In the fickle world of technology, it doesn't take long for an innovation held out as a shining hope to become a commercial flop. If any telecom technology were such a case, it would be CDMA 1X evolution data/voice, more commonly known as EV-DV.

Just last year, vendors were performing live demos of EV-DV at CTIA, and Sprint was insisting the technology would comprise the basis of its 3G migration path. Today, telecom executives are surprised to hear the acronym in conversation. There are still some vendors that have markers on the technology, but for the most part, the technology is on its last legs. Instead, the industry is focusing with laser precision on its competitor EV-data optimized, or EV-DO. Last year, EV-DO was taking hits for what its namesake implied — the limitation of carrying only data on a single channel — while EV-DV was touted because it merged data and voice on a single channel. But from the way carriers and vendors talk now, there seems little EV-DO can't do, whether it be merging voice and data, broadcasting video or bringing immediate gratification to what carriers want for a mobile broadband data network today.

“A year ago, we would go to meetings, and our customers would always ask about EV-DV,” said Amit Jain, director of product management for EV-DO technology vendor Airvana. “Now what they want to know about is how EV-DO revision A is coming along and how VoIP is performing over it. … The momentum for EV-DV is gone. I'd be very surprised if we see a single EV-DV demo at CTIA this year.”

Tracking EV-DV's decline is a fairly simple process — its fall is almost inversely proportional to EV-DO's success, starting with SK Telecom's deployment of EV-DO in 2002 to Verizon Wireless' and Japanese carrier KDDI's adoption of the technology. The critical holdout, however, was Sprint, EV-DV's last champion, and its signaling of an EV-DO rollout last year seemed to be the death knell for EV-DV.

Whether or not EV-DV is disappearing for good, people in the industry think EV-DO will be around for a very long time. Most of the largest carriers in the major CDMA markets have either rolled out or have committed to deploying EV-DO networks. Sprint, Verizon, KDDI and SK Telecom have been joined by Vivo in Brazil, Optus in Australia and most recently by Bell Mobility in Canada, which had also been another major proponent of EV-DV.

Many of those carriers have also indicated they would pursue EV-DO down its migration path to revision A, the first cellular access technology capable of supporting end-to-end voice over IP (VoIP). In response, most vendors have either shelved their EV-DV plans entirely or have backed off from the technology to focus on EV-DO.

“The market demand for DV is rapidly disappearing,” said Mike Iandolo, vice president of CDMA product management for Lucent Technologies. “It's quite possible that we'll never have to build a DV system. We're still doing all of the long-lead items for the technology, but clearly we're focusing all of our resources on DO. The only reason we don't take DV completely off the table is the carriers haven't all made up their minds.”

Before Sprint's commitment to EV-DO, the carrier's view was that both technologies would co-exist, EV-DV coming out a year or two after the initial data-only rollouts. Even after declaring its DO intentions, Sprint said it was still weighing a possible deployment of DV to add voice capabilities to its new broadband data channels. But Sprint's announcement in favor of EV-DO at last year's Supercomm immediately fueled speculation that DV had been handed its pink slip, and despite Sprint publicly keeping its options open, it had set the industry firmly on the EV-DO migration path.

But Steve Searles, Nortel Networks vice president of marketing for CDMA, believes that the decisive point for EV-DO had come long before. The technology was already gaining momentum with Asian carriers, but it was when the world's largest CDMA carrier, Verizon Wireless, announced it would pursue EV-DO in 2003 that the true shift in momentum occurred. Verizon's size wasn't the only consideration. It was also the competitive peculiarities of the U.S. market. Verizon's decision set off a chain of events that caused all North American carriers to re-evaluate their technology plans.

“The day that Verizon announced it was doing DO, the rest of the industry knew it was doing DO,” Searles said. “Sprint needed to come up with a competitive response to Verizon. DO was the only technology available. It was a forgone conclusion.”

Nortel is one of the few vendors that will completely write off EV-DV publicly. The reason is simple market economics, Searles said. The CDMA market simply can't support two competing technologies providing essentially the same capabilities. Vendors would be hard-pressed to support two separate development tracks without high carrier demand for both, but spreading the industry across two distinct infrastructures would undermine the economies of scale from having a unified platform.

The business basically comes down to one of handsets, Searles said. Even if an EV-DV network could offer immediate benefits to operators looking to deploy voice and data on a single 1.25 MHz channel instead of deploying twice the infrastructure, any operational and infrastructure savings would be immediately offset by higher prices for EV-DV handsets, Searles said.

“Handsets prices are determined by chipset prices, and chipmaking is a volume-based business,” Searles said. “That's why Sprint and Verizon jumping on board the DO bandwagon put the final nail in the coffin of DV. The industry simply won't have the volumes to support both technologies.”

One thing you'll find vendors and carriers agreeing on is that both EV-DO and EV-DV were equally viable technologies. In fact, if you look at the standards set out by the 3G Partnership Project, or 3GPP, for both technologies, the final characteristics of each look remarkably similar when the their different approaches to voice are ignored. Clearly, it wasn't technological superiority that caused DO to win out over DV. In fact, one of the main reason's EV-DO gained such a big lead over DV in the minds of carriers was simply its timing.

“DO is what's commercially available now, and it's what we're focusing on,” said Eduardo Conrado, senior director of systems marketing and portfolio management for Motorola, which is continuing its EV-DV development. “DO may be the only thing definite now, but this will be a multi-year play. We'll have to wait for carriers to evaluate their networks and make their final technology decisions.”

While being first to market clearly has its advantages, it didn't necessarily mean that EV-DV didn't have a chance to recover. EV-DV's biggest advantage is its ability to give carriers both voice and broadband data on a single channel, using the same legacy digital voice technology operating over current 1x networks. EV-DO, as its name implies, is an optimized data technology, and its current incarnation, revision 0, requires a dedicated 1.25 MHz channel to deliver broadband data speeds of up to 2.4 Mb/s on the downlink.

In that configuration, a carrier with a 5 MHz license would have to reserve one channel for data and run 1X voice over the other two. That's not a problem for deployments in major markets, where carriers are trying to use their spectrum in the most efficient manner, but in suburban and rural markets, operators often use only one channel for both services. EV-DO forces them to use two channels in those markets. Coupling EV-DV voice capabilities with a higher initial downlink speed (3.1 Mb/s), it's easy to see why EV-DV could still have some appeal.

But then EV-DO revision A hit the scene. The second installment of DO technology not only pushes reverse link capacity from 153 kb/s to 1.8 Mb/s, but it incorporates quality of service (QOS) controls that keep latency from degrading as the network gets overloaded. Upstream capacity and low latency are the two critical elements to deploying VoIP, and EV-DO suddenly had its rebuttal to EV-DV's voice claims.

At first carriers expressed skepticism because end-to-end voice wasn't expected to occur until much farther up the evolutionary ladder. But in the last two years, VoIP on the wireline has taken off, spawning numerous new technologies that optimize IP telephony over phone lines, coax and Wi-Fi. The IP multimedia subsystem (IMS) core also gained momentum in the last year, providing not only a SIP-based conduit to run VoIP but also the promise of convergence and enhanced services to be built off VoIP.

The deal breaker, however, came last year when the 3GPP2 standards body released its first details about how VoIP would operate over a revision A-upgraded network. Operators that were worried VoIP couldn't provide carrier-class voice had the data the industry needed to show VoIP could work, said Gennady Sirota, vice president of product management for Starent Networks, a core technology vendor for CDMA networks.

“VoIP had gone from mere theory to the labs,” Sirota said. “That was when the carriers became convinced.”

It was certainly the argument that convinced Sprint to abandon its EV-DV plans and commit to later versions of DO. Oliver Valente, Sprint vice president of technology development, said Sprint spent a lot of time evaluating both technologies, and while it had been leaning toward revision A, it wasn't until late last year that Sprint was comfortable enough with the technology's support of VoIP to commit fully to the upgrade. (Its vendors agreeing to supply EV-DO and the upgrade for the same price probably helped, too). Sprint is pursuing an aggressive deployment timeline, launching its first trials of revision A technology late this year and in early 2006, with a full rollout in the 2006/2007 time frame and an immediate launch of VoIP services. The carrier is scheduled to launch its actual 3G network this year, leading to speculation that Sprint might choose to launch EV-DO and its upgrade simultaneously.

Sprint's aggressive rollout plans aside, most carriers probably won't have revision A installed until 2008, and even then, most will probably approach VoIP cautiously, launching push-to-talk and video conferencing services first before risking their core voice revenues over the new technology. Meanwhile, some vendors' EV-DV technologies are supposedly in the final stages of development. An EV-DV network could feasibly be up much faster than its VoIP-enhanced DO counterpart, giving carriers immediate access to new voice capacity.

That reasoning has prevented most vendors from letting their EV-DV product lines go stagnate. Ericsson, Lucent and Motorola all have EV-DV in some stage of development.

EV-DV has been in development just as long as EV-DO, even longer if you consider that its part of the standard 1X evolution path standardized under 3GPP2's CDMA2000 (EV-DV encompasses revisions C and D of the 1X migration path). In fact, while EV-DO definitely enjoyed some time-to-market advantages with SK Telecom's rollout in early 2002, many vendors have been preparing commercial EV-DV equipment for the last two years. One of the main reason's that equipment never made it to market was Qualcomm.

Unlike the GSM world, where the 3GPP imposes strict standards that every vendor builds to, the CDMA world is a bit more fluid. And Qualcomm holds an unusual amount of sway for the wellspring of CDMA's core technology. Qualcomm manufacturers the majority of the chipsets that make it into CDMA handsets and base stations, making vendors very reliant on its silicon time tables. In the case of EV-DV revision C, Qualcomm simply didn't release a chipset, choosing instead to focus on revision D and its higher uplink capacity. The move effectively killed off revision C and several vendors' plans to deploy the technology alongside EV-DO.

In 2003, Qualcomm announced the CSM and MSM 6700, a line of chipsets that combined all CDMA2000 technologies through EV-DV revision D and EV-DO. At the time, Qualcomm said it would begin sampling the chipset in 2004, but since then it hasn't publicly announced any sampling or commercial shipments, and news outlets in Korea have begun reporting that Qualcomm has killed off its EV-DV development entirely. If Qualcomm is truly out of the EV-DV business, it represents the biggest hurdle for the technology's future. If there are no Qualcomm chips, there simply won't be any phones, except from that handful of vendors that presses their own 3G silicon.

One such vendor, Samsung, said it's remaining undaunted by the growing obstacles to EV-DV. Though Samsung was one of the original innovators of EV-DO (it launched SK Telecom's network in 2002) and is still committed to its DO product line, it remains the only major vendor that hasn't backed away from EV-DV. It has readjusted its strategy, canceling its plans to launch EV-DV revision C, but Jim Parker, Samsung wireless systems portfolio manager, said the vendor is aggressively developing a commercial product line based on revision D of the EV-DV standard.

What's more, Samsung isn't just touting EV-DV as an alternative to EV-DO. Parker said Samsung still believes EV-DV comes right after DO in the evolutionary path of CDMA 3G — that it's the only technology that can provide both broadband data and carrier-quality voice. Of all the vendors, Samsung remains unconvinced that EV-DO revision A can support VoIP. While VoIP is an interesting concept in theory, latency issues and the other peculiarities of packetizing an already digitalized and highly compressed radio transmission will present far more problems than other vendors are admitting, Parker said.

In addition, voice and data are different animals. Voice requires low bandwidth but extremely high QOS and latency requirements, while data is bursty with high bandwidth requirements. They're polar opposites, and the network must be optimized to handle both aspects. To toss both voice and data onto the same packet stream is just asking for problems, Parker said. When it comes down to the wire, carriers won't be willing to take their chance on a VoIP platform that gives them a lower-quality voice product, no matter what spectral and operational advantages the technology may have, Parker said.

“Think back to the old days of overseas phone calls when you had up to three-second delays,” Parker said. “We don't want to go back to the tin can.”

If anyone has sway with EV-DV, it could be Samsung and fellow Korean CDMA vendor LG Electronics. They're the dominant vendors in South Korea, where the government has allotted specific spectrum for an EV-DV launch, and both vendors have been supporting trials of the technology there. LG actually owns licenses itself and is running a live EV-DV network. Obviously, LG would face a huge setback if its chipset provider, Qualcomm, stepped away from EV-DV, but with an actual EV-DV network running in the hotbed market of CDMA innovation, it's hard to say the technology is completely dead.

“We've always viewed this as a matter of carrier preference,” said Perry LaForge, executive director of CDMA Development Group. “I don't hear much debate anymore over DO versus DV. The market has decided, which is the best way for these decision to be made. But never say never. At the end of the day, EV-DV could be back.”

CDMA evolution technologies

EV-DV:

REVISION C

Forward link	3.1 Mb/s
Reverse link	307 Mb/s

This first mobile broadband implementation of the CDMA 2000 line faded away quietly as it didn't get enough support from vendors or carriers, mainly because of its low upstream capacity.

REVISION D

Forward link	3.1 Mb/s
Reverse link	1.8 Mb/s

With a much faster uplink, carriers sought to skip over revision C for this implementation, allowing them to offer more symmetric broadband services while preserving the voice.

EV-DO:

REVISION 0

Forward link	2.4 Mb/s
Reverse link	153 kb/s

The only commercially deployed EV technology, EV-DO provides impressive bandwidth, but its already high latency degrades beyond 300 milliseconds as network loads increase. That coupled with its small downlink and lack of QOS make it unattractive for many real-time communications applications such as voice.

REVISION A

Forward link	3.1 Mb/s
Reverse link	1.8 Mb/s

With a low latency (between 30 to 50 milliseconds, vendors claim), QOS packet prioritization and a robust uplink, vendors and carriers are pointing to revision A as the perfect platform on which to overlay VoIP as well as other real-time multimedia applications.