USE OF GROUND PENETRATING RADAR FOR LANDMINE CLASSIFICATION BASED ON ARTIFICIAL NEURAL NETWORK B y P . S . L . F e r n a n d o T h i s t h e s i s i s s u b m i t t e d t o t h e D e p a r t m e n t o f E l e c t r o n i c a n d T e l e c o m m u n i c a t i o n E n g i n e e r i n g a t t h e U n i v e r s i t y o f M o r a t u w a i n p a r t i a l f u l f i l m e n t o f t h e r e q u i r e m e n t s f o r t h e D e g r e e o f M a s t e r o f E n g i n e e r i n g University of Moratuwa 83814 O c t o b e r 2 0 0 4 Th e s T s B3&I4 USE OF GROUND PENETRATING RADAR FOR LANDMINE CLASSIFICATION BASED ON ARTIFICIAL NEURAL NETWORK Submitted By P . S . L . F e r n a n d o Examining Committee Dr. V.P.C Dassanayaka (Chairperson) Dr. D.A.I. Munindradasa Dr. K.G.P. Dharmawardena This Research Project was carried out at the Department of Electronic and Telecommunucation Engineering of the University of Moratuwa during the period from February 2002 to October 2004 October 2004 DECLARATION T h e w o r k p r e s e n t e d i n t h i s d i s s e r t a t i o n h a s n o t b e e n s u b m i t t e d f o r f u l f i l m e n t o f a n y o t h e r d e g r e e . D r . D . A . I . M u n i n d r a d a s a S u p e r v i s o r « Acknowledgements The research presented in this thesis was carried out at the Department of Electronic and Telecommunication Engineering, University of Moratuwa during the period from February 2002 to October 2004. First and foremost, I am indebted to Dr D. A. I. Munindradasa, my supervisor for his ever present guidance, encouragement, suggestions and constructive critisim while this research was being conducted. The author is most grateful to Dr V. P. C. Dhasanayaka for acting as the chairperson in the examining committee. I also have to express my deepest thanks to Dr E. C. Kulasekere for his great assistance and continuous support throughout this research project. A very special thanks go to Dr R. Nilavalan, Communication Research Institute, University of Bristol, UK for providing valuable and current information on GPR. I wish to express my thanks to Dr K. G. P. Dharmawardana for the coordination of this project. I am also pleased to acknowledge the contribution everyone has made in developing a specialist area of ground penetrating radar technology. P .S .L . F E R N A N D O University of Moratuwa October 2004 iv ¥ N O M E N C L A T U R E Mo A b s o l u t e m a g n e t i c s u s c e p t i b i l i t y o f f r e e s p a c e p C o n d u c t i v i t y o f t h e s o i l ( S / m ) Gt G a i n o f t h e t r a n s m i t t i n g a n t e n n a ( d B ) Gt G a i n o f t h e r e c e i v i n g a n t e n a ( d B ) Vo I n t r i n s i c i m p e d a n c e o f t h e f r e e s p a c e (Cl) Vi I n t r i n s i c i m p e d a n c e o f t h e s o i l ( Q ) V2 I n t r i n s i c i m p e d a n c e o f t h e b u r i e d o b j e c t (0.) t a n S L o s s t a n g e n t o f t h e m a t e r i a l f O p e r a t i n g f r e q u e n c y ( H z ) eo P e r m i t i v i t y o f t h e a i r ( F / m ) Mr R e l a t i v e p e r m i a b i l i t y o f t h e s o i l € r R e l a t i v e p e r m i t i v i t y o f t h e s o i l Mr R e l a t i v e m a g n e t i c s u s c e p t i b i l i t y o f m a t e r i a l a S i g n a l a t t e n u a t i o n c o n s t a n t o f t h e s o i l ( d B / m ) f Abstract This research is mainly aimed at developing a technique based on neural networks to classify metal and plastic objects buried within a range of soil conditions. In addition, the validity of this technique is also presented. The explosives in land mines are generally cased in metal or plastic containers. Identi­ fication of buried metal and plastic objects using a neural network and a sensing technique based on an electromagnetic method are discussed in this thesis. Neural network simulation results for plastics and metal objects in the range of soil condition are also reported. Finding the appropriate frequency window (FW) for the Ground Penetrating Radar (GPR) operation and the development of a theoretical mathematical model is also presented. Using this model, the appropriate F W for GPR operation is derived. Furthermore the estimation of important system parameters of GPR, modulation and detection techniques, modelling of GPR, and clutter reduction techniques are also discussed in the context of this thesis. Contents A c k n o w l e d g e m e n t s 1 V Abstract v i * List of F igures x List of Tables xii C H A P T E R 1 In troduct ion 1 1 .1 F r e q u e n c y W i n d o w 1 1.2 M o t i v a t i o n 2 t 1 .3 O b j e c t i v e o f t h e R e s e a r c h 2 1 .4 A r t i f i c i a l N e u r a l N e t w o r k A p p r o a c h 2 1.5 P r a c t i c a l l i m i t a t i o n s 3 1.6 O r g a n i z a t i o n o f P r e s e n t a t i o n 4 C H A P T E R 2 Theore t i ca l Frequency W i n d o w for G P R 5 2 . 1 I n t r o d u c t i o n 5 | 2 . 2 S u r f a c e C l u t t e r 6 2 . 2 . 1 M i n i m i z i n g I m p e d a n c e M i s m a t c h 6 2 . 2 . 2 S i g n a l A t t e n u a t i o n 8 2 . 2 . 3 R e s o l u t i o n 1 0 C H A P T E R 3 E s t i m a t i o n of S y s t e m P a r a m e t e r s 12 3 . 1 I n t r o d u c t i o n 1 2 3 . 2 E v a l u a t i o n o f L o s s e s 1 2 * 3 . 2 . 1 P a t h L o s s 1 2 3 . 2 . 2 A n t e n n a L o s s Le 1 3 3 . 2 . 3 A n t e n n a M i s m a t c h L o s s Lm 1 3 v i i 3.2.4 Transmission Loss from Air to Soil Lt\ 13 3.2.5 Retransmission Loss from Soil to Air Lte 13 3.2.6 Spreading Loss Ls 13 3.2.7 Target-Scattering Loss Lsc 14 3.2.8 Material Attenuation Loss La 15 3.2.9 Processing Gain PG 15 3.3 System Equation 15 3.4 Estimation of Buried Distance 16 3.4.1 Calculation of Relative Permittivity of the Soil 16 C H A P T E R 4 M o d u l a t i o n Techniques 19 ¥ 4.1 Introduction 19 4.2 Characteristics of Modulation Techniques 19 4.2.1 Frequency Modulated Continuous Wave 19 4.2.2 Synthetic Pulse Technique 20 4.2.3 Base Band Pulse BBP 20 4.3 Selecting the Appropriate Modulation Technique 21 * C H A P T E R 5 P o s t R e c e p t i o n Synthet i c Focusing 22 5.1 Introduction 22 5.1.1 Spot Focusing 22 5.1.2 Processing Gain and Resolution 22 5.1.3 Post Reception Synthetic Focusing(PRSF) 23 5.1.4 Advantages of the PRSF Technique 24 % C H A P T E R 6 M o d e l i n g G P R 27 6.1 Introduction 27 6.2 Mathematical Analysis 28 6.2.1 Losses due to Dielectric Discontinuities Ldd 29 6.2.2 Antennae Spreading and Attenuation Loss Ls 30 6.2.3 Mathematical Model 30 6.2.4 Generating da ta 31 C H A P T E R 7 N e t w o r k Training and Simulat ion 33 7.1 Introduction 33 7.2 Data Structure for Training and Simulation 3 3 ^ * * , ^ v v i i i ; I IWJ § 7.3 BPN Architecture 34 7.4 BPN Training 35 7.5 Network Simulation 37 7.6 Conclusion 37 C H A P T E R 8 Conc lus ion and Future Work 42 8.1 Conclusion 4 2 8.2 Future Work 4 3 Bibl iography 44 A P P E N D I X A Back P r o p a g a t i o n Network 47 * A P P E N D I X B Intrinsic I m p e d a n c e and Skin D e p t h 48 B.l Instrinsic Impedance n 4 8 B.2 Skin Depth Ds 4 8 A P P E N D I X C M a t h e m a t i c a l Analys i s of G P R 50 « ix List of Figures 2.1 Variation of magnitude of impedance with frequency in wet soil 7 2.2 Variation of phase angle of impedance with frequency in wet soil 8 2.3 Variation of magnitude of impedance with frequency in dry soil 8 * 2.4 Variation of phase angle of impedance with frequency in dry soil 9 2.5 Variation of skin depth with frequency in wet soil 9 2.6 Variation of skin depth with frequency in dry soil 10 3.1 Transmitter output against target distance. This variation is exponential. G P R demands high power in soil having a higher attenuation constant. Peak to peak voltage required by GPR to detect a object at 50 cm, in a moderate * soil is around 25 V 17 4.1 Commonly used base band pulses and their frequency spectrums 21 5.1 Near field spot focusing 23 5.2 Post-Reception Synthetic focusing GPR 25 5.3 Time domain coherent addition of target signals at the target location . . . 26 * 6.1 Schematic diagram for a typical GPR 28 7.1 Variation of Signal Level with Buried Distance. Scanning a buried object with closely spaced different frequencies, distinctive signal levels can obtained. These signal levels are a function of attenuation constant of the soil and relative permittivity of the object 34 7.2 Variation profile of validation error of Metal and Plastic over the 32 number of training steps. The minimum error for metal and plastic occurs at 2 1 s t , 25rd, and 25th training step 38 7.3 For Metal and Plastic in the soil condition ranging from 1 d B / m to 10 d B / m network gives two distinctive outputs below 33 cm mean depth 40 x 7 . 4 V a r i a t i o n o f t h r e s h o l d l e v e l s w i t h s o i l a t t e n u a t i o n i n s t e p o f 1 d B / m f o r t h e r a n g e o f 3 d B / m t o 1 0 d B / m . O b j e c t c a n b e c l a s s i f i e d a s M e t a l o r P l a s t i c w h e n n e t w o r k o u t p u t i s g r e a t e r t h a n 0 . 4 4 a n d l e s s t h a n 0 . 4 1 r e s p e c t i v e l y . . 4 1 * A . l T h r e e l a y e r e d B P N a r c h i t e c t u r e 4 7 C . l P r o p a g a t i o n o f s i g n a l i n G P R a t a p r e s e n c e o f a b u r i e d o b j e c t 5 0 x i « List of Tables 2 . 1 S o i l t y p e s a n d t h e i r e l e c t r i c a l a n d d i e l e c t r i c p r o p e r t i e s 7 3 . 1 T a r g e t t y p e a n d k v a l u e 1 4 3 . 2 R a d a r C r o s s - S e c t i o n s 1 6 3 . 3 M a t e r i a l L o s s a t 1 0 0 M H z a n d 1 G H z 1 8 6 . 1 A t t e n u a t i o n a n d l o s s t a n g e n t m e a s u r e d a t 1 G H z 3 1 x i i c