Global Backbone Project

From AB9RF

One of my major objections with Echolink and IRLP is that both rely on the Internet for transit. This means that they're dependent on non-ham resources to operate, and that means that they cannot be counted on in the event of a telecommunications failure.

This illustrates one of the major areas in which ham radio currently falls short: we don't have a mechanism for high-bandwidth long-range transit of digital signals. We have lots of mechanisms for low-bandwidth transit, ranging from good old CW through HF packet. None of these is fast enough to send even one channel of digitally-encoded speech, though. A new replacement packet protocol is needed.

The proposal here contains a number of interrelated parts:

1. A mesh of high-bandwidth line-of-sight links covering, ideally, as much of the world as possible; at the very least the entire United States and as much of Canada as possible

   This is envisioned as being done using line-of-sight point-to-point communications in, probably, 10 GHz. This gives typical hop distances of between 25 and 200 miles depending on terrain. Between 2000 and 3000 nodes will be required to cover the entire continental United States. Europe would require somewhat fewer.

2. A means for operators local to a node in the mesh to gain access to the mesh and connect with another operator local to some other node in the mesh

   This can be an omnidirectional station directly connected to a mesh node operating in SHF, UHF, VHF, or even HF running (a variation on) the packet protocol used by the mesh

   It can also be a modified voice repeater coupled with software to encode and decode voice into packet data which is then injected into the mesh

3. A new packet protocol engineered around IPv6 (probably) that can support the dynamic mesh routing that this project requires.

   IPv6's hardware address field is long enough to directly encode a call sign along with several bits worth of "transmitter number"

   There are already existing IPv6 distributed mesh routing algorithms out there that can be adapted to this purpose

Problems:

1. The radios are going to be expensive

   Possibly not. We may be able to use commodity 802.11 radios transverted to 10 GHz. We might even decide to tweak commodity 802.11a or 802.11b radios as both of the ISM bands used for wireless networking are adjacent to, or even overlap, ham allocations. We think we can keep the node cost at around $1000 each. 10,000 of them (3000 nodes with two to four radios each) will cost about $10 million, or $13 per licensed ham in the United States.

2. We need to site and install thousands of them

   We may be able to convince existing repeater operators to colocate a node with their existing repeater(s).

   We can use tools like splat and the FCC tower database (in the United States) to identify potential candidates for installing nodes to fill gaps, and then lobby tower owners for permission to install. The emergency communications value of this network may help us gain access to towers operated by public authorities.

3. There's no way we'll get line of sight between continents or to isolated islands like Hawaii

   We can use satellites. Latency may be an issue, but this isn't a problem for non-real-time data and can be acceptable for voice to a point.

   The protocol can readily be defined to scale down to the lower bandwidths allowed in SHF, UHF, VHF, and HF, and communications not requiring high bandwidths can be sent over links in these bands. HF, in particular, allows over-the-horizon operation and can be used worldwide, conditions permitting.