Cellular Nanoscale Sensory Wave Computing (eBook)

Preface 5
Contents 7
1 A Brief History of CNN and ONR 9
2 Cellular Wave Computing in Nanoscale via Million Processor Chips 13
2.1 Introduction 13
2.2 From Standard CNN Dynamicsto the Cellular Wave Computer 16
2.3 Various physical implementationsof the Cellular Wave Computer 19
2.4 Virtual Cellular Machine 20
2.4.1 Notations and Definitions 20
2.4.1.1 Core=Cell 20
2.4.1.2 Elementary Array Instructions 21
2.4.2 Physical Implementation Types of Elementary Core/Cell Array Instructions (A, B, C) 21
2.4.3 Physical Parameters of Array Processor Units (Typically a Chip or a Part of a Chip) and Interconnections 22
2.4.4 Virtual and Physical Cellular Machine Architectures and Their Building Blocks 22
2.4.5 The Design Scenario 24
2.4.6 The Dynamic Operational Graph and its Use for Acyclic UMF Diagrams 25
2.5 Recent, Non-Standard Architecture Combining Spatial-Temporal Algorithms with Physical Effects 26
2.6 Hints for Architectural Principles for Non-CMOS Nano-Scale Implementations 29
2.7 Biological Relevance 30
References 32
3 Nanoantenna Infrared Detectors 34
3.1 Introduction 35
3.1.1 Project Overview 36
3.1.2 Infrared Detectors 37
3.1.2.1 Thermal Infrared Detectors 38
3.1.2.2 Quantum Infrared Detectors 40
3.1.2.3 Radiation-Field Infrared Detectors 41
3.1.3 Detector Characterization 41
3.1.3.1 Figures of Merit 42
3.1.3.2 Electrical Noise Considerations 43
3.1.3.3 Detector Comparison 45
3.2 Antenna-Coupled MOM Diodes 46
3.2.1 Dipole Antenna 47
3.2.2 MOM Diodes 48
3.2.2.1 MOM Diode Design 51
3.2.2.2 Point Contact MOM Diodes 52
3.2.2.3 Thin-Film MOM Diodes 53
3.2.3 Conduction Mechanisms 54
3.2.4 Substrate and Antenna Effects 59
3.3 Fabrication 61
3.3.1 Substrate 61
3.3.2 Bonding Pad Fabrication 61
3.3.3 ACMOMD Fabrication 62
3.3.3.1 Electron Beam Lithography 62
3.3.3.2 Metal Deposition 66
3.3.3.3 Packaging 70
3.4 Detector Characterization 70
3.4.1 Current-Voltage Characteristics 71
3.4.1.1 Air-Oxidation 72
3.4.1.2 Controlled Oxidation 75
3.4.2 Infrared Response Characteristics 76
3.4.2.1 IR Detector Characterization 78
3.4.2.2 Polarization Dependence 81
3.4.2.3 Antenna Length Dependence 83
3.5 Comparison to Current Technologies 85
3.5.1 Comparison with Currently Available IR Detectors 86
3.5.2 Integration with CMOS Imaging Chips 86
References 89
4 Memristors: A New Nanoscale CNN Cell 94
4.1 Introduction 94
4.2 Background Information on Memristors 95
4.2.1 HP Memristor 100
4.2.2 How to Read and Write Memory States 102
4.3 Memristive Devices and Systems 106
4.3.1 Potassium Memristor 111
4.3.2 Sodium Memristor 111
4.4 Lossless Nonvolatile Memory Circuit Elements 113
4.4.1 Memory Capacitor 113
4.4.2 Memory Inductor 117
References 121
5 Circuit Models of Nanoscale Devices 123
5.1 Introduction 123
5.2 Vacuum Fluctuations in Nanocircuits 126
5.3 Mixed Quantum Classical Electromechanical Models 127
5.4 Circuit Model of a Double-Band Infrared Sensor 130
References 132
6 A CMOS Vision System On-Chip with Multi-Core, Cellular Sensory-Processing Front-End 134
6.1 Introduction 134
6.2 Architectural Concept of the Eye-RIS System 136
6.3 The Eye-RIS Chip 140
6.4 The Eye-RIS' Front-End: The Q-Eye 141
6.5 The Eye-RIS Chip in Operation 143
6.6 Discussion 147
References 150
7 Cellular Multi-core Processor Carrier Chip for Nanoantenna Integration and Experiments 152
7.1 Introduction 152
7.2 Algorithmic Considerations 154
7.2.1 Numeric Precision 155
7.3 Architecture of the Nanoantenna Carrier Chip 155
7.3.1 High-Gain Sensor Readout Channel 156
7.3.2 Digital Processor Architecture 159
7.3.3 Partitioning 159
7.3.4 Control Processor 161
7.3.5 SIMD Processor Array 162
7.4 Nanoantenna Integration 163
7.4.1 Antenna Coupled Nanodiode Interfacing 163
7.4.2 Physical Integration of the Nanoantenna Array 165
7.5 Measurement Environment 166
7.6 Concluding Remarks 166
References 167
8 Circuitry Underlying Visual Processing in the Retina 168
8.1 Introduction 168
8.1.1 Background Circuit Organization 169
8.1.2 Extreme Complexity of Amacrine Cell Interactions 171
8.1.3 A Dozen Different Representations 171
8.1.4 Each of the Ganglion Cell Outputs Extends over a Specific and Different Space--Time Domain 173
8.1.5 Crossover Circuitry of Vertical Amacrine Cells Affects Bipolar Amacrine and Ganglion Cells 174
8.1.6 The Visual Functional Roles of Crossover Circuitry 176
8.1.6.1 Active Surround Mediated by Crossover Inhibition in Ganglion Cells 177
8.1.7 Crossover Inhibition Helps to Distinguish Brightness from Contrast (Molnar et al. 2008) 177
8.1.7.1 Crossover Inhibition Allows Neurons to Linearly Add Intensities Distributed Across the Receptive Field Center for Ganglion Cells (Molnar and Werblin 2007b) 177
8.1.7.2 In-layer Interactions are Mediated by GABAergic Pathways 179
8.1.8 Specific Ganglion Cell Circuitries 180
8.1.8.1 Directionally Selective Ganglion Cells 180
8.1.8.2 Alpha Ganglion Cells 181
8.1.8.3 Local Edge Detectors 182
8.1.8.4 ON Beta Cells 183
References 184
9 Elastic Grid-Based Multi-Fovea Algorithm for Real-Time Object-Motion Detection in Airborne Surveillance 186
9.1 Introduction 186
9.1.1 Unmanned Aerial Vehicles 186
9.1.2 Multi-Fovea Approach 187
9.1.3 Airborne Motion Detection 188
9.2 Independent Motion Analysis 189
9.2.1 Images and Video Frames 189
9.2.2 Background and Objects 189
9.2.3 Global Image Motion Model 190
9.2.4 Motion Detection, Object Extraction, and Global Background Mosaic 191
9.3 Multi-Fovea Framework: Abstract Hardware Model 192
9.4 Algorithms 194
9.5 Corner Pairing Algorithm 197
9.5.1 Block Matching Algorithms 197
9.5.2 KLT Algorithm 198
9.5.3 SIFT Algorithm 199
9.5.4 Global Registration-Based Detection 200
9.5.5 Elastic Grid Multi-Fovea Detector 202
9.6 Performance of Methods 204
9.6.1 Metrics for Quality 204
9.7 Comparison 206
9.8 Summary 208
References 217
10 Low-Power Processor Array Design Strategy for Solving Computationally Intensive 2D Topographic Problems 219
10.1 Introduction 219
10.2 Architecture Descriptions 220
10.2.1 Classic DSP-Memory Architecture 221
10.2.2 Pipe-Line Architectures 223
10.2.3 Coarse-Grain Cellular Parallel Architectures 225
10.2.4 Fine-Grain Fully Parallel Cellular Architectures with Discrete Time Processing 226
10.2.5 Fine-Grain Fully Parallel Cellular Architecture with Continuous Time Processing 227
10.3 Implementation and Efficiency Analysis of Various Operators 228
10.3.1 Categorization of 2D Operators 229
10.3.1.1 Execution-Sequence-Variant VersusExecution-Sequence-Invariant Operators 231
10.3.2 Processor Utilization Efficiency of the Various Operation Classes 233
10.3.2.1 Execution-Sequence-Invariant Content-DependentFront-Active Operators 233
10.3.2.2 Execution-Sequence-Variant Content-Dependent Front Active Operators 235
10.3.2.3 1D Content-Independent Front Active Operators (1D Scan) 236
10.3.2.4 2D Content-Independent Front Active Operators (2D Scan) 238
10.3.2.5 Area Active Operators 239
10.3.3 Multiscale Processing 239
10.4 Comparison of the Architectures 240
10.5 Optimal Architecture Selection 242
10.6 Conclusions 247
References 248
Index 250