A broadband slot-loaded trapezoid microstrip antenna

✍ Scribed by Mon-Chun Pan; Kin-Lu Wong

• Publisher: John Wiley and Sons
• Year: 2000
• Tongue: English
• Weight: 135 KB
• Volume: 24
• Category: Article
• ISSN: 0895-2477
• DOI: 10.1002/(sici)1098-2760(20000105)24:1<16::aid-mop6>3.0.co;2-y

✦ Synopsis

of Targets Correctly Classified with Respect to the Number of Tests Considered for Different Signal-to-Noise Ratios Signal-to-Noise Ratio Targets Correctly Classified SrN s 50 dB 30r30 SrN s 40 dB 30r30 SrN s 30 dB 28r30 SrN s 20 dB 27r30

than 90%, while for a signal-to-noise ratio greater than 40 dB, all of the targets are correctly classified.

Conclusions

In this letter, we have presented the application of a neural network to the electromagnetic classification of cylinders embedded in a lossy medium between a conductor and dielectric objects. We showed that the proposed NN technique allows us to achieve a high percentage of successful classifications, even in the presence of targets characterized by dielectric parameters significantly different from those considered during the training phase and for noisy input data. Moreover, the classifications are performed very quickly. Therefore, the proposed approach seems to represent an interesting solution for on-line applications. Moreover, the possibility to classify a target considering only the values of the backscattered field also seems to reasonably indicate a possible future application for cylindrical targets buried in half-space geometries.