Radio-Parameter Selection Algorithm for Receiver-Directed Packet-Radio Networks(SRNTN-73)

摘要

This report describes the Parameter Selection Algorithm for the SURAP 4 packetradio network and the basis for its design. It presents a Semi-Markov model of the algorithm operation. The algorithm adaptively adjusts three radio parameters: transmitter power, FEC coding rate, and channel bit rate. These parameters can increase the perceived signal-to-noise ratio on the channel and are treated as gain mechanisms. The algorithm must maintain robust radio links between nodes whenever possible, while minimizing interference with other network transmissions. We present one heuristic approach to balancing these two goals. This approach uses barely sufficient gain to maintain the packet error probability below an acceptable threshold value. A threshold probability of 0.1 was chosen. Bit error statistics provided by the sequential decoder and error feedback packets assist the process of determining the appropriate change of gain as noise levels vary. The algorithm can adapt at a fast rate, sometimes on the order of one packet time. It maintains acceptable packet error probability and can recover from estimation and decision errors. A Semi-Markov model can be used to compute the steady state packet error probability and packet transmission overhead due to noise. The model is general enough to include a stationary model of transceiver blocking probability, dynamic CDMA interference, and alternative algorithms based on similar operating principles. Although the model does not incorporate bit error statistics, a way is proposed to model the effect of these statistics without compromising the simplicity of the Semi-Markov description. (jhd)