the range of approximately 100 msec. It's reasonable to estimate that each digital cellular or PCS phone adds 200 msec. of round-trip delay. The end-to-end circuit consisting of two digital wireless phones incurs approximately 400 msec. of delay (Figure 2).

If both phones are digital wireless phones and the delay becomes approximately 400 msec., most people will have problems. They will interrupt each other prematurely and have a difficult time talking to each other.

This was a well-known problem of satellite calls, where the 550 to 650 msec. of round-trip delay produced many awkward conversations. When the calls were restricted to using a satellite segment in only one direction, for a delay of 275 to 325 msec., even that was objectionable to most users. In fact, time delay caused the demise of satellite as the preferred long-distance transmission medium. Today, satellites are used only when no other option exists.

Modem usage The low bit-rate codecs of digital cellular and PCS do not support dial-up modems operating at 9.6 kb/s or higher. When any of these modems are plugged into a digital wireless voice input port, the modem signal will not pass through the voice path.

The only method for transmitting data over a digital wireless system is to transmit a digital data stream. The data terminal must connect on a serial data port basis rather than on a telephone line basis (Figure 3). Laptop computers may end up with three different communications ports: a 10baseT LAN PCMCIA card for the office, a V.34+ modem PCMCIA card for use at home and on the road, and the serial port for digital wireless connectivity.

Even with a digital connection, the present digital wireless systems will support 14.4 kb/s data rates at most. Using digital cellular or PCS for dial-up data will cut the data rate by more than 50% with dial-up 33.6 kb/s V.34+ modems.

Moreover, when a wireless digital data call is made to a traditional wireline modem, the system must convert the digital signal data back into a regular modem signal. The wireless network incorporates a modem pool to realize the conversion. All this adds more complexity, round-trip delay and modem-to-modem interoperability issues.

Even if these problems could be solved, the delay problem remains. Many end-to-end data protocols use an acknowledgment to ensure that the data being sent was received correctly. If the acknowledgment takes 200 msec. or more to receive, there could be a huge reduction in throughput while the data devices wait for the acknowledgment.

If packet sizes are small, reduction in throughput will be large. If packets are large, reduction in throughput will be less but retransmission of errored packets will be much higher. In other words, there is no way out of this except to move to a windowing packet scheme, which is often not used by dial-up modems.

Data rate notwithstanding, the digital connection is a big problem for the typical fax machine. The fax machine has a telephone line connection, not a serial port, and it cannot connect directly to a digital wireless system.

Wireless faxes will mostly be used in offices and homes served by fixed wireless local loop service. They will either need to be modified to include a digital connector or have an interface box.

When digital wireless is used for wireless local loop, all the impairments discussed above come into play (Figure 4). Typically, the wireless local loop provides the traditional POTS loop start telephone line at the customer premises. When customers served by a wireless local loop use existing devices, they will discover that voice quality is worse; modems, if they work at all, will operate at substantially reduced data rates, and fax machines will be inoperable unless the wireless local loop includes an interface box for fax modems.

If wireless local loop technology goes up against landline service in a service area, there will be a significant difference in service quality. The local exchange carrier has toll-quality voice and supports 33.6 kb/s V.34+ modems and fax machines. In addition, the LEC can offer ISDN service with simultaneous 64 kb/s digital data service and 64 kb/s high-fidelity voice.

Can digital wireless systems improve themselves? Can cellular and PCS reach toll-quality voice, reduce delay and support voiceband modems? First, digital wireless standards can be changed to allocate more bandwidth for voice. By using two time slots instead of one, the data rate can be doubled and some digital wireless systems could approach a data rate that supports toll-quality voice codecs as well as fax and conventional modem calls.

The tradeoff is system capacity, which would be reduced by at least 50%. Digital wireless technologies cannot reduce delay, however, without totally redesigning the existing standards, which seems unrealistic at this point.

To support the dial-up V.34+ modems, a digital wireless system would have to support a 64 kb/s PCM codec. To do this throughout the system is, again, not a realistic possibility, even with modifications to the standards.

To support data, digital cellular and PCS providers will have to offer a digital data interface, which introduces a complex set of data interworking capabilities in the network to convert a digital wireless data and one of the standard V series modem signals. With minor modifications to digital wireless standards, the PCS network will support digital 28.8 kb/s data rates, but supporting fax will be difficult because it interfaces on the analog voice path over a telephone line.

Multiple technologies Wireless service will be offered in three distinctive environments: urban areas, including small towns, suburbs, cities and their associated highways; rural areas, including farms, ranches and rural highways; and the global environment, which includes mountainous regions and deserts as well as airplanes and ships.

Service providers want to be able to offer service in all three environments at a reasonable price. The critical issues become which technology is best for which environment and how to continue a call when a user migrates from one environment to another while talking on the phone-especially between urban and rural areas. Service quality, coverage and cost become the three determining factors, and the tradeoffs are different for each environment.

Existing analog and digital cellular technologies are well-suited for farms, ranches and rural roads because they support very large cells several miles in diameter. The foremost issue in this environment is cost-effective area coverage, which implies the largest practical cell size. Both the analog and digital cellular technologies realize large cell sizes. The choice of technology becomes primarily a choice of voice quality, which may be superior with an analog system, especially as wireless-to-wireless calls become more prevalent-or enhanced features and security-which can be achieved more easily with digital cellular.

In the global environment, coverage takes the highest priority, and service levels must be compromised for the sake of cost. The proposed low earth orbit satellite systems and existing synchronous satellites, such as American Mobile Satellite, are the only alternatives for providing service in these environments because they are the only systems that can economically cover vast areas containing very few people.

In the city environment, where toll-quality voice and capacity are as critical as coverage, the technology of choice would be one that gives subscribers toll quality on their home cordless phones, wireless PBX phones and public mobile telephones. In addition, the system must provide security for access and communication and be tightly integrated with wireline networks. The cost for this level of service goes up, of course, as the coverage area expands.

The carrier would like to use the same base stations for mobile and wireless local loop service to spread the cost of the infrastructure over both. The wireless local loop would have to support the subscriber's existing suite of telephone, modem and fax equipment.

In addition, the wireless local loop technology also would be used to supply high-speed data access, Internet access and metropolitan data communications transport. Again, the carrier would like to use the same switch for local, mobile and long-distance service to further minimize network infrastructure costs and simplify the delivery of uniform services to all phones used by a single subscriber. Finally, the technology should allow subscribers to seamlessly migrate from city to rural systems without losing a call or having to use a different portable telephone.

The PACS solution Personal access communications system (PACS) appears to be the most promising technology to deliver wireless service to suit a wide range of needs (see sidebar on page 54). It is the only current alternative that has been designed to meet all the quality targets required for all urban services. Additionally, it is designed so the subscriber unit is inexpensive. In volume its cost will be closer to that of a cordless phone than a cellular or PCS telephone, and the base stations should be in the $1000 to $4000 range.

With PACS, standard 32 kb/s ADPCM voice channels will support toll-quality speech. The less than 3 msec. one-way delay does not cause delay problems because a PACS-to-wireline network call will have an end-to-end delay of less than 5 msec.; a PACS-to-PACS call will take less than 10 msec. round trip. The delay will go undetected in a voice call and will be very difficult to detect in a data call. Since the delay is 10 msec., echo control is not mandatory.

PACS is also the only technology that has demonstrated interoperability with both wireline and mobile switches. It has run off Lucent 5ESS switches configured with both local and interexchange service, Northern Telecom DMS-100 switches and Siemens local switches. It uses existing standard subscriber carrier or national ISDN interfaces to connect the wireless network to the switch.

With conventional POTS, all the customer premises equipment operates over a loop start copper line. The PACS wireless local loop substitutes for the copper wire of the subscriber loop and provides all the same POTS services by implementing a loop-start or ground-start interface with the inside building wiring. It can even improve performance where long runs of copper lack the capability of supporting all voiceband modem speeds.

Additionally, the PACS radio system integrates seamlessly with the switch over the same trunks that the subscriber carrier systems use. Subscribers won't notice that the local loop is provisioned with wireless technology any more than they realize it is provisioned with a digital loop carrier system.

Current standards activities are considering enhancing the PACS standard to include both the existing 32 kb/s ADPCM and 64 kb/s PCM channels. By offering both these channel types within one system, wireline services can be provided on a wireless local loop with enough spectrum efficiency for mobile service on the same wireless system. The 32 kb/s ADPCM channel supports toll-quality voice, while the 64 kb/s PCM channel will support all voiceband modems, present and future.

To conserve spectrum, the wireless local loop can start out on a 64 kb/s channel and then downshift to a 32 kb/s ADPCM channel, once a call is determined to be a voice call. If it is a voiceband modem call, the channel continues as a 64 kb/s PCM to preserve the voiceband data modem signal.

Interstate highways within the urban environments require wireless technology to handle vehicular speeds. While primarily envisioned for low-speed mobility applications, recent PACS tests have demonstrated satisfactory operation at speeds up to 70 miles per hour.

PACS supports all the traditional wireless mobility services such as handoff, roaming, secure voice and authentication. It also comes with short messaging services and packet data capability. The existing standard can handle Internet protocol data calls, and subscriber units could have a special data connector such as a 10baseT interface to support Ethernet and Internet services.

The subscriber unit could also carry a 64 kb/s dial-up data call through a 9-pin connector that attaches to a computer's serial port and uses the TIA 602 Hayes-compatible AT command set for the standard modem-to-computer interface. PACS could further provide basic rate ISDN service on a wireless local loop and support leased data services-fractional DS-1, 64 kb/s, and 28.8 kb/s. There's plenty of room to expand data and voice services with incremental enhancements to the existing standard.

With PACS, toll-quality speech does not have to be sacrificed for mobility; the same mobile phone can be used for office wireless service and home cordless telephone service with no loss of voice quality.

In the face of service degradation and other limitations evolving from widespread use of digital wireless, it's reassuring to know subscribers will have alternatives. Because it does not have a digital cellular heritage, PACS stands at the forefront of the new technologies for offering wireline quality in the wireless world.

C.R. Baugh is a consultant specializing in wireless technology based in Bellevue, Wash.