Well, looks like I feel like pulling out my soapbox again. This note is not as good as the first one, but here goes.
Direct Sequence Spread Spectrum - IS-95 CDMA
The new cellular system standard for "2-G", that is "Second Generation", is IS-95 CDMA. First Generation was clear NBFM analog voice, something very close to the VHF/UHF FM in common use on the hambands. It was FDMA, sort of. It was closer to trunking (see below). Now, in order for the operators to have about twice the capacity, DS-CDMA has been introduced. What is known as "3-G" is planned to be DS-WCDMA, "wideband CDMA", at least in this country and in Japan. What will really be in 3-G in this country is still undetermined.
DS-CDMA
Both 2-G and 3-G CDMA use direct-sequence spreading. They both have the same problem: they create a wideband rise in the noise floor. This is not due to some inadequacy in the particular implementation, but is a characteristic of direct-sequence spreading. 2-G is about 1.5 MHz wide, while 3-G is proposed as 4.5 MHz wide. In either case, they both rely on the operator having TOTAL control over all emissions in the 1.5 or 4.5 MHz channel. The system is VERY susceptible to interference from uncontrolled carriers. As for causing interference to other more narrow-band receivers, their noise floor will rise substantially for MILES around the DS transmitter's antenna. That's one reason why IS-95 CDMA has only been deployed in the U.S. and in the HiTACS band in Japan. It was proposed for another band in Japan, but tests showed that the band was populated by "pirates", rogue mobiles, not unlike the "HFers" between CB and the 10 meter amateur band. They couldn't guarantee to get rid of ALL of them, so they had to CANCEL the deployment of CDMA in that band. I would expect the same kind of issues with any deployment of 3-G CDMA.
Moral 1: Direct sequence spread spectrum is usually wider than a whole HF band. It could really only be useful on VHF or UHF.
Moral 2: Direct-sequence spread spectrum and conventional narrow-band modes (CW, SSB, NBFM) are NOT compatible, even on VHF and UHF. The DS station requires total control of his wide channel.
Land Mobile Service and the Amateur Service on VHF
It seems that traditional Land Mobile users are closer to typical ham radio users, at least in their needs and style. Traditional Land Mobile users are dispatch services, like police, fire, taxi, etc. They could be simplex or use a repeater. The transmitter is usually NBFM, and the channels are either dedicated to one user or are "community repeaters", used by several users, each with a small traffic load. The repeaters have a fairly large coverage area. The adjacent chanels, maybe 25 or 30 kHz away, are protected against causing interference by splatter filters on the transmitters and from hearing interference by narrow IF filters in the receivers.
Sometimes, "trunking" is allowed. Here several channels are "trunked" together to form a pool of channels. A mobile finds, or is assigned, an available channel from this pool. The people he wants to talk to, his "fleet", also moves to this channel. By the way, hame do this on VHF when they move to another repeater to carry on their conversations when their usual machine is occupied.
It is not hard to see that the need to protect the next channel for another user, not knowing who will have the next channel or where he'll be located, or how much power he will have, is the same concern we have on the VHF ham bands. It is usually addressed by good radio design and frequency coordination.
DS Spread Spectrum on Satellites
Well here we have some possibilities. Satellites are up some 100 miles or more. They run low transmitter power and low ERP, usually just enough to make the path. They are closest when directly overhead and farther away when at the horizon. They have Doppler shift, but it's predictable. This might be just the application for direct sequence spreading.
A properly designed satellite transmitter for a 1 MHz wide spread signal of 10 kbaud at 430 MHz would have to contend with at least 130 dB of path loss. The system would still want a 10 dB S/N at the receiver, or a sensitivity of -120 dBm or so. The transmitter needs to put out only +10 dBm. That means that a typical NBFM receiver only picks up -140 dBm ... a barely perceptible rise in the noise floor. CW and SSB filters would allow even less to get through. Yes, here the terrestrial narrowband systems would not be effected by a DS satellite transmitter.
But, the other way around ... the ground-based satellite receivers would probably be bothered by terrestrial transmitters. Terrestrial transmitters run high power (100 watts typical) into high gain antennas (10 dB typical) and are much closer (10 miles typical). The ground-based satellite receivers would have to contend with a -60 dBm or so interfering signal and can be interfered with ANYWHERE in their 1 MHz receiver bandwidth. Even the 20 dB of processing gain to the desired signal and 20 dB rejection to an interferer would not be anywhere near enough. They would be blasted.
Moral: Direct sequence spread spectrum requires control of ALL emissions in its channel.
Other Spreading Schemes
As you may be aware, spreading a carrier with a high-speed sequence through a DSB modulator (DS-SS) is only one way to do the job. There are slow frequency hopping, fast frequency hopping, and time-slot hopping methods for creating a spread spectrum signal. Maybe the best is slow frequency hopping. Here the information signal is transmitted for maybe a few milliseconds on one channel before the transmitter hops to a new channel. The channel hopping is in a pseudo-random sequence. The receiver must hop to the new channel in step with the transmitter, but it knows the channel-hopping sequence too, so that is not a problem. The advantage of this scheme is the ability to hop over interferers or chunks of spectrum used for something else, with only a loss in how many "hoppers" can be allowed. The disadvantage is collisions with other hoppers will result in lost information, so maybe only using enough power to make the path is a good idea to minimize interference.
Other Direct Sequence Spread Spectrum Users
If we were to set aside a chunk of spectrum, say 1.5 MHz, for DS-SS, we might bypass the problems of co-existing with narrow modes. But what about other DS-SS operators? Will they see interference? Well, maybe ...
Let's just consider two links on the same channel, Link 1 and Link 2. A receiver will hear both the desired and the undesired transmitter signal. It will treat the undesired signal as noise, and reject it by 20 dB or so. With an undesired signal at the same level as a desired signal, we have no problem. The undesired signal is gone. But what if the undesired signal is 3 S-units louder ... about 20 dB. Now the interfernce is still there. To minimize interference, the receiver will have to control the power of the transmitter at the other end, trying to get the desired transmitter to put out more power. If it can, then the receiver has no interference. But now the transmitter is louder and can cause more interference to the other link. Then that receiver may need a louder signal, telling the other end to turn up, increasing interfernce to the other link, which has to turn up .... Well, you could get kind of a runaway situation. Running full duplex would probably help ... for two links. But for three or more links, there is little hope. Only increased propagation loss ... distance ... can help.
Internet Access ?
I will tell you right now that this is not my area of expertise. But ...
If the spreading sequence is 100 times the data rate, then 2-G systems can transfer data at about 10 kilobits/s to 15 users, or 150 kb/s total, and 3-G systems can run at 10 kb/s to 50 users, or 0.5 Mb/s or 500 kbytes/sec ... on the average. Unless one can live with 9600 bps transfer rate (when even home users are used to 40000 bps or more), a user has to borrow time, power, etc. from someone else. The key is to find a way to handle these higher transfer rates for a second or two. One can see why this is possible on a cellular telephone system (which has the capacity for a large number of users), but may not be practical on the ham bands.
Recommendation
Well, it looks like I can't find a good way to have direct sequence spread spectrum or IS -95 style CDMA co-exist with narrow modes, except by frequency coordination. Even then spread spectrum may not be able to co-exist with itself. Since frequency coordination may not be the direction amateur radio will go, I don't think I see a place for spread spectrum, except as a developmental system or as a curiosity. Now I am all for the diversity of modes of communication in ham radio, everything from CW to fast-scan television. (I've run both.) It's just widespread (bad pun) use of DS-SS may not be desirable. Large open frequency bands with narrow antennas (microwave? optical?) may be just the thing.