| Autor: |
Stranne, K. O. F., Asraf, D. E., Robinson, J. W. C., Lindqvist, P., Sigray, P. |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2000, Vol. 502 Issue 1, p603, 6p, 1 Diagram, 8 Graphs |
| Abstrakt: |
In this work we demonstrate information theoretic ‘resonance’ type phenomena in the performance of a simple static magneto-resistive device. We focus on signal detection applications where probability of detection together with so called &fgr;-divergences, computed for the output, are used as performance measures. In particular we show that, when detecting a DC signal injected on the input under constant SNR, for all but very small input noise levels the performance deteriorates very rapidly with increased noise level. However, using nonlinear theory and by applying an appropriate bias one can ‘tune’ the device to give high performance for a whole range of noise strengths. Maximization of statistical information processing capabilities as a performance criteria for a nonlinear system represents a new approach to sensor/device design. Such an approach is more appropriate than the classical linearization approach in applications where the system is expected to be ‘noise saturated.’ © 2000 American Institute of Physics. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
