SIGNAL OPTIMIZATION FOR CHANNELS WITH FEEDBACK.
STANFORD UNIV CA STANFORD ELECTRONICS LABS
Pagination or Media Count:
In this work, a communication receiver is regarded as a dynamical system described by a linear difference equation where the last state is the test statistic upon which a decision is based. When noiseless feedback to the transmitter ia allowed, the signal selection problem becomes essentially a stochastic control problem. By choosing a minimum mean square error criterion, signals are found that exploit the feedback to achieve considerable reduction in coding and decoding complexity over what would be needed for comparable performance with the best signals for the one-way channel. The schemes developed exhibit error probabilities that decrease in a double exponential manner with increasing signal duration for both additive and multiplicative noise channels. This control theory approach depends only on the first- and second-order statistics of the noise, handles multiplicative noise in addition to additive noise in the forward channel, and extends naturally to consideration of noise in the feedback link. The schemes developed could be very important for satellite communication, since they allow for a substantial decrease in the coding effort while permitting the satellite to transmit its information at a rate arbitrarily close to channel capacity. Author
- Statistics and Probability
- Unmanned Spacecraft
- Non-Radio Communications