DARPA Packet Radio Network
This article is still in construction!
Much of today‘s wireless technology (see Wireless Networks) is based on the inspiration, ideas and concepts that have been developed at the DARPA PRNet Project. The fundamental problems of mobility and adhoc networking are still the same as 30 years ago, so it is worth having a look on the first solutions which finally led to what we know as wireless networks today.
In 1973, the Defense Advanced Research Projects Agency (DARPA) initiated research on the feasibility of using packet-switched radio communications to provide reliable computer communications. This development was motivated by the need to provide computer network access to mobile hosts and terminals, and to provide computer communications in a mobile environment. The DARPA PRNET has evolved through the years (1973-1987) to be a robust, reliable, operational experimental network. Capabilities have been increased step by step as new algorithms were developed and implemented. During 14 years of development the DARPA PRNET has been in daily use (experimental purposes) for about 10 years. The DARPA PRNET projects includes network devices, routing protocols and protocols for automatic distributed network management.
The Packet Radio (PR) was the main component of a PRNet.
Firmware can be loaded into a PR either locally (via serial interface) or from the PRNET. If a new version runs on one PR all the devices that have contact to the PR with the new firmware will receive & install this new version automatically. The firmware in each PR gathers information about bidirectional link quality, nodal capacity and route characteristics and provides this knowledge to debugging and monitoring. Debugging can be done by connecting to a PR, either locally via terminal on RS-232 interface or remotely via PRNET.
As one can tell from the name packet radios use radio frequency technology to transmit and receive data. The implmented packet radios support omnidirectional, spreadspectrum, half-duplex transmission and reception at 400kbit/s and 100kbit/s rates. They implement the physical, data link and network layer (OSI model).
The devices implement chip modulation at a 12.8 mchips/s rate which produces a direct-sequence, spread-spectrum waveform using a pseudo-noise sequence. This pseudo-noise sequence is a function of 56bit-Key,25bit-Slottime and a 32bit-qualifier. DES (Data Encryption Standard) produces the pseudo-noise-code (packets are encoded). If slot times are used to produce the code the network needs to be time-synchonized.
Radio Frequency Capabilities
20 frequencies between 1718.4MHz and 1840.0MHz are available for the channel selection. Channel-hopping is possible on a packet-by-packet basis with 3ms settling time. Also attenuation to the signal can be changed on a packet-by-packet basis (adaptive power control) from 0dB to 24dB in 8-dB Steps to effectively control connection quality and power consumption.
Several measurements are available to the firmware: Receive power, signal/noise, noise, multipath, FEC error count They can be used to better quantify the qualities between neighbouring PRs.
Forward Error Correction „repairs“ received packets that are "broken" and so prevents retransmission. To achieve this repair capability convolutional codes & sequential decoding are applied (half-rate coding).
Cyclic Redundancy Checksum (32bit) is used to provide a "good" error detection. If a packet fails, it is discarded and the receiving node waits for retransmission, which follows automatically if there is no acknowledgement.
Burst noise is mitigated by interleaving
Routing and Network Management
The routing protocols used in PRNet are designed to enable reliability, speed and correctness and thus include network management facilities. Routing can be split up into two main protocols: forwarding and transmission.
Automatic and distributed network management results from the use of the protocols above. The result of this network management is that the DARPA PRNet can be installed and deployed quickly and easily. Almost nothing needs to be configured. Devices need to be put into their locations and powered on, then the network initialises itself.