3G, HDR, and 1X
Funny that a QCOM fan should call MCOM dishonest:) First of all, there are several different ways to deceptively quote your data rates so that they appear faster. QCOM has a tendency to use all of them. Here are a few:
QUOTE RAW DATA: Not all of the data that is transmitted to you from the tower is the data you are requesting from the internet. Much of it is necessary additional data that the wireless network attaches to your packets in the form of added protocols, bits used for error correction, etc... In short, you may be receiving 144kbps from a CDMA tower, but if you were to do a bandwidth test, you may find that you are only receiving 56kbps of actual internet data, which is all you care about.
QUOTE BURST SPEEDS: There are time gaps between your packets that when accounted for, will dilute your sustained data rates. By quoting burst speeds, you can pretend that those periods of DEAD TIME, where no data is passed, don’t exist. In the real world, quoting burst speeds inflate your data rates by 15% or more..
QUOTE THEORETICAL SPEEDS: Theoretical numbers mean nothing because the ideal conditions that they assume never exist. Theoretically, a 1.25MHz CDMA channel can support 64 voice users. In reality, it often tops out at 20 users. Limitations of the electronics as well as adverse RF conditions stemming from multipath, a noise floor, and fading will drag down your numbers.
QUOTE SPEEDS ON AN UNLOADED NETWORK: There is only so much bandwidth in a channel and this bandwidth has to be shared. Many people think that because the maximum data rates are much higher with 1X and 3G compared to current day CDMA networks (IS-95, also called cdmaOne), that somehow more bandwidth is created with 1X and 3G. In actuality, the main reason why you can get higher data rates with 1X and 3G than with cdmaOne is NOT because more bandwidth has been created but because 1X and 3G are able to “funnel” more of the channel’s bandwidth to one user. QCOM tries to insinuate otherwise by saying that there should be a 50-100% increase in voice capacity with 1X but this is pretty sketchy based on their reasoning and even the CTO of AT&T Wireless has questioned this. I think Moore’s Law has spoiled people into thinking every technology will increase in performance exponentially. Unfortunately, increasing spectral efficiency is much harder than increasing the number of transistors on a computer chip.
To elaborate, current CDMA and the future CDMA 1X networks divide their spectrum into channels that are 1.25MHz wide. A CDMA channel this wide gives you roughly ~250kbps of total bandwidth in the channel. With current day cdmaOne, you can have a bunch of (~20) voice users or a bunch of data users at 8-14kbps. With 1X, which theoretically can give you 144kbps, you still have the same 250kbps total bandwidth in the channel (perhaps a little more in some circumstances where many users are closer to the towers), but unlike cdmaOne, you have the option of variably breaking up the bandwidth allocation to a bunch of voice users ( a bunch because they only take up ~8-13kbps) or you can cut a bigger piece of the 250kbps pie to a FEW high speed data users and any combination in between. The same is true for 3G, but there is much more bandwidth in the channel so you have higher maximum data rates (384kbps) but you are also using more spectrum because 3G channels are also much wider (3.75MHz or 5MHz) so does this really buying you anything? Much of the time, the answer is no. When you are the only person in the channel, you can hog it all and get pretty good data rates and take advantage of those maximum data rates. This is the scenario they like to use for their quotes. In the real world, voice users are using most of this bandwidth most of the day. That leaves a just fraction of the total bandwidth in the channel for data users and with multiple data users, they have to divide this left over bandwidth among themselves even FURTHER resulting in each data user getting perhaps 40kbps much of the time. This number will of course fluctuate greatly but consider that in most major cities, the spectrum is saturated at peak times (noon, rush hours, etc..). Here, you normally cannot even establish an 8kbps voice connection with your phone. Doing data, you may not get any data packets through at all and if you do, the data rates will be extremely low (below 10kbps) and the latencies very high (possibly several seconds). Also consider that if you are a working professional, these may be the times you need your wireless connection the most.
Unlike cdmaOne, 1X and 3G, HDR is data only. By being designed purely for data, it can achieve a higher total bandwidth in a channel than cdmaOne, or 1X by transmitting full power to a single user. By going full power to one user, it allows for higher modulation schemes (up to 16QAM), which results in more total bandwidth. Unfortunately though, this sacrifices the ability of more than one user being able to communicate with the tower at any instant so now, users have to take turns talking to the tower like in TDMA based networks. This is because using CDMA for multiple access requires the power to be distributed among all users in the channel. Soooo… although QCOM increases the maximum bandwidth within the channel with HDR compared to 1X or cdmaOne, it has to queue users in a round robin fashion and this means higher latencies as well as drastically reduced data rates when more users are added. According to QCOM, a SINGLE user on an HDR channel can get ~600kbps. Because all the bandwidth is concentrated on this one user, this 600kbps represents the total bandwidth in that channel. If the HDR user employs antenna diversity (modem uses 2 antennas and processes the signal to mitigate adverse RF conditions and boost signal strength), that single user can get ~700kbps. The Adventis white paper suggests this figure is closer to 200kbps under real world situations for HDR and MCOM suggested ~150kbps at the shareholder meeting. Either way, the white paper used the 600kbps figure and since this bandwidth is divided between all HDR users in the same channel, they arrived at the 33kbps figure for 15-20 users in the channel, which sounds reasonable IMHO.
Even better than all the QCOM spin tactics mentioned above, perhaps my favorite trick is the one that QCOM fans like to use where they produce links to papers and articles and scream, “This is the reason why CDMA/HDR will kill Ricochet!” Yet, they have no idea what the link is saying;) Caxton, all that slide shows is the trade off in voice capacity and data bandwidth when you are dividing up your spectrum between mix matched technologies. If anything, it shows how much your data service will suffer when you make voice your #1 priority as every carrier has clearly stated they will do. And why wouldn’t they? They can sell voice packaged at $2/MB while in the case of data, even Jacobs has said the price points need to be $0.30/MB or less. Ricochet service is flat rate and some of it’s users pass 500MB of data per month. That price per MB for R2 is $0.14/MB.
About when HDR will appear, it may be never as far as the US is concerned. US West tested it but rejected it. Sprint and Verizon are TRIALLING it but have yet to adopt it. Nokia rejected it after testing it in Spain. Even Seybold has said it may be questionable that US carriers will deploy it anytime soon. It really is not hard to see why carriers have not yet adopted HDR if you think about it. Let’s pretend we are Sprint. We own 30MHz of spectrum, which cost us $2.6B. 30MHz is enough for us to afford 24 channels 1.25MHz wide. Even with 24 channels that can support up to 20 users each, our customer service gets complaints constantly because of busy signals and drops due to overloaded cells. We simply do not have enough capacity. Can we afford to give up even 1 channel for HDR? Especially when more and more voice users are added every day? Does it make financial sense when we can fit 20 voice users in that channel and charge them 20 cents a minute but the HDR channel can only support a few data users at reasonable speeds and these data users want flat rate pricing? What about the costs we have to pay for the hardware and to reengineer the cells and networks? Despite all the hype and marketing, HDR has yet to even get off the ground. Another question I have is how many channels will a carrier really have to give up to implement HDR? Since it cannot have the same users on the same channel at the same time, will it require carriers to give up 7 channels to provide just 1 HDR channel per cell (assuming N=7 frequency reuse scheme)?
Bottom line, MCOM’s quote of 128kbps uses none of the “cheats” mentioned above and the latencies are comparable to that of DSL or cable and far better than the latencies you would see with dial up, 3G, HDR, or any other voice based technologies. Instead of “killing” Ricochet, I’d bet that 3G and HDR will bring it more respect when people see how poorly the competing technologies really perform when/if they show up. |
