It is one of the more pleasant task as moderator to consolidate a couple of excellent posts for future reference.
Thanks to propitious7 for great questions and special thanks to engineer for the answers. Suspecting these questions will be asked over and over again during the coming months, this is just a post to put the questions and answers together.
From propitious7: Who can help me understand the rebuttal of IJ's points; in general, who can help fill in a report card on wCDMA. A lot of smart people are working on it and a lot of carriers are committing capital, market position and reputation to build wCDMA so what are the respects in which it is -- or will be when evolved -- be superior to cdma2000??
Can we grade wCDMA on the following:
propitious7 Data rate -- Will wCDMA meet or beat the 2.4mbps for do?
engineer Data rate. No, WCDMa will not meet the DO data rate. that comes later in EDGE and UMTS systems.
propitious7 Capacity increase -- Will wCDMA permit a 2x or greater voice capacity increase? Does wCDMA have capacity enhancements comarable to the promise bruited by IJ to 2x capacity again with dual antennas and other H.Potter magic?
engineer WCDMA will give a capacity increase over GSM, the system which it replaces, but it will be less capacity than CDMA2000 or 1x. WCDMA can benifit from dual antennas and half rate voice coders just like CDMA2000 and 1x, but any improvement for any form of CDMA can be lic and used for the other form of CDMA.. the issue is that FIRST they have to get WCDMA out there amd make sure that the standard that they have is stable and then, and only then, can they start making it improved. that is part of the current problem, in that the standard is not yet completed, so that they cannot completely implement it yet.
propitious7 Throughput Cost -- Q cost projects data throughput cost per 1M bits at two cents for do and seven cents for wCDMA; Is this eyewash? What's the counter-position. I never read about throughput/cost in wCDMA materials
engineer Will Q's projections hold up? do the math. If the carrier can provide the channel at 2 cents a minute, then if it does an average rate of 1M bps (real data rate I have observed and used), then 1M costs 2 cents. due to other factors listed below, WCDMA will cost more to implement and this will drive the carriers ability to sell air time minutes for less than 5 cents. WCDMA is projected to be less than 1 MBPS. Again, do the math.
propitious7 Power Usage -- We read about great computation demands on w and much higher power usage? So? or NO.
engineer GPRS has higher power demands, have not seen anything on WCDMA higher power demands, except in the handoff case I will cover below. WCDMA has higher DSP computation needs than GSM or GPRS, since it is using higher coding gain type modulation. This is true for any of hte newer standards where they are adding procssing to increase the coding gain. the same argument was made in 1993 by Q thread debates on IS-95 that it could nto be made economically because we had about 3x the amount of circuitry over US TDMA. the thing they forgot was that the technology of ASICs came down faster than the time rollout of the standard and by the time the standard was rolled out, the technology caught up and the cost was close to the same. This was by design.
propitious7 Drop Rate -- We read about handoff problmes with wCDMA; is this just normal evolution of a new system or is there a congenital problem with synchronous operation?
engineer Drop rate - asynch handoff problems were known around Q a couple of years ago and only now is coming to light. the issue is that in conventional IS-95 or 1x the timing of the pilot channel allows the phone to "know" where the handoff will happen in time and acquire the signal fast, no matter what the signal level has dropped to. the voice channel is run at about 7 dB EB/No and the pilot timing channel at about 10-14dB. This increase allows the timing chains in the phone to find the new handoff channel in a very fast, very low fading event, which is exactly what happens at handoff. In the case of the WCDMA handset, they not only have an asynch channel which must find timing between basestations, but they have to do it at exactly teh wrong time. When the signal starts to fade and the cell indicates handoff, the phone must go find a new signal to lock onto to handoff to. but since you may be going directly away from the old cell you were hooked up to the signal fades very fast. The new cell may not be coming up as fast as you need, so you may enter a 2-5 second fade hole. the WCDMA standard has a timing filter which has a training time, but this training time is based on the signal to noise ratio of the lowest fading channel level. What happens is that as the signal fades, the ability of this filter to find timing goes up exponentially and the problem becomes harder for it to solve. the three solutions to this are to add more cell towers in the middle, or add a timing channel over the top, or to add alot more processing power to the timing filter, which may not be realizable in todays technology, but perhaps will be available in the newer 0.1 or 0.08 technolgy available in 2003.
the increase in cell towers is not a factor of two, it is a geometrically driven problem. Imagine 16 circles where the edges just touch. each one of the "holes in the middle" represents another tower that is needed. For a multiple tower system, it is not just 4 new towers, but more like 5 new ones to take into account mountains, valleys, buildings, etc. This accounts for the increased cost out there.
propitious7 Propagation -- IJ and others have said that dynamic range problems with synchronous can be overcome but only by increasing base stations; I have read three times more per area of service. Is this true, false or just temporary?
engineer Same issue as drop rate. If you do not handoff a call wihtin 4-6 seconds, it is dropped. You can also turn up the power of the handset or cell tower during handoff to try ot give more signal, but this proves to be a dynamic nightmare. It is a true fact though that during a low level handoff, the reverse channel transmits higher power due to dynamic power control issues. i.e as the rec channel is perceived to be weaker, then the transmit power gets stronger to insure that the handset can reach teh cell tower.
propitious7 Bandwidth Requirement -- Eric writes that minimum for cdma2000 is 2.5mhz (Why is that, Eric, I thought it was 1.25mhz; is it that two bands are minimum?) and 10mhz for wCDMA ( I thought it was 5mhz -- again why double). Of course, to get comparable performance you have to compare 10mhz to 10mhz; but is there an advantage to using wider band or to using narrower band apart from obvious advantage that wider bandwidth may not be available until new spectrum is allocated for mobile use.
engineer Eric is wrong. It is 1.2288 Mhz spreading for CDMA2000 hence the 1.25 Mhz bandwidth and 3.884 Mhz spreading for WCDMA, hence with much more sloppy guardbands, 5 Mhz. Notice that the useage effecieny is also much more poor than CDMA2000, but they save on filters in the handset. BTW - this benifit over IS-95 is negated when teh ZIF technology comes around, as 1x and IS-95 can do this in a simpler way.
propitious7 Async vs. Sync -- Is there some performance advantage to asynchronous service apart from not having to depend on U.S. provided GPS for time signal? When wCDMA service is normalized and optimized will there be advantages to async service? Should we assume that if EC does build a Euro GPS, then wCDMA service will be converted to sync timed to EuroGPS?
engineer synch versus asynch was an issue spawned by the WCDMA guys to try to get China and europe to use WCDMA because the CDMA200 used a "US Military controlled system" component in it. this is why Europe is now proposing a new GPS system. they sold the idea too well and for them to go to synch now would mean eating those words. Heard that alot while working in Korea also when the new spectrum wars were going on for 3G. trouble is that if they shut off cell phones, most airliners would get lost......they both use GPS.
propitious7 do for wCDMA. The advantages of data only service, segregating voice and data, seem so obvious why is this not adapted to wCDMA to provide flexibility and upgrade data rate?
engineer WCDMA V+D or 1xev-do? Matter of choice. the DO channel is MUCH more effcient than the voice channel at handling data. there is no equivalent in WCDMA yet. They are thinking about it. Here is the case. If your a 1x/HDR carrier, then you can have teh CSM modulate 7 channels of 1.25 voice and 1 channel of HDR during the 5:00 rush hour. then at 7:30 when everyone is home and wants to surf, then put in 5 HDR channels and 3 voice. Change the modulation to match the channel loading. In WCDMA, they only use the voice channels and you really have no float gate except denial of service during high peak hours. Again, it is a matter of choice, but the HDR system seems more logical to me.
the other two issues not mentioned by you are the lack of a solid ratified standard yet on WCDMA and the maturity of the CDMA2000 as a standard. Also not mentioned is the direct fallback of components in manufacturing so that volume discounts on components is possible. In 2003 perhaps we will see the volumes that are needed to drive WCDMa down. but the fight now is for 100 little WCDMA vendors out there who all have a peice of the pie. they will grow volumes slower than one or a few sources would. In the CDMA2000 case, teh radio and modulation of hte channel is basically the same with the digital ASIc being the differnce. Even for HDR, the radio remains the same. this means that all the RF experience that IS-95 has gained will be directly carried over to CDMA2000. In WCDMA you have the same guys who have not been able to perfect CDMA or IS-95 components trying to come up with WCDMA ones. If Nokia (who I started helping in 1992) couldn't get IS-95 right by 2000, then how will they get WCDMA right by 2003? Perhaps they will buy Q chipsets to get it working first....<ggg....sorry could not resist the jab..> Same for MOT, Same for Samsung. VLSI?, Philips?, DSPC? the WCDMA group has about 20 or so chip companies that I know of out there trying to build an ASIC to capture a part of this market.
Last point. I hold to this as well. WCDMA can work, but it is different than the working parameters of CDMA2000. If the operators want it to be teh case, then it will probably come out. It is just sad that on some of these cases, they took a pretty hard lined approach for marketing reasons that seem to have come back to haunt them. I think IMJ wants it to work so that we can all get on with life and make bigger roaming agreements and grow the pie. get apps going and let the wireless data revolution get flying. right now we are not arguing about the car race, but the tires on the car. |
