2024-12-13 — What's New at DLARC — 2024-12, Explaining the Use Case for Data Over Repeater - Part 3 - SuperPeater!, Recs for Portable / Mobile 70cm / 23cm DVB-T [Video] Station, Raspberry Pi 500
I have use chatGPT for humorous graphics for our club newsletter, it does show a good sense of fun. It knows I am an Amateur Radio operator so occasionally signs off with 73.
Hey Steve the Super Repeater sounds a lot like a mental exercise I went through a few years ago. I think that asymmetrical modulation makes a whole lot of sense and pretty much follows what commercial data systems do already.
What you describe is basically how DOCSIS cablemodems work. The key to making it all happen is that the downstream DOCSIS carrier is always transmitting. I'll describe the initialization process for a new modem, maybe it will give you a few ideas.
After the modem's POST completes it begins tuning the downstream tuner, starting at 55 MHz (channel 2). It is looking for 6 MHz wide QAM carrier with an MPEG transport stream. Once it finds one it demodulates the data, looking for a specific MPEG header (D000). When one is discovered (there are others that carry traditional/legacy digital video streams), it then looks for information about the system. This gives the modem a channel plan and basic instruction for how to talk back to the headend (cablemodem termination system, or CMTS). The modem can then move to another downstream channel or continue to use this first channel. IIRC at this stage the CMTS can tell modems to tune to other downstream channels if programmed to do so. The critical information to the modem at this point is what's called the Upstream Channel Descriptors (UCDs). These tell the modem what frequencies and modulation are available on the system to establish a link. The modem chooses a compatible UCD (not all modems can use all modulation options), tunes the upstream modulator (transmitter) to one and starts transmitting. First at minimal power, then ratcheting up the power at 3dB increments until the CMTS acknowledges the modem. Then the modem and CMTS do a little dance to agree on timing and power levels, what other UCDs the modem can/should use, and other downstream carriers that are available such as OFDM (most cable systems have 32 or more 38.8Mbps QAM carriers and 100+ MHz OFDM carriers in the downstream and several different QAM/OFDMA carriers upstream). Once the modem has established a reliable PHY link the CMTS passes along authorization and feature data. Then the modem will start passing customer data.
The thing I can't really figure out is how a radio out of the box will discover the Super Repeater. Sure, there will be more published guides, but why? And how thick will the repeater book get when you have 5 different inputs, each with 4 different modulation options? (Oh, yea, the modems can change modulation on the fly on a per-packet basis even though most cable companies haven't implemented that feature) Are you getting through but the channel is saturated? Will the community tolerate being told when their transmitter can key up? Is your noisy signal causing problems for everyone else? If so will the repeater tell your station to shut down or just pass garbage data (the R2D2 problem)? And hams being hams there will always be those who think they just have to use the channel that is the bestest fastest one even if they're QRMing the airwaves.
A real life example of the cathedral and the bazar problem.
APRS can probably get most of the basics information out there, and GPS disciplined transmitters can help tighten up timeslots (as proven by WSJT-X). But I think a repeater transmitting full time will be the best way to utilize such a repeater (maybe I can't see beyond my cable cowboy past though). Again, this is pretty standard in the commercial/cellular world. The end terminal looks for a host carrier before transmitting (and only with permission from the host). And you can take full advantage of TDMA as well as FDMA. I don't know the legalities of an amateur repeater transmitting 24/7 like a broadcast TV station, and who knows how well the equipment would stand up to such abuse, but hey, we're experimenting, right?
The other nice thing about this sort of set up is that the terminal station can be pretty cheap to build. Get yourself an RTL-SDR dongle, an old TNC and you'll get started at least. Pick up a Kenwood TM-D700 and you get 9600 too. Once you get your feet wet, start saving up for a 1.2GHz radio or a LimeSDR with a 5 Watt Lime RFE amplifier. And as you point out, there are people with repeaters at the QTH (I happen to be sitting next to one as I type this) for DV modes. Building out cross band repeater networks, or remote receivers in QTH locations (with geolocation information known to the terminal radio), would help fill in gaps too. Again, it's all about the UCDs and allowing the terminal radios to QSY based on repeater instructions.
Eric - Your comments illustrate that there's very rarely something truly new... we just keep developing better technology that makes implementing old, bold ideas, first possible, then reasonable, then really great, then easy, cheap, and ubiquitous - like cable modems are now.
One of my favorite examples is that OFDM was known and implemented more than a decade before it came into practical use. The person that told me about it said that we COULD have been using OFDM a decade earlier... but it would have required a dedicated VAX computer to do all the complex calculations that make OFDM work.
Just like your example of a brand new user cable modem listens for a pilot signal to know what system it's connected to and what to do next (what channel it should use to reach out on / phone home), with a Software Defined Receiver as part of the user station, it could simply listen to all of the VHF / UHF bands to hear the most local SuperPeater - best SNR. Worst case, turn it on in the evening and by the next morning, it will have identified all of the SuperPeaters in the area.
Alternatively, a concept I'm "encouraging" the APRS Foundation to start seriously exploring and hopefully developing is to evolve APRS to incorporate a "hailing channel" capability on APRS. Basically, any radio system periodically beacons out its presence on 144.39 MHz (in the US) 1200 bps AFSK. So the SuperPeater would have a small transmitter on 144.39 that sends a periodic beacon out, advertising its presence on another frequency. All the SuperPeaters do this, so your system would know the frequency and location of the SuperPeaters in the area. Then it monitors each one and selects the one that it can receive the best, and also notes any others that seem feasible to be able to use as fallback systems to try to use should the primary become unavailable.
Agreed wholeheartedly that coordination and advertising of systems like SuperPeaters should be automatic and dynamic - not depending on static listings, etc.
Tightly integrated CAT control would probably be necessary too. The receiver wouldn't be a problem of course, SDR dongles can tune easily and the preamp gain controls are out of the box adjustable. But cheap V/UHF transceivers usually don't come with CAT controls, so something would need to be done there. Perhaps hacking the remote mic keypad commands could at lest allow for switching bands and memory channels. But automatically adjusting RF output might not be simple, and automating it based on feedback from the repeater (the "long loop AGC") might not be too popular with the community either. Maybe not a big deal, but dynamic range becomes a problem for the repeater side if there are transmitters blasting the receiver and others barely above the noise. Like a DX pileup the loudest station wins and the little guys quit in frustration. If the repeater can tell the loud terminal to lower power the input could have a lot of extra gain to help with fringe reception.
Eric - I view the use of "simple, stupid, low speed" radios to be a bridge into the use of Software Defined Transmitters. Once we crack those, then all of those features - frequency control, transmit power, modulation index, channel size, etc. all "come for free" as the exact waveform to be transmitted is composed in the software. We have all those features already in the various Software Defined Transmitters for HF - we just have to translate all of that to VHF / UHF.
As previously stated, SDR dongles trade sensitivity for bandwidth. They are as deaf as a post and unsuitable for repeater use. If you do the math for the noise floor at wider bandwidths, you will see that a lot of sensitivity is lost, whereas a narrow band conventional receiver is much more suitable.
Martin - Good point about SD Receivers for use on repeaters. Given that equipment for a repeater is "big investment... but usually one 1x", it may well make more sense to use discrete (sensitive) receivers on a repeater system.
I built a first practical OFDM modem on an FPGA back in 1997 at WiLAN. The first prototype used 6 Xilinx 4000 series parts, the second 2 VIrtex parts, and the third was a single gate array. The FPGA did most of the heavy lifting, including the FFT, the processor the rest of the frame level stuff. I found several practical improvements to the FFT which lowered some of the processing requirements, and was able to process OFDM frames with 256 carriers in a 10 MHz bandwidth, yielding up to 26 Mb/s. We were responsible for rolling out 802.11a in the 5GHz band, which never saw the light of day due to a lack of suitable semiconductors, but it did resurface later as 802.11g.
Martin - Thanks for your inputs. As we discussed in our personal conversations, this quote:
One of my favorite examples is that OFDM was known and implemented more than a decade before it came into practical use. The person that told me about it said that we COULD have been using OFDM a decade earlier... but it would have required a dedicated VAX computer to do all the complex calculations that make OFDM work.
was from your boss at WiLAN during a visit to their facilities to demonstrate WiLAN's (impressive) implementation of OFDM, including true NLOS capabilities.
I'm surprised to see your statement that 802.11a (yes, "a", not "g" because it only worked on 5 GHz) "didn't see the light of day" as I recall buying 802.11a gear for a consulting customer in lieu of deploying (expensive) wired LAN equipment in an office, only to find later that the encryption was minimal at best and easily hacked. That kind of thing where I didn't quite know what I was doing caused me to stop dabbling in IT consulting.
I have read all the specs since DOCSIS 1, and also done work on the OFDM transceiver for the latest creation. There are two major nails in its coffin that make it unsuitable for use over the air, the first is that is assumes a near perfect channel, the second is it has no knowledge of multi-path or fading, as cable systems do not require it. I would suggest as far as using it for amateur radio, let sleeping dogs lie.
Sure. No one is suggesting putting cablemodems on the airwaves (although IIRC WiMAX had a lot of similarities with DOCSIS). I was thinking more along the lines of the handshake that takes place where modems and CMTS devices come to an optimized state of operation as an example of a possible way to get the SuperPeater and terminal radios to function.
Martin - There have been at least two significant adaptations of cable modem technology to non-cable radio that I'm aware of (neither of which persist to the current day). The first was my friend Dewayne Hendricks WA8DZP (Silent Keyboard) use of Com21 cable modems on Amateur Radio spectrum (or maybe not - I forget) in the Bay Area to provide Internet Access at broadband speeds to bunch of friends. The second was a radio manufacturer for Broadband Internet Access, whose name escapes me at the moment, who simply added a transverter to a cable modem. Their usage was envisioned for the never quite realized Sprint-owned spectrum at 2.5 - 2.7 GHz (the former MMDS band) that presumably would have been "quiet and stable" enough for cable modems to work.
I am really, really happy that Jason brought back the SMT TARPN NinoTNC back into production. It makes the NinoTNC accessible for those that aren't comfortable soldering and troubleshooting their build. I understand that building it yourself from mostly through-hole components was a design goal of the NinoTNC, but the SMT TARPN NinoTNC really makes that technology more accessible.
Nice to see a small compact unit back on the market. However, can we understand that 'bit rate' and 'baud rate' are not the same in all cases? The rates that you speak of in your documentation are actually the bit rate. In some modulation methods, such as 2400 DPSK, where 2 bits are coded at a time, the baud rate is actually the symbol rate, which in this case is 1/2 the bit rate. Granted some are the same when one bit is encoded at a time, but in higher orders of modulation this is not the case. For example, 14.4Kb modems actually had a baud rate of 1200.
Regarding your SuperPeater: You omitted a split-site repeater. Although matching RX and TX coverage can be difficult, it makes the repeater MUCH simpler. No duplexer, no desense from the TX. Once you have the split-site infrastructure, multiple RX sites are easy to add; terrific for portable coverage. Another interesting avenue to explore is some form of LTE which allows voice and data and all sorts of other stuff. -de NI0K
John - The article was long enough so I decided to omit mention of split site repeaters. Yes, that is indeed a great capability. I've reported on at least a couple of projects to implement experimental LTE systems on Amateur Radio spectrum - it's no longer exotic, especially not with GNU Radio which, if memory serves, already has LTE modules.
Interesting concept, but I think the need for software defined receivers is overrated. First, I have a lot of experience with the dongle you mention, (RTL-SDR), and their main issue is that they are a a deaf as a post. Second, digital modes used in the amateur world today are all based on C4FM, the variances are the coding scheme and frame formats. All of this has already been done on the MMDVM project. What has not been done so far, as you mention, is a common platform that can control and analog and digital radio simultaneously, and transcode between the two in the same location. I am also not convince that a high speed link is really necessary, as you state there are several solutions already out there, include AREDN which leads the pack. In an 'normal' link between two repeaters voice data can be sent already compressed into lower bit rates inherent in the digital mode, an there is ample bandwidth to overlay other data such as telemetry, APRS and weather information, and keep the occupied bandwidth down to fit existing bandplans.
Martin - We're in violent agreement (with one major exception) that AREDN is the answer to all Amateur Radio data communications requirements for present and future.
The exception is that AREDN is only been usable on microwave frequencies because to date AREDN has only been developed on off-the-shelf microwave radios. In rare places like Southern California where high locations are plentiful and easily (and inexpensively) accessible... great! AREDN is also easily feasible in places / organizations that are fortunate enough to have good relations with tall site owners (skyscraper buildings, broadcast / communications towers, etc. In such situations, microwave is "easy" - users just point their dish at the nearest node, and you have high bandwidth data communications, on "Amateur Radio" frequencies, suitable for video, audio, data, streaming audio, etc.
But most of us in North America aren't so fortunate to have easy, inexpensive access to high locations suitable for microwave networking. AREDN has chosen to keep their development in the software realm for off-the-shelf 5 GHz (and some 2.x GHz) units, but not create new hardware unique to Amateur Radio such as a AREDN unit for 1.24 - 1.30 GHz or 902-928 MHz, or 420-450 MHz, or even higher power units for 5 GHz or 2.x GHz. It's possible to do so... but no one has taken up that challenge. Thus we're left tinkering at the margins for effective data communications with radio hardware that we can get our hands on.
It might not be that way forever given rising use and familiarity with Software Defined Transceivers. It seems feasible to me to port AREDN to, say, a BladeRF unit, choose to transmit on 902-928 MHz, and then add a 902-928 MHz amplifier. It's just that no one has (chosen) done so.
I have use chatGPT for humorous graphics for our club newsletter, it does show a good sense of fun. It knows I am an Amateur Radio operator so occasionally signs off with 73.
Tom.. VK3DMK
Hey Steve the Super Repeater sounds a lot like a mental exercise I went through a few years ago. I think that asymmetrical modulation makes a whole lot of sense and pretty much follows what commercial data systems do already.
What you describe is basically how DOCSIS cablemodems work. The key to making it all happen is that the downstream DOCSIS carrier is always transmitting. I'll describe the initialization process for a new modem, maybe it will give you a few ideas.
After the modem's POST completes it begins tuning the downstream tuner, starting at 55 MHz (channel 2). It is looking for 6 MHz wide QAM carrier with an MPEG transport stream. Once it finds one it demodulates the data, looking for a specific MPEG header (D000). When one is discovered (there are others that carry traditional/legacy digital video streams), it then looks for information about the system. This gives the modem a channel plan and basic instruction for how to talk back to the headend (cablemodem termination system, or CMTS). The modem can then move to another downstream channel or continue to use this first channel. IIRC at this stage the CMTS can tell modems to tune to other downstream channels if programmed to do so. The critical information to the modem at this point is what's called the Upstream Channel Descriptors (UCDs). These tell the modem what frequencies and modulation are available on the system to establish a link. The modem chooses a compatible UCD (not all modems can use all modulation options), tunes the upstream modulator (transmitter) to one and starts transmitting. First at minimal power, then ratcheting up the power at 3dB increments until the CMTS acknowledges the modem. Then the modem and CMTS do a little dance to agree on timing and power levels, what other UCDs the modem can/should use, and other downstream carriers that are available such as OFDM (most cable systems have 32 or more 38.8Mbps QAM carriers and 100+ MHz OFDM carriers in the downstream and several different QAM/OFDMA carriers upstream). Once the modem has established a reliable PHY link the CMTS passes along authorization and feature data. Then the modem will start passing customer data.
The thing I can't really figure out is how a radio out of the box will discover the Super Repeater. Sure, there will be more published guides, but why? And how thick will the repeater book get when you have 5 different inputs, each with 4 different modulation options? (Oh, yea, the modems can change modulation on the fly on a per-packet basis even though most cable companies haven't implemented that feature) Are you getting through but the channel is saturated? Will the community tolerate being told when their transmitter can key up? Is your noisy signal causing problems for everyone else? If so will the repeater tell your station to shut down or just pass garbage data (the R2D2 problem)? And hams being hams there will always be those who think they just have to use the channel that is the bestest fastest one even if they're QRMing the airwaves.
A real life example of the cathedral and the bazar problem.
APRS can probably get most of the basics information out there, and GPS disciplined transmitters can help tighten up timeslots (as proven by WSJT-X). But I think a repeater transmitting full time will be the best way to utilize such a repeater (maybe I can't see beyond my cable cowboy past though). Again, this is pretty standard in the commercial/cellular world. The end terminal looks for a host carrier before transmitting (and only with permission from the host). And you can take full advantage of TDMA as well as FDMA. I don't know the legalities of an amateur repeater transmitting 24/7 like a broadcast TV station, and who knows how well the equipment would stand up to such abuse, but hey, we're experimenting, right?
The other nice thing about this sort of set up is that the terminal station can be pretty cheap to build. Get yourself an RTL-SDR dongle, an old TNC and you'll get started at least. Pick up a Kenwood TM-D700 and you get 9600 too. Once you get your feet wet, start saving up for a 1.2GHz radio or a LimeSDR with a 5 Watt Lime RFE amplifier. And as you point out, there are people with repeaters at the QTH (I happen to be sitting next to one as I type this) for DV modes. Building out cross band repeater networks, or remote receivers in QTH locations (with geolocation information known to the terminal radio), would help fill in gaps too. Again, it's all about the UCDs and allowing the terminal radios to QSY based on repeater instructions.
Eric - Your comments illustrate that there's very rarely something truly new... we just keep developing better technology that makes implementing old, bold ideas, first possible, then reasonable, then really great, then easy, cheap, and ubiquitous - like cable modems are now.
One of my favorite examples is that OFDM was known and implemented more than a decade before it came into practical use. The person that told me about it said that we COULD have been using OFDM a decade earlier... but it would have required a dedicated VAX computer to do all the complex calculations that make OFDM work.
Just like your example of a brand new user cable modem listens for a pilot signal to know what system it's connected to and what to do next (what channel it should use to reach out on / phone home), with a Software Defined Receiver as part of the user station, it could simply listen to all of the VHF / UHF bands to hear the most local SuperPeater - best SNR. Worst case, turn it on in the evening and by the next morning, it will have identified all of the SuperPeaters in the area.
Alternatively, a concept I'm "encouraging" the APRS Foundation to start seriously exploring and hopefully developing is to evolve APRS to incorporate a "hailing channel" capability on APRS. Basically, any radio system periodically beacons out its presence on 144.39 MHz (in the US) 1200 bps AFSK. So the SuperPeater would have a small transmitter on 144.39 that sends a periodic beacon out, advertising its presence on another frequency. All the SuperPeaters do this, so your system would know the frequency and location of the SuperPeaters in the area. Then it monitors each one and selects the one that it can receive the best, and also notes any others that seem feasible to be able to use as fallback systems to try to use should the primary become unavailable.
Agreed wholeheartedly that coordination and advertising of systems like SuperPeaters should be automatic and dynamic - not depending on static listings, etc.
Tightly integrated CAT control would probably be necessary too. The receiver wouldn't be a problem of course, SDR dongles can tune easily and the preamp gain controls are out of the box adjustable. But cheap V/UHF transceivers usually don't come with CAT controls, so something would need to be done there. Perhaps hacking the remote mic keypad commands could at lest allow for switching bands and memory channels. But automatically adjusting RF output might not be simple, and automating it based on feedback from the repeater (the "long loop AGC") might not be too popular with the community either. Maybe not a big deal, but dynamic range becomes a problem for the repeater side if there are transmitters blasting the receiver and others barely above the noise. Like a DX pileup the loudest station wins and the little guys quit in frustration. If the repeater can tell the loud terminal to lower power the input could have a lot of extra gain to help with fringe reception.
Eric - I view the use of "simple, stupid, low speed" radios to be a bridge into the use of Software Defined Transmitters. Once we crack those, then all of those features - frequency control, transmit power, modulation index, channel size, etc. all "come for free" as the exact waveform to be transmitted is composed in the software. We have all those features already in the various Software Defined Transmitters for HF - we just have to translate all of that to VHF / UHF.
As previously stated, SDR dongles trade sensitivity for bandwidth. They are as deaf as a post and unsuitable for repeater use. If you do the math for the noise floor at wider bandwidths, you will see that a lot of sensitivity is lost, whereas a narrow band conventional receiver is much more suitable.
Martin - Good point about SD Receivers for use on repeaters. Given that equipment for a repeater is "big investment... but usually one 1x", it may well make more sense to use discrete (sensitive) receivers on a repeater system.
I built a first practical OFDM modem on an FPGA back in 1997 at WiLAN. The first prototype used 6 Xilinx 4000 series parts, the second 2 VIrtex parts, and the third was a single gate array. The FPGA did most of the heavy lifting, including the FFT, the processor the rest of the frame level stuff. I found several practical improvements to the FFT which lowered some of the processing requirements, and was able to process OFDM frames with 256 carriers in a 10 MHz bandwidth, yielding up to 26 Mb/s. We were responsible for rolling out 802.11a in the 5GHz band, which never saw the light of day due to a lack of suitable semiconductors, but it did resurface later as 802.11g.
Martin - Thanks for your inputs. As we discussed in our personal conversations, this quote:
One of my favorite examples is that OFDM was known and implemented more than a decade before it came into practical use. The person that told me about it said that we COULD have been using OFDM a decade earlier... but it would have required a dedicated VAX computer to do all the complex calculations that make OFDM work.
was from your boss at WiLAN during a visit to their facilities to demonstrate WiLAN's (impressive) implementation of OFDM, including true NLOS capabilities.
I'm surprised to see your statement that 802.11a (yes, "a", not "g" because it only worked on 5 GHz) "didn't see the light of day" as I recall buying 802.11a gear for a consulting customer in lieu of deploying (expensive) wired LAN equipment in an office, only to find later that the encryption was minimal at best and easily hacked. That kind of thing where I didn't quite know what I was doing caused me to stop dabbling in IT consulting.
I have read all the specs since DOCSIS 1, and also done work on the OFDM transceiver for the latest creation. There are two major nails in its coffin that make it unsuitable for use over the air, the first is that is assumes a near perfect channel, the second is it has no knowledge of multi-path or fading, as cable systems do not require it. I would suggest as far as using it for amateur radio, let sleeping dogs lie.
Sure. No one is suggesting putting cablemodems on the airwaves (although IIRC WiMAX had a lot of similarities with DOCSIS). I was thinking more along the lines of the handshake that takes place where modems and CMTS devices come to an optimized state of operation as an example of a possible way to get the SuperPeater and terminal radios to function.
Martin - There have been at least two significant adaptations of cable modem technology to non-cable radio that I'm aware of (neither of which persist to the current day). The first was my friend Dewayne Hendricks WA8DZP (Silent Keyboard) use of Com21 cable modems on Amateur Radio spectrum (or maybe not - I forget) in the Bay Area to provide Internet Access at broadband speeds to bunch of friends. The second was a radio manufacturer for Broadband Internet Access, whose name escapes me at the moment, who simply added a transverter to a cable modem. Their usage was envisioned for the never quite realized Sprint-owned spectrum at 2.5 - 2.7 GHz (the former MMDS band) that presumably would have been "quiet and stable" enough for cable modems to work.
Please note that RPC Electronics SMT TARPN NinoTNC has been discontinued. Only the original thru-hole version of the NinoTNC remains.
We're bringing the SMT NinoTNC back. The website has been updated to show this and we'll start taking orders in about a week...
https://www.rpc-electronics.com/smtninotnc.php
I am really, really happy that Jason brought back the SMT TARPN NinoTNC back into production. It makes the NinoTNC accessible for those that aren't comfortable soldering and troubleshooting their build. I understand that building it yourself from mostly through-hole components was a design goal of the NinoTNC, but the SMT TARPN NinoTNC really makes that technology more accessible.
Nice to see a small compact unit back on the market. However, can we understand that 'bit rate' and 'baud rate' are not the same in all cases? The rates that you speak of in your documentation are actually the bit rate. In some modulation methods, such as 2400 DPSK, where 2 bits are coded at a time, the baud rate is actually the symbol rate, which in this case is 1/2 the bit rate. Granted some are the same when one bit is encoded at a time, but in higher orders of modulation this is not the case. For example, 14.4Kb modems actually had a baud rate of 1200.
Regarding your SuperPeater: You omitted a split-site repeater. Although matching RX and TX coverage can be difficult, it makes the repeater MUCH simpler. No duplexer, no desense from the TX. Once you have the split-site infrastructure, multiple RX sites are easy to add; terrific for portable coverage. Another interesting avenue to explore is some form of LTE which allows voice and data and all sorts of other stuff. -de NI0K
John - The article was long enough so I decided to omit mention of split site repeaters. Yes, that is indeed a great capability. I've reported on at least a couple of projects to implement experimental LTE systems on Amateur Radio spectrum - it's no longer exotic, especially not with GNU Radio which, if memory serves, already has LTE modules.
Interesting concept, but I think the need for software defined receivers is overrated. First, I have a lot of experience with the dongle you mention, (RTL-SDR), and their main issue is that they are a a deaf as a post. Second, digital modes used in the amateur world today are all based on C4FM, the variances are the coding scheme and frame formats. All of this has already been done on the MMDVM project. What has not been done so far, as you mention, is a common platform that can control and analog and digital radio simultaneously, and transcode between the two in the same location. I am also not convince that a high speed link is really necessary, as you state there are several solutions already out there, include AREDN which leads the pack. In an 'normal' link between two repeaters voice data can be sent already compressed into lower bit rates inherent in the digital mode, an there is ample bandwidth to overlay other data such as telemetry, APRS and weather information, and keep the occupied bandwidth down to fit existing bandplans.
Martin - We're in violent agreement (with one major exception) that AREDN is the answer to all Amateur Radio data communications requirements for present and future.
The exception is that AREDN is only been usable on microwave frequencies because to date AREDN has only been developed on off-the-shelf microwave radios. In rare places like Southern California where high locations are plentiful and easily (and inexpensively) accessible... great! AREDN is also easily feasible in places / organizations that are fortunate enough to have good relations with tall site owners (skyscraper buildings, broadcast / communications towers, etc. In such situations, microwave is "easy" - users just point their dish at the nearest node, and you have high bandwidth data communications, on "Amateur Radio" frequencies, suitable for video, audio, data, streaming audio, etc.
But most of us in North America aren't so fortunate to have easy, inexpensive access to high locations suitable for microwave networking. AREDN has chosen to keep their development in the software realm for off-the-shelf 5 GHz (and some 2.x GHz) units, but not create new hardware unique to Amateur Radio such as a AREDN unit for 1.24 - 1.30 GHz or 902-928 MHz, or 420-450 MHz, or even higher power units for 5 GHz or 2.x GHz. It's possible to do so... but no one has taken up that challenge. Thus we're left tinkering at the margins for effective data communications with radio hardware that we can get our hands on.
It might not be that way forever given rising use and familiarity with Software Defined Transceivers. It seems feasible to me to port AREDN to, say, a BladeRF unit, choose to transmit on 902-928 MHz, and then add a 902-928 MHz amplifier. It's just that no one has (chosen) done so.
Yet.