Beyond 3G (LTE) but not on to 4G (yet)
>> The Extended Post-3G Roadmap
John C. Tanner
Wireless Asia
Feb 23, 2006
tinyurl.com
With 3G barely out of the gate and still evolving, the cellular industry is already plotting a course toward 4G. But while it looks linear and neat on a PowerPoint slide, the details are mind-bogglingly complex - which is why the first of the standards won't be ready until 2008 and 4G itself won't arrive before the end of the decade.
It was this time last year that NTT DoCoMo and 26 other companies announced plans to sit down and start seriously working on the next evolutionary steps for W-CDMA. Until that point, of course, everyone knew that the next "stage" for W-CDMA was HSDPA. Beyond that, however, was some vaguely defined cloud referred to generically as "4G".
NTT DoCoMo and friends didn't help matters by describing their project as "Super 3G" - which would feature data rates and technologies often associated with 4G, but wouldn't actually be 4G. The group did say, however, that it would be keeping the 3GPP in the loop on its work. The 3GPP, meanwhile, has not been idle, and now "Super 3G" finally has an official name.
This is it: UMTS Terrestrial Radio Access Node Long Term Evolution (UTRAN LTE). Not exactly snappy, which may be why some vendors have shortened it in their PowerPoint slides to LTE, or "Evolved UMTS", or even (perhaps a bit cheekily) 3.99G.
Whatever you call it, by the time the 3G World Congress last November rolled around, a lot of delegates and speakers were talking about it. And not just in regards to W-CDMA. On the CDMA side of the cellular industry, CDMA vendors were talking up cdma2000's evolution beyond EV-DO Rev A, which gives the technology more symmetrical data uplinks and downlinks. Little of it was a mystery - the CDMA Development Group has said for years that EV-DO would essentially be followed by a multi-channel, all-IP version, which is now known as EV-DO Rev B (formerly EV-DV). Still, the CDMA camp had clearly shifted from simply talking about it to actively promoting and developing a specific technology roadmap.
That said, the LTE (for both W-CDMA and cdma2000) isn't so much an evolution to 4G itself but a roadmap to help the cellular industry get there and do its part in bringing 4G to life. That's because 4G - at least as the telecoms/IT industries now understand it - will be nothing like 3G. 3G was primarily vendor and regulator driven, with the two main industry coalitions developing their own standards under the ITU umbrella standard IMT-2000. 4G will be a far more complex animal, encompassing a variety of technologies, from wide-area cellular and wireless broadband networks to mobile TV and personal-area networks using UWB, Bluetooth and ZigBee, all IP-enabled for seamless mobility between them.
Consequently, the details are mind-bogglingly complex, and the roadmap itself is still riddled with gaps. Little wonder then that for all of the grumblings among some industry players and analysts that it's too soon to be talking about 4G, others are saying that its sheer complexity demands that all players in the telecoms value chain start working on it now.
The Many Faces of LTE
Indeed, even the LTE itself is still in some ways a work in progress, as vendors continue to promote their own ideas to the 3GPP.
For example, it's well understood that the next step after HSDPA and HSUPA (the uplink version of HSDPA that will give W-CDMA more symmetrical data throughput speeds in the multi-megabit range) will involve adding technologies like OFDM and MIMO to the mix. Nortel Networks has already assigned a name to it - HSOPA.
Nortel has also been promoting its idea regarding IP in the LTE network, which it says should be a "flat" architecture that's as close to all-IP as possible to maximize mobility between multiple access technologies. Nokia, meanwhile, has been talking up an idea called Internet-HSPA (I-HSPA), which also promotes a simplified network architecture but recommends providing a direct link from the Node B to the Net, which would do an end run around the network completely.
Such technologies are not running outside the scope of the 3GPP's standards for Release 6 (which is finished) and Release 7. The same can't be said for initiatives such as Qualcomm's DMMX (DO Multicarrier Multilink eXtensions) and HMMX (HSDPA Multicarrier Multilink extensions), which the company says isn't a challenge or threat to the future roadmaps of UMTS and EV-DO, but rather, as Qualcomm's senior marketing VP Jeffery Belk puts it, "a platform-slash-philosophy out of Paul Jacobs' brain" that Qualcomm refers to as "concurrence".
"None of this requires anyone to make any changes to the 3GGPP or 3GPP2 standards," Belk says. "EV-DO will still evolve to Rev A and Rev B as planned, and UMTS will still upgrade to HSDPA and HSUPA."
What Qualcomm is essentially proposing to add to the mix is a series of its own optional techniques to enhance the performance and capacity of EV-DO and HSDPA that improve capacity on top of what the standards already call for. For example, a key objective is to enable "multicarrier multilink" - multiple wireless transmission protocols in multiple frequency bands being used simultaneously. Examples include using 3G uplinks to provide interactive services for mobile TV, and tying assisted GPS technology with HSDPA.
"It's about enabling concurrence as well as convergence," Belk says. "It's one thing to build a multi-mode, multi-band terminal. It's something else to be able to use two or more modes at the same time on different bands."
Examples of this "concurrence" could include watching real-time TV streams on a handset while it continues to monitor the paging channel for voice calls and cellular data, or enabling multiplayer online games with voice capabilities.
Belk says the first product announcements for DMMX and HMMX are due some time this year, but of the UMTS vendors Wireless Asia contacted, all declined to comment on their take on HMMX. Given Qualcomm's current legal wrangling over anti-competitive royalty schemes with the EU, the UMTS camp may treat HMMX with suspicion at the outset.
Belk acknowledges as much, but responds that Qualcomm would gain nothing by derailing the post-3G roadmap on either side.
"We own IPR to technology that's core to 3G evolution either way, so we won't be successful unless the manufacturers and OEMs and operators are successful," he said. "We support what they want."
Challenges Identified
Still, despite the individual vendor pushes on technological direction, there's little debate over the basic technological direction and the elements involved, says Markku Ellil? Asia-Pacific director of radio networks product management for the networks division of Nokia.
"Some of the technologies planned for 4G will be realized within 3G and its spectrum, as can be seen in 3GPP future evolution work," he says. "Graceful evolution of the 3G systems can provide a viable path to 4G networks. For example, 4G radio interfaces could be plugged in to the evolved 3G network."
Fawzi Behmann, director of strategic marketing for the Networking and Computing Systems Group for Freescale Semiconductor, adds, "When we look beyond 3G, at the very least we know which technologies will be involved - data speeds between 100 Mbps and 1 Gbps, an IP core, OFDMA, MIMO, voice and video over IP, mesh and seamless mobility."
"We also know that it will not be one technology but a convergence of multiple technologies - HSDPA. EV-DO, WiMAX/WiBro, Wi-Fi, UWB and DMB, forming a network of networks supporting multiple services."
Helpfully, Behmann adds, we also know what the chief challenges are in achieving all this: spectrum availability, deployment costs and developing terminals that can cope with a multiple-access environment and - perhaps more importantly - all the multimedia that the environment will be throwing at it.
The spectrum issue is of course mainly up to regulators while the cost of deployment will vary from one operator to another. One constant, says Behmann, is that existing cellcos will need a lot more base stations than they have now to get the kinds of speeds that 4G will demand.
"To support a data speed increase of ten times 3G, you will need three times as many base stations," he says.
Redesigning the Phone
The real challenge, though, will be the terminals, and the frontline for getting them ready for post-3G will be at the electronics level. At an IEEE wireless symposium in San Diego in January, Terry Lingren, VP of engineering for Kyocera Telecommunications Research Corporation (KTRC), provided a laundry list of requirements for the handsets of the future: integrated reconfigurable front ends, tunable filters, integrated baseband processors and DSPs that can handle multiple protocols, a separate application processor for niche markets to ease customization, multiple reconfigurable antenna structures, very large memories, and innovative UIs such as virtual displays, virtual keypads and intelligent voice recognition. And that's the short list.
At the core of all this is the trick of doing all this without draining the battery in a matter of minutes, says Behmann.
"4G will be about high-speed and advanced applications, so it's going to be intensive in terms of processing power, and it's going to drain the battery fast," he says. "Imagine a PC running at 100 MIPS. The challenge is how to create that kind of environment on a cellphone."
On top of that, it will have to be done on the assumption that battery capacity isn't going to get much better in the next few years, says Lingren.
"Batteries don't follow Moore's Law. Li-ion capacity has increased 9% since 2002. There needs to be a way to get higher storage on batteries," he says.
Fuel cells are believed to be the solution to that problem, but Lingren says that despite their promise, "they need to be smaller, and that's still several years away."
That's why semiconductor companies have been on the case for some time, developing simpler, integrated chipset architectures that pack more processing power and consume less power. Companies like Qualcomm, Intel, Texas Instruments, Philips Semiconductors, STMicroelectronics and Freescale are working on microprocessors with process nodes from 90 nm down to 65 nm, while companies like RF Micro Devices are working on the front end of the handset to develop low-cost high-volume solutions on a single chip die.
"Front-end modules need to switch between different frequencies and antennas, different bands, transmit and receive, so you need to add more of that functionality in a smaller area," says Frank Della Corte, RF Micro's marketing manager for its digital cellular product line.
Behmann of Freescale says that OFDM can also be exploited in combination with parallel processing to save on power consumption.
"OFDM divides data into sub-bands, and each of those can be processed independently. You could do the same with space and time slots in parallel," he says. "If you increase that and do hardware acceleration, you can minimize drainage."
That said, hardware acceleration and parallel processing are still in the fledgling stage themselves, Behmann adds. "It will take a few years to get it into a proof of concept."
Out Of Sync
Even once such issues as processing speeds and power consumption have been sufficiently tackled, the real challenge for many will be multiple access integration in the handset. Many see it as a design challenge in terms of radio design and footprints, but Della Corte says that the uneven standards process between different access technologies is also proving to be a significant obstacle.
"The GSM standard is stable and mature, but W-CDMA is still somewhat unstable, and WLAN is still moving forward on several levels. Then you've got Bluetooth, ZigBee and UWB, which are at different levels of maturity," he says. "Our customers don't want to select a solution that's going to hamstring them six months later. Until the standards you want to integrate are mature and stable, the integration trend isn't really there."
There are also external factors, says Behmann - most of them regulatory.
"We're seeing some early initiatives [for 4G], but they're separate. A lot of it is driven by local regulations and spectrum allocation plans, which makes interoperability difficult," he says. "Within a country or community it's more feasible, but doing it between countries is more difficult."
Consequently, says Nokia's Ellil "4G is still something we're unlikely to see - at least in any meaningfully standardized form - before the end of the decade."
"The earliest expected time for initial 4G standards to be ready is around 2010 to 2012, subject to the outcome of the World Radio Conference in 2007," he says.
Carriers Want In
Little wonder that some industry players and analysts are saying that it's too soon to be talking about 4G when the industry is still getting its head around 3G. They may have a point. The Global mobile Suppliers Association regularly touts optimistic sounding statistics on W-CDMA adoption - the latest numbers report just over a hundred W-CDMA operators at the end of 2005. However, out of a base of around 670 GSM operators worldwide that are slated to go 3G eventually, that works out to less than 15%, suggesting even 3G still has a long deployment road ahead of it.
On the other hand, given the amount of work ahead - and given that other wireless technologies like Wi-Fi, WiMAX, ZigBee and others are moving forward to assume whatever role they will play in the 4G future - it arguably makes sense to get cracking on the technology now while cellcos put 3G technology and next-gen business models through the hoops.
One interesting development in all this is that the operators themselves are so determined not to be left out of the loop this time that they are talking directly with the semiconductor companies, says Behmann of Freescale.
"Carriers are coming directly to us now. They want to be more in the loop with the vendors and the semiconductor manufacturers as the technology develops so they can direct it and have a voice," he says. "They want to collaborate with us to develop and understand this so that they can get what they want and need to serve their customers' demands. And they're now starting to recognize the complexity involved."
4G: a Pocket Guide
What you'll need:
- HSPA
- EV-DO Rev A/B
- WiMAX/WiBro
- Wi-Fi
- UWB
- ZigBee
- Mobile TV (DMB/DVB-H/MBMS/MediaFLO)
- IP core
- OFDM
- MIMO
- VoIP
- Video over IP
- Mesh networking
- A way to make all this seamlessly mobile
Top Three Challenges
- Mobile terminal design (including processing power, battery demands and radio integration)
- Spectrum availability
- Overall cost of deployment
Handset Requirements
- Integrated reconfigurable front ends
- Tunable filters (MEMS or ferro-electric, etc)
- Integrated baseband processors and DSPs that can handle multiple protocols
- Separate application processor for niche markets to ease customization
- Multiple reconfigurable antenna structures
- A whole lotta memory (see: iPod)
- Innovative UIs <<
The article below by Robert Clark references Tanner's article above and speaks to the issue of the the "Politicization of 4G."
If you thought 3G standardization was politicized and complicated, wait till you see 4G. 3GPP will plod on for another five years to deliver a gargantuan, bloated compromised "family" of standards, backwards-compatible with every known radio and meeting the political needs of the US, the European and Chinese lobbies. ... Good luck with that, but by then the IP horse will have bolted from the IMS stable. Just as IP is eating everything in the enterprise networks and is becoming the default for consumer wireline, it will likewise dominate wireless.
>> The Short Life and Death of 4G
Robert Clark
Telecom Asia
Feb 28, 2006
tinyurl.com
Wi-Fi and cellular have been cagily eyeing each other for half a decade, varying between suspicion and indifference.
But at Barcelona two weeks ago the cellular industry delivered its verdict on WLAN: inevitably, it's about integration.
The news came from across the board. Intel and the GSM Association agreeing that Centrino will support HSDPA, Hutchison adding a Skype client, Motorola and Nokia announcing a bundle of new Wi-Fi-enabled phones.
Some traditional comms vendors will trumpet this as the triumph of HSDPA over Wimax; that with its capacity to deliver 1 Mbps or so in real bandwidth the 3.5G standard makes mobile broadband a reality, forcing the Wi-Fi-Wimax crowd to a compromise.
But the reverse is just as true. HSDPA handsets in volumes and variety are a good two years away (though some operators are claimed to be planning a PC card service later this year).
Meanwhile, mobile operators are under a pincer attack from the device segment and the growth of Wi-Fi.
The cellular-Wi-Fi phones will see a slice of operator revenues at the top end of the market.
Wi-Fi networks continue to expand. Mesh networks are a reality in a number of US cities, in the London financial district and places like Taipei and Kaohsiung. Some might question the capacity of mesh, but that is not an issue for early adopters. They will help create a community of wireless IP voice users, bypassing cellular.
So the news from Barca shows cellular acknowledging reality and going where customers want it to go, which is to have access to data through the wireless tech of their choice, be it UMTS, HSDPA, Wi-Fi or some short-range standard.
All to the Good.
And whatever you say about it, the business convergence of UMTS-HSDPA and the Wi-Fi-Wimax camps is pretty simple. There's no difficulty in putting HSDPA into laptops and Wi-Fi into phones.
Would that you could say the same about 4G. Colleague John Tanner, writing in Wireless Asia, has described it as "mind-bogglingly complex". Were talking HSDPA, EV-DO Revs A and B, Wimax, WiBro, Wi-Fi, UWB, ZigBee, the various flavors of mobile TV, IP core, OFDM, MIMO, VoIP, video over IP, and mesh.
Why bother? Well, the 3GPP community will. If you thought 3G standardization was politicized and complicated, wait till you see 4G. 3GPP will plod on for another five years to deliver a gargantuan, bloated compromised "family" of standards, backwards-compatible with every known radio and meeting the political needs of the US, the European and Chinese lobbies.
Good luck with that, but by then the IP horse will have bolted from the IMS stable. Just as IP is eating everything in the enterprise networks and is becoming the default for consumer wireline, it will likewise dominate wireless.
3GSM Barcelona marks the end of top-down cellular governance by the old communications world. IP over high-speed radio puts the customer in charge and makes 4G an irrelevance.
Robert Clark is a Hong Kong-based technology journalist and analyst <<
- Eric - |
