Post of the Day
June 21, 2000
Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!
I posted this on some of the other boards and thought it would be of interest here.
MCOM is a company named Metricom that is owned ~45% by Paul Allen and ~35% by MCI WCOM. They are quietly, almost to the point of secrecy, building a nationwide network for pure data that they call MCDN which stands for Micro-cellular Data Network. The service is called Ricochet.
Many people think that the future of wireless data will be via the cell phone based networks, but talk to any RF engineer worth his salt and you'll find that a network based on voice does not do data with the speed and capacity that is needed.
The reasoning is simply the fundamental architectures of the networks. Voice based networks, whether CDMA, TDMA, iDEN or GSM, were designed to use a clean but narrow spectrum to carry voice transmissions which stream at ~8kbps. Despite voice transmissions using little bandwidth, voice networks can still barely handle these 8kbps connections because there is not enough capacity to carry all these transmissions. This is usually the cause for when your phone does not work at peak hours.
Now comes along data. Data is even more bandwidth intensive, much more. For every ~10 kbps of bandwidth that a carrier allocates to a user, he will displace 1 voice customer. In other words, if a carrier decides to give 128kbps data service to a user, the carrier has to bump 13 voice customers. What will he have to charge that data customer when those voice customers are willing to pay 25 cents a minute? Here is the opportunity cost in the shared bandwidth for voice and data.
If you are transmitting low level data like text for WAP phones or low quality video, bandwidth is not as big of a concern although it certainly still has the ability to overload a network (just look at NTT's iMode). But say you want to use the wireless connection for your laptop so you can browse those graphics laden sites, download MP3s, download email attachments, view decent quality streaming video, etc.... You will have to pay very high fees for using all that bandwidth on the voice based network, since carriers will not price it cheaply given the opportunity costs mentioned above. Of course the carriers can make it affordable by giving you slower speeds where you will tie up less resources. This is how CDPD(Omnisky, PocketNet) is made affordable, by providing slow 14.4kbps connections. If you can stomach waiting for pages to download at these speeds you can save yourself a few bucks.
These same problems of voice and data fighting for bandwidth will still be present even with the improvements made by 3G. At the next 3G spectrum auction (Oct '00 for the US), carriers will have to pay billions (UK netted $36B overall) to buy spectrum in paltry 5MHz blocks in order to add capacity. That is only ~17% increase in spectrum for some carriers. Increases in spectral efficiency with 3G equipment will roughly allow for a doubling of voice capacity but much of this capacity is already spoken for by the overselling of the service by the carriers. The rapid growth of future voice subscribers will also claim more of this capacity. Irregardless, if you look at the white papers for WCDMA(which is the 3G standard pushed by ERICY, NTT, NOK, and others), you can see the capacity concerns where it discloses that a triple base station WCDMA system that can carry 80 voice connections, will only be able to manage 3 data users at 384kbps.
Does anybody think that a carrier will shut out 80 of their voice customers so they can serve 3 data customers at 384kbps? Absolutely not and that is why data rates for 2.5G are already coming in below what was promised. GPRS was supposed to do 144kbps. The new figure from NOK is more along the lines of 20kbps.
Once again, the reason the speeds are slow is because the carrier wants to limit the bandwidth available to the data customer so they won't hog all the spectrum. Even QCOM admitted at the last PCIA conference that their CDMA2000 1X which also was promised to do 144kbps would only do ~40kbps in actual use.
This is where MCOM comes in. MCOM's MCDN network will serve data traffic SEPARATELY from the voice networks so there is no trade-off between the two. They have positioned themselves as a wholesaler so as to make potential partners out of Sprint PCS, AWE, AOL, NXTL(& others) instead of competitors. They will announce their reselling partners in a few weeks. Their Ricochet service will also light up in a few weeks at 128kbps(faster than GPRS, or 1X) and next year, they will have new modems for speeds up to 384kbps.
How does MCOM create capacity when the others cannot? First, they use unlicensed spectrums. Now with voice networks, you can't do this because the noise and interference of unlicensed spectrums may cause intermittent static, which would be suicide for a business catered towards sound clarity. With data however, any corrupt transmissions can simply be resent and it may never even be noticed. MCOM also has developed technology that makes its networks extremely robust and resilient to interference (they have been doing wireless IP since the early 90s before it was even talked about in the media). The advantage of unlicensed spectrums is firstly cost. MCOM will build out an entire nationwide network in 46 markets for ~$2B - which is less than the $3.3B that ATT paid yesterday for just Houston, San Diego, and San Francisco! The other advantage is capacity. By using unlicensed spectrums, MCOM will utilize 139MHz to transmit data. That is a huge pipe. The largest spectrums that voice networks have are 30MHz wide. ATT and Sprint have 30MHz of spectrum while NXTL and others have much less.
Another way that MCOM creates more capacity is by using a microcellular network. One fundamental of RF design is that the smaller you make your cells (areas covered by a tower) the greater your capacity will be. This is because larger cells aggregate more traffic creating a bottleneck as well as more complicated reasons such as frequency reuse. With MCOM's MCDN, the cells are ~1/2 mile in radius. PCS systems are much larger at ~5miles in radius (depending on topology of course). That is a huge advantage for MCOM considering this factor is squared when measured area wise. The disadvantage for MCOM is that they have to put up more radios but that's where they have created low cost (~$1,500) transceivers that can be hung on light poles instead of setting up a $400,000 PCS tower.
This is not to say that MCOM will cut into QCOM's or NOK's or ERICY's profits. The 3G players are going to sell their equipment regardless since networks will need 3G to increase voice capacity and to provide WAP features at higher speeds (albeit on the same small screens), but I think the real direction of wireless internet is to retain the graphics, java, mp3s, video, and all other goodies that desktop users have been accustomed to. In order to provide these bandwidth hungry features at an affordable price and high speeds, the capacity has to be there and that's what MCOM's MCDN is all about.
To lay the speeds on a timeline, MCOM's MCDN will provide 128kbps this year and 384 kbps next year. The fastest of any mobile IP network. A plus will be it will be less expensive also. For 2.5 � 3G networks, GPRS and 1X will bring speeds up to ~40kbps until ~2003 where EDGE, WCDMA, and/or 3X promise to match MCOM's then 384kbps speeds but given that GPRS and 1X figures are coming in ~30% of promised rates, the actual data rates in practice are highly suspect. If anyone is thinking that 3G will do 2Mbps, they may want to take a close look at the 3G framework called IMT-2000. The 2Mbps figure is only given for indoor situations. This would only occur if a 3G carrier decides to spend the large sums of money to set up a local base station inside a specified building in spite of the limited number of potential customers. The original idea was to have 3G act as a LAN within buildings but that was before the availability of 802.11 LANs and before the skyrocketing of spectrum prices. This idea for 2Mbps speeds is now admitted by 3G proponents to be unlikely for both economic and logistical reasons.
Concerning the prospects for HDR, the package appears to be a misleading twist of facts from the information that I have found. Raw data rates are quoted instead of actual throughput, burst speeds are given instead of sustained data rates, the question of capacity is avoided. The simple fact is any company can create a wireless device that transmits at high speeds. It's as close as the nearest CompUSA where you can buy such technology in the form of wireless LANs, but developing a system that provides the speeds to many users (mobile ones at that), in city wide applications where factors such as multipath and interference work to reduce speeds - that is the real challenge. If HDR was as effective as touted, both WCDMA and CDMA2000 would boast far greater data rates considering they are all based on the same underlying technology - but they don't despite the 3G systems (WCDMA and CDMA2000) being able to utilize 24X more spectrum (or more). If this is not evidence enough, look at the simple fact that no US carrier has chosen to implement it. US West was the only ones to test it and they decided not to use it. Sprint did not even care to test it saying it was not of value to them in its present form. The only places where HDR is showing up is Korea and Japan, both countries where users are accustomed to high per minute/hour charges for internet access.
Either way, 3G will be exciting with the added features on phones and uses for laptop where metered pricing is not a concern, but even more exciting may be a laptops or webpads with an MCOM modem. The service will be affordable (~$60/month) and will allow users to surf wirelessly at 128kbps with unlimited usage.
Excerpt from Lehman Brothers:
"The rates talked about for both 2.5G and 3G for broadband mobile networks are for raw data rates, which are often accompanied by the adjective burst, as in, not sustained. (The other thing that is becoming clear about 3G is that it is an incredible spectrum hog. The amount of voice capacity that will need to be sacrificed to offer 3G is daunting. The opportunity cost of taking voice capacity out of commission to offer data means that it will be very unlikely that the name carriers will be able to price high-tier laptop connectivity at a flat rate. We think this could create significant resale potential for Metricom.) Metricom's raw data rate is pushing 500 Kbps, but management believes it is misleading to promote raw rates when users will not experience that. Current beta users of Ricochet2 are regularly running speed north of 128 Kbps."
FD: I am not associated with Metricom in any way other than as a shareholder
Read More Posts by This Author
Go To This Post
More Recommended Posts