MCI on EDGE
>> A Genuine Contender?
Vaughan O'Grady
Mobile Communications International
Issue 98
01 February 2003
Once apparently sidelined, is EDGE about to make a triumphant return? It will soon take on CDMA2000 in the Americas, and both Asia and Europe are reassessing this narrowband high-speed data solution. But it's not that simple. Questions are still being asked about whether the promise of high data speeds at low relative costs will be met - and indeed whether other technologies might have done a better job ....
Prove it! That's what the GSM industry said in 1997, after it heard the voice capacity claims of supporters of the then upstart IS-95 CDMA technology. And often they couldn't - stymied by new networks and unwilling to let competitors know how many base stations they were actually rolling out.
But the tables have turned. In 2003 what is now called CDMA2000 is in the market and identifiably performing. Its proponents now finally have the luxury, when they hear the data capacity claims of the EDGE community, of sitting back and repeating exactly the same words they heard six years ago.
Not surprisingly, the result has been claim and counter-claim and, in an industry less buoyant than in 1997, there is - appropriately enough - a real edge to the debate. But what is there to prove? Essentially, the debate rages around technology, cost, likely take-up and alternatives.
The official version of what EDGE is and does goes something like this: EDGE is a technology developed to enable the transmission of large amounts of data at a high speed. It uses the same TDMA frame structure, logic channel and 200kHz carrier bandwidth as today's GSM networks, all of which allows existing cell plans to remain intact. It uses a new modulation scheme within the existing GSM carrier frequencies and makes more efficient use of the available spectrum - specifically, data rates three times higher than GPRS. With a maximum eight timeslots up and downlink, each with a maximum 59.2kbps capability, a staggering 473 kbps data rate is possible.
The story so far is a brief one, as Jussi Ware, VP marketing and sales, GSM and EDGE systems, of Nokia points out. "The standard is not that old. The actual standardisation work started in 1998 and before that in 1997 there was a feasibility study for what at that time was the GSM2+ system. It was standardised and finalised in March 2000."
Leo Nikkari, chairman of the GSM Association's EDGE Taskforce representing GSM operators, brings us up to date: "I think when attention shifted to the planning and the investments for UMTS spectrum and plans for UMTS deployment, a lot of promotion - and even understanding - of EDGE got left by the wayside." Then, of course, there was the brief introduction of an alternative - TDMA-EDGE - before the decision of US TDMA operator AT&T Wireless Services to choose the GSM route effectively rendered only one solution possible.
And now? "With the reality of the marketplace, and a much more heavily scrutinised review of any proposals for capital expenditure in your business plan, EDGE has come back as a serious discussion point for all of the GSM operators," says Nikkari.
If this is so, in which countries will EDGE be introduced in 2003? Lacking any immediate source of new spectrum to make large-scale UMTS rollout likely, the US will almost certainly be first through AT&T and Cingular.
Other operators from the US should follow in 2004, but Jussi Ware is surprisingly optimistic about the rest of the world, suggesting that Europe and Asia could be added to the list by the end of the year. Doron Elinav, director of product management of Comsys, a company that develops and integrates physical layer solutions for a number of wireless standards, including EDGE, suggests that for tier two operators, with fewer customers, "WCDMA will not bring the same amount of savings of increased capacity because the traffic isn't there. It would be worthwhile for these operators to use EDGE just from a cost perspective... I see this mostly for, today, South East Asia and the Asia Pacific rim." In Europe, by contrast, he feels this approach is less feasible, "because of internal politics and how the financial markets would regard an announcement to go to EDGE versus the amount of money that was put into the licensing procedures for 3G."
Svetlana Issaeva, manager, Europe/EMEA of Pyramid Research, looks at it from a slightly different angle: "We were very sceptical about it a year ago and did not in fact see it taking off at all in Europe [but] we are basically reconsidering it now because there is solid ground for it to grow." However, she suggests a few considerations. "Number one is financial - and in western Europe there are pretty stringent 3G rollout requirements. From that viewpoint very few operators have the money. So in Europe it's most likely to get rolling probably after the 3G rollout - I would say 2004/2005 even for EDGE - and it's going to be complementary to WCDMA, covering the densely populated areas where WCDMA doesn't have sufficient coverage, or alternatively for the rural areas where it doesn't make sense to roll out WCDMA at all. And because there's going to be interoperability between these two nothing stops operators from rolling out both."
But how quickly will EDGE roll out - even in its key markets? That may be defined by market size, Elinav suggests: "I would estimate the first rollout would just be to give basic coverage. There won't be many EDGE handsets, there won't be the need for many services, so the first phase would just be a broad service upgrade and then, once service picks up, an upgrade for the capacity will be added."
The likelihood of a phased rollout makes it difficult to find an immmediate answer to one contentious question. Can EDGE deliver the data rates promised - or will they degrade seriously as one moves away from the centre of a cell?
As wireless consultant Andrew Seybold puts it: "My understanding is, if I put EDGE into a GPRS network, the data speeds at the edge of my cell are going to be no better than GPRS. In other words I'm actually, to take advantage of EDGE, going to have to build out more cell sites."
Geoff Varrall, executive director of research group RTT programmes, expands on this: "EDGE is going to disappoint in terms of its radio layer performance.
When you're close to a base station it'll get quite high data throughputs - even up to 380kbps. That's assuming your noise thresholds are OK; there are even radio engineering aspects when you're close to a base station. Assuming you've got quite a good RF link budget, then it is reasonable to expect some improvement in performance. The problem is it tends to be quite variable, and inconsistent - and of course you never know whether you're close to a base station or not."
But, if proponents are to be believed, EDGE is more than just 8-PSK modulation.
Ware lists such key enhancements as incremental redundancy (higher reception probability, which gives gain in poor radio conditions), new modulation coding schemes (both GMSK and 8-PSK) and enhanced link adaptation (which he suggests gives highest gain in good radio conditions). "These features together," he says, "form an EDGE system that in all conditions offers higher capacity, better coverage and improved quality than GPRS."
But the likely data speeds - at least in the early stages - will not be anywhere near the top rates in any case, in part because of the need to guarantee terminal availability. As Nikkari points out, rather than the near-impossibility of asking for fully-featured and tested phones, operators have agreed a "prioritised consensus list of features for rollout". Beyond that, he says: "We'll be defining four phases of development for the EDGE devices, through and just beyond fourth quarter 2004."
What this means in practice is fewer timeslots and a lower maximum data throughput. Ware notes that, "the first terminals are expected to have two to three timeslots and the networks are likely to be dimensioned to average 40 kbps per timeslot; this would introduce 80-120 kbps data rates with EDGE."
Northstream's Hakan Olofsson suggests good reasons for this. To guarantee attractively priced terminals on time "you trade that technical capability...
The handsets launched now for EDGE - for example the Nokia and Motorola ones - are only utilising a subsect of EDGE's maximum capabilities, like most commercial terminals for other technologies." However, he suggests that, user experience would still be more than acceptable.
Eventually, Ware says, "most of the GPRS terminals that are coming to the market - at least from Nokia - will have EDGE inside towards the second half of this year. Beyond this, Issaeva notes, Nokia has plans to make all WCDMA handsets EDGE-enabled - and is unlikely to be alone in either offering.
In fact EDGE-capable kit has been shipped by vendors for a while. But is that really to the advantage of new networks? Are there other, hidden costs? And what about old legacy networks? Will they not require new equipment?
Certainly, neither Varrall nor Seybold are convinced. Seybold argues that, "along with GPRS and EDGE, there's something a lot of people ignore which is that you need then to spend time beefing up your back end - which can be expensive. This means increasing the number of wirelines you have going to a cellsite, increasing from a T1 to a T3 or upgrading your backhaul capacity in some way." Varrall has similar objections, arguing that it is, "a major physical change...you're trying to make GSM do something it was never really designed to do and it means that both in the handset and the base station - in physical terms - you've got to upgrade the power amplifiers both in the base station down to the handset and handset back to the base station. That's a fairly major upgrade."
He admits that, "Ericsson claim quite a lot of base stations are EDGE-compatible. But then you're putting these EDGE power amplifiers into the handsets as well, and that adds to the handset cost, the handset power budget, which means you've got to have more linearity in the handset."
And again he raises the question of base station deployment: "You've got to have your base stations with sufficient density to make it work nicely and you've got to have them in the right place, you've got to watch your noise floors, you've got to look at interference issues - so you do have to revisit your radio engineering and make sure that you've got a good robust radio layer... And it involves a lot of driving around which is incredibly expensive. You've got to visit these sites; you've got to take out racks, you've got to put in power amplifiers and suddenly you find you haven't got enough air conditioning ...subtle but significant details."
The EDGE operator and vendor community remain farily sanguine about such comments. The answer - from Nokia and a number of others - is that many such problems are overcome either because present networks (like AT&T's) are EDGE-capable, or that most legacy networks have already gone some way towards this by being GPRS-capable or, as Northstream's Olofsson says: "In many cases, operators have renewed their installed base anyway and old equipment has been written off long ago. So, for many operators, the additional effort of bringing in EDGE is not that great."
What EDGE means to Nokia in most cases, says Ware, is an EDGE transceiver - which of course supports GPRS and GSM as well. "Then within the base station subsystem it means a software upgrade for the BSC. But the only new hardware that is needed is really the EDGE-capable transceiver and many operators are already installing those." Echoing Olofsson's point, he adds: "The incremental investment is relatively small, taking into account that the operators are anyway evolving their network and purchasing more transceivers for coverage and capacity reasons."
There may be power issues, says Ware, but that's not just an EDGE problem: "When you have high processing power need and high data rate the energy consumption increases independently of the technology."
Leo Nikkari adds, from the point of view of upgrading TDMA operators: "I'm not aware of any significant challenges with EDGE that we didn't already encounter and address with GSM/GPRS deployment. Any discussion of a large number of base stations I think is more related to the frequency band you're deploying. You're obviously going to need more base stations to deploy GSM or EDGE in the 1800 band than you do in 900 - or similarly 1900 versus 850... But I don't think there's anything I know of inherent in EDGE that adds any additional complexity to the physics."
He continues: "Changing out and overlaying TDMA networks, we're very pleasantly surprised that we can share sites, share antennas much more so than we expected. We actually expected much more interference problems, possibly the changes to radio planning - but radio planning is exactly the same, we've made no major changes there and we're quite pleased with progress so far. Where we are in the learning curve is how we best use some of the new capabilities like link adaptation, incremental redundancy, improvements in some of the automation and some of the planning techniques to really optimise these systems."
And issues of power? "We're certainly making it clear to all the vendors that it will be unacceptable to have any EDGE devices that have any penalties in power requirements."
So should there be no regrets for the TDMA community about not going to CDMA2000? In one sense, no, says Comsysi' Elinav: "Because of the economies of scale they chose GSM/GPRS... to enter the worldwide market rather than the North American market."
Also, the data rates viable for first phase EDGE would, if achieved, vie easily with those of CDMA2000, but, as Seybold points out, "Sprint PCS spent a total of $1.4 billion upgrading from cdmaOne to CDMA2000 1X and that includes backhaul. Cingular is spending three billion dollars just to get to GSM/GPRS. Then they have to go to EDGE, then they have to go to UMTS."
Also, he adds: "TDMA as a voice technology is more spectrally efficient than the current flavour of GSM so they actually took a step backwards as far as their voice capacity's concerned." Furthermore, he argues: "I believe when you have a limited resource like spectrum and there's no new spectrum on the horizon, the best thing you can do is take the most spectrally efficient technology and implement it. And there is no doubt that CDMA20001X is much more spectrally efficient than GSM/GPRS."
As for data, he would go beyond 1X. "The bottom line is wherever you mix voice and data there is always an impact. If you're going to use EV data only, then it becomes a totally different ballgame. I'm a proponent of separate voice and data carriers."
But CDMA2000 is not just an alternative for TDMA operators. Beyond that, both Seybold and Joe Nordgaard, founder of consulting firm Spectral Advantage, mention the interest of (unnamed) European operators in tests on GSM1X overlays.
Seybold says: "It uses the same backend as GSM so it's relatively easy to put in. It is more spectrally efficient so in the same amount of spectrum you can have more users and higher data rates than you have for GPRS."
Nordgaard goes further: "The real constraining factor for GSM is the rigid time slots that have served the industry so well. The essence of a GSM 1X offering is to keep all the great things that have made GSM so successful, but use advancements in coding and processing techniques to tune-up GSM's invisible air interface. To GSM customers, GSM 1X is still good old GSM. To the operator it marks the opportunity to transition to a network with greatly improved capacity while at the same time gaining the flexibility to introduce a continued array of capacity improvements and new enhanced services."
But Nikkari suggests that much still remains that can be done to optimise the GSM/EDGE solution: "AMR, improvements and optimisation of techniques for frequency hopping, signal antenna interference cancelling techniques that I think will come to market very soon... I think there's quite a long life for a GSM-based system that focuses very clearly on optimising everything we can get out of a 200KHz GSM circuit-switched carrier for voice and a 200KHz-based optimised packet EDGE-based or EDGE-evolved system for data."
Varrall meanwhile, has another preferred route. He sees no point in using the 'band-aid' of EDGE or GPRS at all, when a more elegant solution - WCDMA - is on the way: He suggests that the fundamental advantage with the WCDMA air interface is that as you increase data rates you occupy more radio bandwidth rather than use more power. This is a more elegant, flexible and power-efficient way to deliver higher data rates from and to a handset.
"You will end up with a much better and more consistent user experience with WCDMA."
But does all this really matter? Transformed by the pivotal events of late 2000, when AT&T chose it, the backing of many major vendors, and the temptation to use kit that often is EDGE-enabled anyway, will the EDGE bandwaggon become a juggernaut? And will the suspicion of Seybold and Varrall that technical issues will arise after deployment hold any weight or is it the case, as Nikkari puts it, that the attention now is on the commercial side not the technical side.
However, despite the optimism of its supporters, one thing is clear: EDGE has not happened yet - as Seybold says: "I guess the real issue with EDGE is you get different answers from different people about what it really is and what its capability is and what it really can do - and until it's out there we really won't know." <<
- Eric - |
