sl> Front Range Tech Biz ("Aerie Networks ricochets into new
sl> market approach")
no, I hadn't heard about this. Thanks!
That was a really odd article. They don't seem to take Ricochet very
seriously.
I think Ricochet is a viable competitor to 3G. 3G comes in two
variants: cdma2000 which is 144Kb/s and WCDMA which is 384Kb/s (both
these figures describe ``shared'' bandwidth, like cable modems).
According to this guy I bumped into, Ricochet was 100-300Kb/s per
customer in practice.Now, look what the European 3G specturm auctions went for. The buz is saying it was catastrophic. At least it's nothing to laugh at! Where's the extra bandwidth that this money should have bought over Ricochet?
I suppose Ricochet isn't taken seriously mostly because there are no Ricochet telephones yet. However, Japan's PHS network is architected very similarly to Ricochet, and several competing PHS ``virtual carriers'' offer both telephones and data cards. For this and other more technical reasons, I don't think that whether or not they offer phones will end up being the most interesting difference between 3G and Ricochet.
The big dispute between 3G and Ricochet's network designers is between 120-degree towers (those triangular arrangements that cellular uses) and omnidirectional towers (Ricochet's ``pole-top repeaters'' and ``Wired Access Points'').
| 120-degree | omnidirectional |
|---|---|
uses:
|
uses:
|
| big towers. high tower elevation. expensive cel sites---expensive equipment, expensive cables, large footprint and expensive rent, large power consumption, competition for good ``sites'', eyesore complaints, luddite complaints about mysterious ``radiation'' effects. | small, cheap towers---costs and sizes on the same order of magnitude as user devices. |
| towers have more transmit power and more receive sensitivity than devices. | towers and user-devices have similar or equal power output and sensitivity. |
| devices declare affinity to a single tower. | devices use multiple towers at the same time. |
| 120-degree arrangement creates a honeycomb of ``cels''. The towers cooperate with each other to move phones in cars through cels. The physical arrangement of towers can't be arbitrary--- it has to match the arrangement. works great at high speed. | omnidirectional antennas preclude ``cels''. There are problems with fast-moving devices. PHS and first-generation Ricochet had problems at 25mi/hr or less. Current second-generation Ricochet is supposed to work at 70mi/hr. There's no need to plan the spatial arrangement of sites. |
|
coverage pattern reflects its American AMPS-standard ancestry, and the
American preoccupation with holocaust and emergency. Take your 3-watt
phone in a covered wagon on a horse-drawn trip to ``the west'', and
you will get good coverage. Set foot inside a cement mall with some
antennaless piece-of-shit Nokia, and suddenly you don't get calls
anymore. What a shocker! The ``coverage'' as plotted on a map looks
good and has rounded edges, because each tower throws a large pattern,
but it is very expensive to put up enough towers to eliminate ``cones
of silence'' or ``dead spots'' or ``shadows'' or whatever you want to
call that thing where you move a celfone two feet to the west and it
suddenly starts working again.
|
Works great in urban environments with dense population. It can work
indoors simply by placing a site indoors, because the size, power, and
cost of sites is about the same as cordless phone base stations. Works poorly in farms with long government-subsidized telephone and electricity runs. The outer edges of the coverage area are very irregular because the sites each have a small pattern, but it is cheap to tune the coverage pattern toward perfecton. When I had first-generation Ricochet, many subscribers called Metricom to complain about poor coverage at their house, and ended up with a site on their roof. Ricochet's Bay Area coverage was far better than AMPS cellular, back in 1996 when they were just getting started. I had the same experience in Pennsylvania a few months before their network went dark: they had better coverage than a Bell Atlantic CDMA phone. However, I didn't get around as much in Pennsylvania.
|
The point is that 3G is needlessly expensive. I do not think 3G can afford Ricochet/PHS prices of unlimited use of the shared 384Kb/s channel for $70/month. That is certainly not the type of pricing structure that carriers are discussing so far.
Ricochet can scale w.r.t. traffic or population density better than 3G, because short-range Ricochet ``towers'' can go up more quickly and in a more tightly-controlled pattern than 3G towers.
Early Ricochet's coverage was better than early cellular's coverage. Ricochet can affordably deliver coverage that is more consistent than what 3G can afforably deliver.
Cellular's architecture is fundamentally based on the 120-degree tower. The 120-degree tower is motivated by limitations that no longer exist, such as:
New networks want more bandwidth, either because of as-yet-unspecified broadband-wireless ``applications'', or because there are a lot of talkative subscribers in a small area. However, the cost of spectrum hasn't changed. This means cel sizes should shrink.
3G is a case-in-point. When 3G actually meets its bandwidth promises, its cels shrink, presumably until they end up being the size of low-power Ricochet mesh-patterns. However, the 3G equipment doesn't get much cheaper at small cel sizes.
Another thing which hasn't changed is the wavelength and the properties of radio. Refraction, shadows, and multipath around natural and urban geography hasn't changed. When cel sizes are large, we can pretend these things don't exist and draw pictures of elegant honeycomb cel-patterns spreading across a perfectly flat Donald-Duck-in-Mathemagic-Land pink rectangle representing the infinite plane. In a flat desert, one can always see the sun. When cel sizes are small, we realize that cities are three-dimensional. We realize that often we can't see the sun. We can't stick cel sites just any old place we want to. The whole cellular metaphor breaks down at small cel sizes. If modern signalling systems continue working with tiny cels, with little mock-tower 0.5ft^2 flat white designer 120-degree antennas inside malls, it just proves that they never needed cels to begin with! But does that mean Ericsson will sell companies indoor 3G infrastructure for the price of Ricochet repeaters? I rather expect their reasoning will be ``infrastructure price is proportional to number of subscribers served,'' mostly-independent of how far apart that group of subscribers is spread.
The Ricochet network is designed for small cels. With small cels, transmitters don't need to be powerful but do need to be cheap. Small cels make directional antennas silly because the cellular metaphor breaks down when your antenna is situated in a hallway: signal strength and direction don't change slowly or with a predictable pattern based on subscriber location, so they aren't useful for a cellular-style routing regime based on modelling each subscriber's physical position.
Directional cel sites, compared to omnidirectional technology, are:
Cel sites inside malls and on different floors of buildings is something I saw with the GSM networks ``Smart'' and ``Globe'' in the Philippines.
To get information about Ricochet, you need to use the Wayback machine at archive.org. Look for the oldest pages possible, so as to avoid the goofy management that took over, changed to that dorky footprint-logo, made service twice as expensive and horribly confusing to sign up for, and ran the company into the dirt. After their website redesign, most of the actual information disappeared. Here are a couple good pages:
A Ricochet repeater in the wild, on Warren Road in Malvern, PA.