A Proposal for a Wireless Internet of Portable Nodes
The prospects that the Y2K problem, or other such cascade failure arising
from other causes, might bring down power grids and telecommunications systems,
suggests that we consider establishing an alternative communications system
that could support emergency services in such an event.
One of the main ways that emergency communications has been provided in the
past is through amateur radio networks. However, the advent of packet radio and
other forms of digital communications on amateur bands suggests a better use of
the bandwidth, especially since a nationwide or worldwide failure of essential
systems would be likely to overwhelm the ability of traditional amateur voice
communications to cope with the situation.
A number of amateur and commercial researchers have been exploring expansion
of the Internet to radio links, using the TCP/IP and AX-25 protocols and making
use of both the amateur bands and the newly available spread spectrum bands.
These are represented by such efforts as FlexNet and AMPRnet, and such services
as Ricochet. See http://www.constitution.org/wipnet/wipnet.htm
This is to suggest that we get together to establish a much more ambitious
system. The effort is complicated by the need to confine the use of amateur
bands to licensed amateurs, which is not a limitation for the use of spread
spectrum bands. The following are the design specifications for such a system.
- All nodes would be portable computers and transceivers. There would be no
wired backbone, although there could be connections to the wired backbone to
make use of it, if it is available, such as to span what would otherwise be
breaks in the network.
- The geographic range of the system would be nationwide or worldwide. Any
node would be able to send an email message to any other anywhere on the
network, regardless of the number of hops in between. There would also be
support for newsgroup, ftp and WWW services, but these might be restricted in
the event of an emergency.
- The primary bands would be spread spectrum, with perhaps some use of HF
bands for longer distances, to span breaks in the network.
- The addressing and routing scheme would permit all nodes to move about
freely, and perhaps rapidly, without every node having to have a complete
database of every node and its location, although such a database might be
available on some nodes, making them DNS servers.
- There would be a priortization protocol to permit high priority messages
to have routing priority over lower ones.
There are many advantages to the use of spread spectrum. It supports higher
bandwidths, makes more efficient use of a given band, and provides better
security. At the moment its use is limited to the 900 MHz and 2.4 GHz bands,
and to 1 W, but amplifiers are available for use outside the U.S., and we can
expect that use at higher powers would be excusable in the event of a national
emergency, as would the use by non-licensed personnel. A protocol would have to
be adopted for exchanging sequence codes among cooperating transceivers, but
PGP-encrypted messages on a standard contact sequence could be used for that
and for authentication of the nodes. It would also be better suited for
preventing packets from going astray.
It would be possible, in principle, for nodes within range of each other to
update one another's link databases, and for those databases to be transmitted
to other nodes to update their, on a continuing basis, so that every node would
maintain an updated database of every node and which was in range of which
other, however, this would involve too heavy an overhead. A different approach
would be for each node to transmit its approximate grid location, either set
manually or taken from GPS, and messages routed along a node path that would
take them in the general direction of the target node until a link was
The line-of-sight range of both VHF FM and SS is about 20 miles, under good
conditions, and using antennas that could be carried by a person. That means
that to establish a nationwide network, and to provide sufficient bandwidth for
heavy message traffic, there would need to be a fairly dense positioning of
nodes not more than 20 miles apart, and preferably closer, across the country.
This could be a problem in some of the Western states, especially desert
country, where population density is low, but in an emergency node-bearing
operators could scatter themselves and position themselves to make sure that
coverage was complete. Three or four operators in each of the 3000 counties in
the United States might be sufficient.
Small notebook or wearable computers with large disk drives are now becoming
available, and they should soon become fairly inexpensive. SS 1 W transceivers
are also becoming available at a cost of about $200 each. The only other things
that would be needed would be power packs, a good antenna, and perhaps extra
disk drive capacity, all of which could be carried by an operator in a
The importance of portability must be emphasized. In an emergency, the
availability of fuel to operate a vehicle cannot be assumed, although vehicle
based mobile systems might be made use of if available. But operators should
plan on being able to operate on foot, perhaps using solar cells to recharge
their batteries. The operator will also have to carry water, food, medical, and
other survival supplies on him, so the communications equipment needs to be as
small, light, and easy to use while on the move as possible.
The military has such systems, but, typically, they are too expensive for
civilian use. Indeed, they are too expensive for widespread use by the
military, which is why we cannot depend on military or national guard personnel
to provide such services in an emergency. This needs to be a civilian,
autonomous system that is not under central authority or control, like the
Internet itself, and that therefore cannot easily be interdicted or interfered
with by anyone.
We need to vigorously pursue the development, testing, and deployment of
such a system, with the goal of full deployment not later than October, 1999.
Comments for ways to do this are welcome.