Wide-Area Wireless Data

The landscape is further complicated by other technologies trying to leapfrog over cellular networks, many of them based on OFDM (Orthogonal Frequency Division Multiplexing), a radio technology also employed by IEEE 802.11a and IEEE 802.11g. Examples include Flarion's FLASH-OFDM and the high-visibility WiMAX, standardized as IEEE 802.16. Some Wi-Fi enthusiasts even dream of blanketing entire metro areas with 802.11 access points. These upstart ideas don't have support from major operators yet, but they could be a potent disruptive force (see "WiMAX and Beyond,").

We recently conducted a reader poll to gauge enterprise adoption of wireless WAN services. After tallying results from 638 respondents, the majority in organizations with revenue between $10 million and $1 billion, one finding stood out: Most still aren't familiar with these data services, especially compared with wireless LAN technologies, such as IEEE 802.11. The sheer number of offerings may be one reason. Plus not many use these data services yet.

As they move forward with 3G deployments, however, operators hope that wireless WANs will gain traction, and there are reasons to be optimistic. The higher speeds enable a wider range of applications. Meanwhile, nearly all enterprise software vendors, including IBM, Microsoft, Oracle and SAP, are adding features to make their applications work better over wireless connections. These features include caching information locally, compression, minimizing back and forth traffic, and sending only information that has changed. Add 3G capabilities, and enterprises might find the combination irresistible.

For years we've speculated about the possible "killer application" for wireless. Many thought it would be wireless e-mail; others said phone microbrowsers. We think the real enabler is broadband capability. Only broadband provides a highly responsive, always-connected networking experience that lets us be productive (and entertained) across a wide range of applications while mobile. Moreover, broadband lets companies use existing applications with no modifications--the need for wireless middleware to improve performance has always stunted uptake. Only with broadband can enterprises truly start to deploy their full range of applications, including unrestricted e-mail, Web and client-server apps, and vertical market applications, such as field automation, transportation and logistics. With wireless broadband, the killer app is, in essence, any application.

Mass-market broadband came into its own with consumer and small-business adoption of DSL and cable-modem services. It continued with rapid adoption of Wi-Fi, which widened broadband access in some areas. The next logical step is to extend broadband into the wireless wide area. Operators are in the earliest stages of a process that will unfold over the rest of the decade.Do current 3G technologies deliver a true broadband experience? Sort of. The FCC defines broadband as 200 Kbps of throughput, and using this benchmark, all the emerging wireless WAN technologies pass, though just barely compared with the 1-Mbps to 3-Mbps speeds users are experiencing with wireline technologies. In our poll, 75 percent of respondents said a performance threshold of 1 Mbps must be reached for wide area wireless to be as broadly appealing as a general-purpose enterprise data service.

Equally important is latency--the time it takes packets to traverse the network. New cellular technologies still have relatively high latency, in the range of 200 to 300 milliseconds. This is better than previous wireless services, though still high compared with Wi-Fi and wireline networks, where latency is usually under 30 ms. For perspective, Internet latency typically ranges from 20 ms to 100 ms, depending on the number of hops. For downloading e-mail, file transfers and streaming audio/video, latency in the 200-ms range is not an issue. Even VoIP (voice over IP) becomes feasible at 200 ms maximum latency. But for applications that must transfer a lot of individual objects, such as database transactions, that latency feels sluggish compared with other broadband connections. Still, these new networks are faster than any previous wireless WAN technology, and more than capable of supporting a wide range of apps. And latency will continue to decline, with future 3G technology versions expected to reach about 100 ms.

Technologies and Services

To expand your company's wireless WAN usage you must sort out the technologies, the networks built from them and the services offered.Many operator-provided services use underlying data capabilities. These include SMS; MMS, which supports picture messaging; hosted e-mail; microbrowsers on phones that can access a broad range of mobile-formatted content; downloadable ring tones and games; streaming video; and general-purpose IP networking. Demand for these services comes from both consumers and businesses, so there's a robust market for a wide range of devices, including phones, smartphones, PDAs and modem cards.

From an enterprise and IP networking perspective, all the technologies offer similar capabilities. During the connection, the network authenticates the mobile device (phone, PDA or modem) against credentials stored on the device, such as in its SIM (Subscriber Information Model) card. This authentication is independent of any enterprise identity system. The network then assigns the mobile device a dynamic or a static IP address. Operators usually assign private addresses, but most make public addresses an option.

The mobile device uses this same address even as it moves across wide geographic areas. Because the system is packet-based, the mobile device can maintain sessions with servers, but it will consume radio resources only when it needs to send or receive packets. This selective use of the radio interface is invisible to applications, and they can act as if they have constant connections. Billing is flat rate or based on volume of data transmitted. To connect to enterprises on the fixed-end side (as opposed to the mobile side), mobile systems can communicate over the Internet, possibly using a VPN for security, or over private circuits, like frame relay PVCs (permanent virtual circuits), that some operators offer between their core networks and customer networks. The 2.5G technologies like GPRS and EDGE offered data encryption with 64-bit keys and one-way authentication; 3G technologies will offer robust security architectures using 128-bit keys and two-way authentication.

As we mentioned, the two main cellular technology families deployed in the United States and around the world are CDMA2000 and GSM/UMTS. CDMA2000 is influenced strongly by Qualcomm, which holds a lot of its intellectual property rights, and is based on CDMA's direct-sequence spread-spectrum technology. Qualcomm's patents relate to the application of CDMA to cellular networks. The version deployed throughout the Americas, Asia and parts of Eastern Europe is called CDMA2000 1xRTT. Carriers using 1xRTT peg typical throughput rates at 50 Kbps to 70 Kbps, and using ping, we measured latency at 400 ms to 450 ms between mobile systems and Internet sites (see "Wireless on the Road,").

The version of CDMA2000 being deployed for broadband data is CDMA2000 1xEV-DO. Here, the carrier assigns a CDMA radio channel (1.25 MHz wide) for data. In contrast, WLAN radio channels are 20 MHz wide.EV-DO evangelist Verizon Wireless says its service, available in Las Vegas, San Diego and Washington, offers average throughputs of 300 Kbps to 500 Kbps and peak rates of 2.4 Mbps. Although they sound good, these "peak rates" are meaningless, because they can be achieved only under optimal radio conditions. So unless you're right next to a base station and the only user in the coverage area, forget about it.

Verizon says it plans to offer EV-DO in most major metropolitan areas next year. Sprint was holding out for a more sophisticated version of CDMA2000, called 1xEV-DV (DV stands for data/voice), that enables the same radio channel to have both voice and data channels. But because of competitive pressures, Sprint now plans to deploy 1xEV-DO to top U.S. metro markets in 2005.

Allocating a radio channel purely for data has the advantage of optimizing the channel for high performance, but it does limit high-speed data capacity to that one radio channel. As demand grows, the operator can add EV-DO channels, if it has the spectrum, or fall back to 1xRTT networks that support both voice and lower-speed data. Eventually, as demand for data grows, operators could migrate their networks to support voice over IP over EV-DO, but not all the pieces are available yet to do this. For example, quality-of-service mechanisms such as traffic prioritization within the wireless network are lacking. Alternatively, 1xEV-DV could mature sufficiently so that operators could upgrade their networks to this technology.

EV-DO achieves high throughput rates using a number of mechanisms. In the higher-speed downlink, it employs a single high-speed channel allocated to different users in the time domain, a more efficient approach than allocating channels in the code domain, as 1xRTT does. EV-DO uses an efficient scheduler, which emphasizes transmissions to users with the best instantaneous radio conditions. It also employs higher-order modulation, allowing more bits to be communicated in each radio symbol (shift of amplitude, phase or frequency to the carrier signal). Finally, it uses dynamic coding, which employs different levels of forward-error correction based on channel conditions. These mechanisms also boost spectral efficiency, resulting in high aggregate data throughput per cell site for a given amount of spectrum.

Whereas the CDMA2000 family has about 100 million subscribers globally, GSM/UMTS has more than 1 billion. The best-established data service is GPRS, available from operators worldwide. GPRS provides average throughput rates of 30 Kbps to 40 Kbps. EDGE, a version of GPRS with an enhanced radio interface, boosts rates to average throughput rates of 100 Kbps to 130 Kbps, assuming a PC Card modem, which has higher throughput than phone devices. AT&T Wireless has deployed EDGE throughout its coverage area, and Cingular Wireless expects to complete its EDGE deployment this year. T-Mobile also plans to deploy EDGE. For 2.5G services, competition is essentially EDGE from GSM carriers versus 1xRTT from CDMA operators.Still, GSM, GPRS and EDGE are all based on a TDMA (time division multiple access) approach in the radio link. To move fully into the 3G realm, GSM carriers need UMTS. UMTS is a CDMA technology like CDMA2000. Transmissions in CDMA are differentiated by their spreading codes, as opposed to time-slot assignments. UMTS, however, differs from CDMA2000 in practically every implementation aspect. For example, UMTS, also referred to as WCDMA, uses 5-MHz radio channels, where CDMA2000 uses 1.25-MHz channels. These wider radio channels enable higher peak data throughputs but, of course, also require operators to have more spectrum to deploy the technology. In Asia and Europe, governments have auctioned 3G spectrum. But in the United States, operators are scrambling to squeeze 3G services into 2G spectrum while the FCC prepares 3G auctions for next year.

Some obstacles, including just how the proposed 1.7-GHz and 2.1-GHz bands will be cleared, are not resolved, and much of the FCC's attention is on other wireless services, such as unlicensed spectrum for Wi-Fi and WiMAX. Meantime, operators are amassing spectrum through acquisitions--for example, Verizon Wireless is buying Qwest's spectrum assets and Cingular Wireless has merged with AT&T Wireless.

Unlike 1xEV-DO, which allocates the entire radio channel to data, UMTS uses the same radio channel for voice and data, allowing for simultaneous operation. For example, you could be on a voice call and have a simultaneous Bluetooth connection between your notebook computer and your phone, using the phone as a modem to surf the Web or read e-mail. Designs of UMTS devices limit peak data rates to 384 Kbps, though rates as high as 2 Mbps are supported by the network. AT&T Wireless claims average rates of 220 Kbps to 320 Kbps for its UMTS services in Seattle, the San Francisco Bay area, Detroit and Phoenix. San Diego and Dallas deployments are planned for later this year.

To improve on the WCDMA radio interface for data, standards groups have specified a souped-up enhancement, HSDPA. Using some of the same techniques as EV-DO, including sharing of channels in the time domain, efficient resource schedulers, higher-order modulation and dynamic coding, HSDPA will boost average throughput rates to about 1 Mbps, with peak theoretical rates of 14 Mbps. Cingular Wireless plans to test HSDPA in Atlanta and says it aims to have it widely deployed by 2006.

Meanwhile, NTT DoCoMo in Japan, the first carrier to deploy UMTS, plans to upgrade its Japanese UMTS network to HSDPA next year. Some marketers are calling HSDPA 3.5G technology, though that is not an official term. HSDPA also has other enhancements in store, such as multiple receive antennas and equalizers on devices.The investment and technical effort required to develop, manufacture and deploy these next-generation networks is so vast that it is feasible only as a global effort, with R&D distributed across many different companies. The CDMA Developer Group reports that there are 87 CDMA2000 1xRTT and 10 1xEV-DO networks across Asia, the Americas and Europe. 3G Americas indicates that GPRS is available in 201 countries from 217 operators, that 34 operators have turned on EDGE service, and that 46 commercial UMTS networks are in operation in 21 countries.

Efforts are ongoing to increase throughput rates, with some systems achieving average rates of more than 100 Mbps in tests. These are sometimes called 4G, but there is no 4G standardization under way. Other 4G concepts include dynamic selection across multiple network types; continuous sessions across networks; and handling all functions, including voice and multimedia, in the packet domain. Other technologies much less distant than 4G, many real and available, offer an alternative to cellular (see "WiMAX and Beyond,").

Nextel has been successful with data applications for its iDEN technology. Speeds are limited to about 20 Kbps, though low latency has made its push-to-talk technology effective. Nextel plans an enhancement, called WiDEN, that will quadruple data rates.

Surprisingly, despite the multibillion-dollar investments being made in these 3G deployments, our poll results indicate that a majority of respondents think Wi-Fi will likely play a more prominent role in future mobile wireless systems than 3G cellular technologies. Clearly, the wireless industry needs to do more to promote awareness of its services (see "Throughput Rates,", for a side-by-side comparison of the technologies).

Enterprise Implications

Never have enterprises had so many wireless data choices. Remember when CDPD was the only IP-based wireless network technology, offering throughput rates of about 10 Kbps in only some parts of the country? That was just four years ago. Speeds have since increased by an order of a magnitude, service is available from an array of operators, coverage is the same as for voice and usage-based pricing has dropped like a stone. Yet data revenue at best represents less than 5 percent of total service revenues. Only 31 percent of respondents to our reader poll had any business-driven wireless deployments under way. What's the disconnect? And what are the implications for future services?

The greatest obstacles are confusion about offerings, concerns about security, cost and sparse coverage. Here's how to work through each:

• Confusion. There are a plethora of technologies from a variety of operators, and it takes some research to determine the best option. Should you use 1xRTT or EDGE today, or wait for 1xEV-DO or UMTS? We don't expect one technology family to prevail over another for the next several years. Today's networks are more than capable for a large number of applications. Most apps that work over the Internet will work over the wireless WAN networks today, and performance will only improve with forthcoming services. The best first step is to test how applications of interest work. If performance lags, consider employing some form of optimization. Most operators have Web optimization, but that won't work if you use VPNs, as encrypted traffic is not compressible. However, some companies offer enterprise optimization solutions so you can enhance your traffic stream before your VPN server touches it. This adds to the development effort but makes sense for wide deployments. But before you deploy anything, make sure your applications are configured for optimal wireless operations.

There's also the question of which applications make the most sense for wireless. In our poll, readers said wireless e-mail has the biggest impact on productivity. Other desirable applications included Web access and organization-specific applications. Wireless instant messaging, field service automation, transportation and logistics, and salesforce automation rated as medium impact. Video phone service rated as relatively low--news to 3G operators that view video telephony as a killer application.With PDAs, smartphones and notebook computers all candidates for wireless communications, another source of confusion has been which platforms to use. All make sense, and all promise significant productivity benefits, but as our previous piece on wireless PDAs and smartphones explained (see "Device Diversity," at ID# 1507f1), there are a lot of problems to solve. In our poll, notebook computers with integrated wireless (built-in or PC Card modem) were rated as the platform of greatest interest, followed by PDAs with integrated wireless, then smartphones; of least interest were cell phones with microbrowsers.

• Pricing. Operators are offering flat-rate plans of around $80 per month. T-Mobile is much more aggressive at $30 per month for GPRS. When we asked readers what monthly flat-rate pricing would be reasonable for wireless service that delivered throughput of 200 Kbps to 300 Kbps, 75 percent said under $50 for general purpose networking. Nevertheless, operators appear to be signing up plenty of customers. Data charges are typically much lower for smartphones and microbrowsers. For example, Sprint charges only $15 per month for unlimited 1xRTT usage with the Treo 600.

• Coverage. Fortunately, services like 1xRTT and EDGE are available nationwide. However, this may mean that, despite all the advantages and capabilities of next-generation services such as 1xEV-DO and UMTS, adoption of these technologies will be slow until service is widely available.

Peter Rysavy is president of Rysavy Research, a consulting firm specializing in wireless networking.

Organizations looking to equip their employees with wide area mobile data services have a daunting range of options. And to add pressure to complexity, some cellular providers are hoping to make end runs around IT by appealing directly to business line managers. If you're smart, you won't let these decisions be made ad hoc--there's too much at stake. First, what technology is best--CDMA, GSM/UMTS, iDEN or a next-gen offering to be named later? Where will FLASH-OFDM and WiMAX fit in? Then there's a gang of carriers, from AT&T to Verizon, vying for your business. Finally, what devices will you support for data access--smartphones, BlackBerrys, notebook PCs or a combination?We try to clear away the evolutionary sludge and help you pinpoint the best choices for your enterprise needs. In "Wide Area Wireless Data: Forever Evolving," we run down current and emerging wide area wireless data services, including broadband developments, and discuss specific technologies and various paths for enterprise adoption.

To help you choose a carrier, "Survival of the Fittest" looks at the strengths and weaknesses of the six major nationwide U.S. cellular carriers: AT&T Wireless, Cingular, Nextel, Sprint PCS, T-Mobile and Verizon Wireless. We also round up a selection of wireless NICs in "Wireless on the Road" and smartphones in "Are Smartphones Lagging Behind?". Finally, for those who live near Research Triangle Park in North Carolina, we took Nextel's Wireless Broadband Service based on Flarion's FLASH-OFDM for a test drive. "High-Speed Cruising in RTP" sums up our impression.

As if today's wireless landscape wasn't confusing enough, a variety of new technologies are being readied for the wide area. These are not attached to the long legacy of cellular technologies, but instead are designed as optimal conduits for spectrally efficient, high-speed data. These technologies provide voice functions, but instead of being handled as circuit-switched traffic, voice is done using VoIP.

Several companies have proprietary approaches; others are standards-based. Flarion Technologies offers its proprietary FLASH-OFDM (Fast Low-latency Access with Seamless Handoff and Orthogonal Frequency Division Multiplexing). Nextel Communications is running trials of FLASH-OFDM in North Carolina (see "High-Speed Cruising in RTP,. With latency of about 50 milliseconds and claimed throughputs of 1.5 Mbps in a 1.25-MHz radio channel, FLASH-OFDM sounds impressive.

Another player is ArrayComm--its iBurst technology offers speeds of 1 Mbps and is being deployed by operators in Australia and South Africa. Of course, a large number of companies offer fixed-wireless service, but this is an entirely different segment.As for standards-based approaches, one company active in this area is IPWireless, which is using the time-division duplex mode of UMTS/ HSDPA specs for a data-only service, achieving data rates of more than 4 Mbps. Service is available in London, and 10 global trials are under way.

The most prominent standards-based approach, however, is based on the IEEE 802.16 standard and promoted through the WiMAX Forum, an industry organization. This standard is complete, though only for fixed environments and hence offering service comparable to cable modems and DSL. Products are expected in 2005. A version of the specification still in development, IEEE 802.16e, is expected to be completed this year and adds mobility functions, including a more robust radio link suitable for mobile operation and handoffs between base stations. Once IEEE 802.16e is completed, it will match or exceed the capabilities of cellular designs, such as 1xEV-DO or HSDPA.

Finally, another standard, IEEE 802.20, in which both ArrayComm and Flarion are active participants, is addressing mobile broadband capabilities. However, 802.20 is much less advanced than 802.16e.

What types of operators will offer service based on these technologies? Existing cellular operators? New providers intending to compete with the cellular operators? There are no clear answers yet, but as the mobile broadband market matures, one thing is certain: There will be no shortage of companies and technologies vying for dominance--and your dollars.