IEEE Spectrum: Build a Long-Distance Data Network Using Ham Radio

(Source: F4HKD via YouTube)

(Source: IEEE Spectrum)

Send data via IPv4 up to 300 kilometers with easy-to-assemble hardware


I have been a hobbyist and maker for almost 15 years now. I like inventing things and diving into low-level things. In 2013, I was looking at a protocol called NBP, used to create a data network over amateur radio links. NBP was developed in the 2000s as a potential replacement for the venerable AX.25 protocol [PDF] that’s been in use for digital links since the mid-1980s. I believed it was possible to create an even better protocol with a modern design that would be easier to use and inexpensive to physically implement.

It took six years, but the result is New Packet Radio (NPR), which I chose to publish under my call sign, F4HDK, as a nom de plume. It supports today’s de facto universal standard of communication—the Internet’s IPv4—and allows data to be transmitted at up to 500 kilobits per second on the popular 70-centimeter UHF ham radio band. Admittedly, 500 kb/s is not as fast as the megabits per second that flow through amateur networks such as the European Hamnet or U.S. AREDN, which use gigahertz frequencies like those of Wi-Fi. But it is still faster than the 1.2 kb/s normally used by AX.25 links, and the 70-cm band permits long-distance links even when obstructions prevent line-of-sight transmissions.

Initially, I considered using different frequency bands for the uplink and downlink connections: Downlinks would have used the DVB-S standard, originally developed for digital satellite television. Uplinks would have used a variation of FSK (frequency-shift keying) to encode data. But the complexity involved in synchronizing the uplink and downlink was too high. Then I tried using a software-defined radio equipped with a field-programmable gate array (FPGA). I had some experience with FPGAs thanks to a previous project in which I had implemented a complete custom CPU using an Altera Cyclone 4 FPGA. The goal was to do all the modulation and demodulation using the FPGA, but again the method was too complex. I lost almost two years chasing these ideas to their dead ends.

Then, in one of those why-didn’t-I-think-of-this-earlier moments, I turned to ISM (industrial, scientific, and medical) chips.[…]

Click here to read the full article.

Many thanks to SWLing Post contributor, Marty, for the tip!

Spread the radio love

3 thoughts on “IEEE Spectrum: Build a Long-Distance Data Network Using Ham Radio

  1. DL4NO

    Here in central Europe we have built a WLAN-like network called HAMNET – see Many OMs dream to access the HAMNET on 70 cm, as 13 cm (2.3 GHz) is a strictly line-of-sight medium.

    F4HDK took a very big step in the right direction, but the dream cannot be realized through this technology:

    * NPR uses FSK hardware. That means that the output power is constant and the frequency is one of four frequencies. This allows to transmit 2 bits at the same time, “per symbol”. Subscract the overhead of premambles, forward error correction and control and about 1 bit per symbol is usable.

    * Any pure FSK modulation is quite low on efficiency in bps/Hz bandwidth. Here in Europe the 70 cm HAM band is 10 MHz wide – in Scandinavia only 6 MHz are available. For a 500 kbps data rate a 500 kHz channel is needed. From OM to OM you might be able to arrange a time slot in the 434 MHz range if other OMs do not unse it for digital TV or other wideband uses. But the band plan only reserves two 200 kHz channels for automatic stations. Here NPR could only provide a real-world bandwidth of 150 kbps. I have seen a HTTP transfer of 12 kB/s.

    * NPR uses a single carrier which means that the symbol rate is up to 500 kHz. This means that reflections very easily mean that one symbol overhaules the previous one. Therefore you will find very few longer tansmission paths that work.

    Summary: NPR is a big step in the right direction and I hope that the low-cost entry will encourage many OMs th experiment with NPR.

    F4HDK needs MUCH help if NPR is to evolve into a standard operating mode. I see several seperate directions for further development:

    * The symbol rate must be lowered by a higher factor, for example 10…50. This would reduce the inter-symbol crosstalk. This means not 1 carrier, but 10…50 parallel carriers. Yes, this would complicate the hardware as described in the article.

    * In the same context more bits per symbol must be transferred – from 2 to 6-8. There are systems that pack 1 Mbit/s into a 200 kHz channel.

    * We should develop a ham-specific protocol stack. TCP/IP is an extremely chatty protocol family. The smallest package is 46 bytes long and there are extremely many of those – for example to acknowledge received packets.

    If you have realized how many packets are sent to transfer a single, very small mail, you will be stunned. We cannot avoid all this on the outside. But within our air gap protocols we could do MUCH better. For example those 46-byte packets cound be reduced to 4 bytes. In FIDOnet times we transferred several 100 mails with one handshake to start and another handshake to acknowledge the end of the transmission.

    1. F4HDK

      Hello, thank you for this comment.
      I know that my NPR project is far from perfect, especially on the modulation used.
      But in my opinion, it has 3 advantages:
      1) it exists, it is available, useable today (unlike other recent 70cm projects)
      2) it is cheap (80$ each modem)
      3) the “multipath/intersymbol” problem inherent to high symbol-rate can be mitigated via using directional antennas (Yagi) at “client” side. At 100kS/s, it gives really good results.

      For the 200kHz chanels available in Germany (100kS/s), you can expect ~130kbps of real datarate (200kbps raw). Once again, that’s not perfect, but that’s truely useable.

  2. Michael Black

    Some if it is hazy, but amateur radio packet came about because someone at tge DOC here in Canada, John de Mercardo (I think I spelled that right) decided tge service needed a new direction. He’dbeen in Hawaii and seen Halohanet, which took packet from ARPAnet and put it over the radio.

    So he implemented a digital license, no code testing and valid onky at 220MHz and up, and did something to allow for packet, maybe he authorized it. It’s been a !ong time. So at the May 1978 meeting of the Montreal Amateur Radio Club, he was there to talk about tge changes, and there was a public display of the new node, said to be the first public display. I’m hazy because around the same time some other group had plans or kits for a packet controller so maybe the idea came independently from two groups.

    Packet radio was big for a while, but it was designwd to ride on everyday 2M FM rigs, so speed was 1200baud. Later there were attempts at higher speeds, but that meant dedicated radio hardware, so other tgan backbone, high speeds never became common.

    Another issue is that ham radio is non commercial, so tgere’s a limit on what can be sent. As the novelty died off, packet radio faded.

    Maybe now it’s different, the hardware cheaper. But any new mode needs users, which in turn brings more users. So even if something is good, it may fail. And what will this high speed protocol be used for?



Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.