Signal Processing for mmWave MIMO Radar

(1)

FACULTY OF ENGINEERING AND SUSTAINABLE DEVELOPMENT

Signal Processing for mmWave MIMO Radar

Óscar Faus García June 2015

Master’s Thesis in Electronics

Master’s Program in Electronics/Telecommunications

(2)

(3)

Preface

I would like to thank my advisor, André Bourdoux, for his supervision, valuable guidance and helpful suggestions during my 6 months of internship in IMEC. Having the opportunity to take part in such an interesting and innovative project like the mmWave radar has been a valuable and enriching experience for me. I am also grateful to Wim Van Thillo for his support during my arrival to IMEC.

I also want to thank all the colleagues in the mmWave group and in the wireless communication department who at some point have helped me with any issue and whose comments and discussions have been highly valuable. I am also grateful to IMEC for providing me with the facilities and means to develop my thesis and to KU Leuven for their help and support during my international scholar stay in Leuven.

While many people have contributed to my experience during my thesis, I would like to thank a few people in particular; Bilal, Georgi, Nico and Yi, for sharing these intense months with me and for offering their professional and personal friendship. It was truly a great experience to share all that time in IMEC, Wisteria and ultimately in Belgium; I am very grateful and I believe that the best is yet to come.

I am especially grateful to Georgi G. for his help, support and suggestions during the writing of my thesis, and also to Hou Yi for his special life approach that so many good fun moments has brought among us.

Also, I want to thank Darya for her help in important moments during these last months. I would also like to thank my family, and specially my parents and sister, for the unconditional support they have always sent me from Spain during all these years that I have been abroad in Sweden, Belgium and Poland. Gracias.

(4)

(5)

Abstract

This thesis addresses the design study, implementation and analysis of signal processing algorithms for a 79 GHz millimeter-wave Phase Modulated Continuous Wave (PMCW) Multi Input Multi Output (MIMO) short range radar; performed in IMEC research institute (Leuven, Belgium). The radar system targets high resolution performance with low power consumption in order to integrate a full MIMO radar transceiver with digital processor and antennas in a compact package featuring a size of 1 cm². Achieving such radar system characteristics requires the exploitation of a highly demanding digital architecture with signal processing gain and high range, speed and angle resolutions. The improved resolution and detection capabilities will be achieved by performing signal processing algorithms on the reflected waveform. The digital front-end implements parallel range gate processing with a bank of correlators that perform: pulse compression, coherent accumulation to further increase Signal to Noise Ratio (SNR) and N-point FFT to extract the Doppler information. The use of MIMO is proposed implementing a code domain technique in the PMCW waveform, the Outer Hadamard Code MIMO. This concept makes use of a unique sequence for all the transmitting antennas that is rendered by an outer sequence to ensure the orthogonality of the transmitted waveforms. The outer code makes use of the good cross-correlation properties of the Hadamard sequences and the waveform uses sequences that exhibit perfect auto-correlation profile, the Almost Perfect Autocorrelation Sequences (APAS). The MIMO implementation results in higher angular resolution and extra processing gain.

The use of beamforming techniques in the radar allows the angle estimation of the detected targets;

using rough and fine beamforming that provides with coarse and precise Angle of Arrival (AoA) estimation in an early and late stage respectively. A Constant False Alarm Rate (CFAR) processing stage is implemented in the stage of the system where higher signal processing gain is achieved. This algorithm allows the variation of the CFAR parameters and analyzes the detections in order to improve the probability of detection (𝑃_𝑑) while decreasing the probability of false alarm (𝑃_𝑓𝑎). A series of simulations with different scenarios and variable parameters are set in order to analyze the performance of the system. The simulations analyze the gain achieved in each stage and their outcomes show an impressive processing gain that can reach SNR improvements as high as 77 dB for a small virtual array while keeping the 𝑃_𝑓𝑎 low with the CFAR adjustment. The use of bigger arrays demonstrates the possibility to enable clear detections for low Radar Cross Section (RCS) targets in far distances of the unambiguous range. The use of beamforming shows interference reduction improvement as the beam widths narrow with the increasing number of virtual array antennas. These results have been achieved while keeping the system design parameters to a range resolution of 7.5 cm for a maximum range of 37.5 meters with speed resolution of 0.2 m/s and a maximum detectable speed of 12.66 m/s. The outcomes support the good performance of the signal processing techniques implemented and the benefits in applying them in a SoC mmWave MIMO radar.

(6)

(7)

Table of Figures

Fig. 1: Maxwell electromagnetism equations... 1

Fig. 2: Telemobiloscope [1]. ... 2

Fig. 3: Radar basic operation principle. ... 5

Fig. 4: Scheme of a bistatic radar design. ... 6

Fig. 5: Scheme of a monostatic radar design. ... 6

Fig. 6: Block diagram of a monostatic architecture pulsed radar system. ... 8

Fig. 7: Block diagram of the architecture of a FMCW radar. ... 9

Fig. 8: Spectrum with the location of the traditional microwave bands and the mmWave bands. ... 11

Fig. 9: Attenuation in the atmosphere of electromagnetic waves according to their working frequency. ... 12

Fig. 10: Radar short and long range types with the mmWave frequencies and capabilities. ... 13

Fig. 11: Basic block diagram of a general monostatic radar system [37]... 18

Fig. 12: Two targets closely placed being radiated by a radar. ... 20

Fig. 13: Spectrum of a pulse of duration τ, with a sinc shape [26]. ... 21

Fig. 14: Several transmitted pulses separated in time by the PRI. ... 22

Fig. 15: Two unambiguous and one ambiguous mesuarement [9]. a)Transmitter: three pulses with a pulse width 𝜏𝑝 and PRI=Tp.b) Receiver: Echoes caused by the first pulse. ... 23

Fig. 16: A basic FMCW radar block diagram. ... 27

Fig. 17: Representation of the FMCW transmitted and reflected signals [9]. ... 28

Fig. 18: Representation of the final FMCW signal in the output mixer [9]. ... 29

Fig. 19: Block diagram architecture of a basic pseudo-noise modulated radar system. ... 30

Fig. 20: Time representation of a biphase sequence with its parameters [9]. ... 30

Fig. 21: An M-Sequence representation in the frequency domain [9]. ... 31

Fig. 22: A pulse being spread through time in order to reduce to peak energy and keep the BW... 33

Fig. 23: Biphase code diagram of pulse compression. ... 33

Fig. 24: A 13-Barker code and its respective autocorrelation function . ... 35

Fig. 25: Autocorrelation response of a m-sequence code with 1023 chips length. ... 36

Fig. 26: APAS code processing of its autocorrelation function [36]. ... 37

Fig. 27: Autocorrelation response of an APAS code of 4080 chips length. ... 37

Fig. 28: Coherent and non-coherent signals. In a) a coherent signal generated from the reference b), in c) a non-coherent pulse. ... 42

Fig. 29: After the Doppler processing the matrix transforms to range/Doppler from fast/slow time to matrix [33]. ... 45

Fig. 30: Shape of the received signal after DFT processing [33]. ... 47

Fig. 31: Thumbtack-like ambiguity function [33]. ... 49

Fig. 32: Basic block diagram of the signal processing in the receiver until the detection stage. ... 50

Fig. 33: Cell analysis procedure of a CFAR analysis. ... 52

Fig. 34: Block diagram of the CA CFAR detector. ... 54

Fig. 35: CA CFAR masking occurring in a multi-target situation with N=32 and 𝑃𝑓𝑎 = 10 − 6 [46]. ... 55

Fig. 36: MIMO radar system principle. ... 59

Fig. 37: Three transmitter antennas and four receiver antennas with their matched filters detectors. ... 61

(12)

Fig. 38: NM virtual antenna arrays formed with the physical antennas from Fig. 36. ... 61

Fig. 39: Transmission part for the signal generation of a MIMO outer code system. ... 64

Fig. 40: MIMO processing for one range gate in one receiver antenna. ... 64

Fig. 41: Transmitter side of the Range Domain Separation MIMO method with four transmitters. ... 65

Fig. 42: Receiver side of the Range Domain Separation MIMO method showing one receiver antenna. ... 66

Fig. 43: Geometry of a Uniform Linear Array (ULA)... 68

Fig. 44: Conventional Beamformer basic block diagram. ... 70

Fig. 45: Adaptive Beamformer basic block diagram. ... 72

Fig. 46: 3D simulation of the targeted SoC radar, featuring the size of 1 𝑐𝑚2 [9]. ... 75

Fig. 47: Chip duration for each sequence 𝐿𝑐 symbol. ... 77

Fig. 48: Unambiguous range detection field of a radar system. ... 78

Fig. 49: One PMCW sequence with a certain chip period. ... 83

Fig. 50: Relation between the unambiguous Doppler frequency and the dwell time. ... 85

Fig. 51: Relation between the number of N and the narrowing of the filter in the Doppler bank filter. ... 85

Fig. 52: System design structure of the mmWave radar project. ... 86

Fig. 53: Basic block diagram of the PMCW radar system under study. ... 87

Fig. 54: Repetition of the transmitted sequence, S, a number of M times. ... 88

Fig. 55: Radar system transmitting the needed N times M sequences repetition. ... 88

Fig. 56: Digital front-end with a bank of parallel correlators for PMCW radar system [20]. ... 89

Fig. 57: Transmitter part of a radar system reaching two targets at different range gates. ... 90

Fig. 58: Receiver part of a radar system echoing the signals from the two targets of the Fig. 56. ... 91

Fig. 59: Complete parallel range gate processor block diagram of a PMCW radar. ... 92

Fig. 60: Detailed parallel gate processing for the example of the Fig. 58. ... 93

Fig. 61: Matlab chain PMCW radar simulation flow. ... 95

Fig. 62: Two dimensional model of the SISO data structure, a radar data matrix. ... 98

Fig. 63: Block diagram of the digital front-end processing a single range gate and saving its samples in each component of the vector that composes the first row of the radar data matrix. ... 99

Fig. 64: Digital front-end processing and storing of samples for the full parallel range gate system. ... 100

Fig. 65: Conversion of the fast/slow time radar data matrix to the range/Doppler radar data matrix. ... 101

Fig. 66: Location of the MIMO processing part in the radar system. ... 103

Fig. 67: Three dimensional model of the MIMO data structure, a radar data cube. ... 104

Fig. 68: A MIMO radar system with N_tx=4 and N_rx=2 and its correspondent virtual array. ... 105

Fig. 70: Simplified diagram of the signals the TX signals being summed and reflected. ... 106

Fig. 69: Forming of the transmitter side of an Outer Hadamard Code MIMO radar system. ... 106

Fig. 71: Receiver side of an Outer Hadamard Code MIMO radar system. ... 107

Fig. 72: Data structure size that each receiver will take after the MIMO processing. ... 108

Fig. 73: Separation of each transmitter signal part with a Hadarmard matrix and radar data cube storage. ... 109

Fig. 74: Separation of each transmitter signal part for every N repetitions and radar data cube storage. ... 109

Fig. 75: Separation of each transmitter signal part for every N repetitions with the full parallel range gate processing besides its storing in a radar data cube. ... 110

Fig. 76: Separation of each received signal for every N repetitions with the full parallel range gate processing and for each receiving antenna, besides its storing in a radar data cube... 111

(13)

Fig. 77: Signal forming of the transmitter side of a Range Domain Separation MIMO radar system. ... 112

Fig. 78: Range Domain MIMO receiver after having correlated a full sequence Lc, giving as a result the unambiguous range of each antenna. ... 113

Fig. 79: Range Domain MIMO receiver detailed process after having correlated a full sequence Lc, giving as a result the unambiguous range of each antenna. ... 114

Fig. 80: Detections of each antenna with a Range Domain Separation MIMO processor. ... 114

Fig. 81: One dimension matrix with the information related to each antenna with length 𝐿𝑐 x M x N. ... 115

Fig. 82: Radar data cube after the reorganization of the one dimensional receiver output with dimension 𝐿𝑐 x 𝑁𝑣 x N. .... 116

Fig. 83: Range Domain Separation structure adjustment with m-sequences. ... 116

Fig. 84: Location of the CFAR processing part in the radar system. Upper figure for a SISO radar configuration, lower figure for a MIMO radar configuration. ... 117

Fig. 85: Data structures in the input of the CFAR detector stage. Left figure, a SISO matrix filled with data, right figure a MIMO data cube structure. ... 118

Fig. 86: Difference in the profile views between the Doppler domain and the Range domain... 119

Fig. 87: CFAR analysis along the range domain for each doppler bin. ... 120

Fig. 88: CFAR analysis along the range domain for each doppler bin. ... 121

Fig. 89: Detection in the radar data matrix and saving of the detected’s coordinate in the detection matrix. ... 122

Fig. 90: Block diagram of the CA/GOCA/SOCA CFAR algorithms. ... 123

Fig. 91: Block diagram of an OS CFAR algorithm. ... 123

Fig. 92: Final detection matrix with several detections. ... 124

Fig. 93: False alarms in consecutive Doppler bin and range gate clearance algorithm. No consecutive/adjacent found. ... 125

Fig. 94: False alarms in consecutive Doppler bin and range gate clearance algorithm. Range migration found. ... 126

Fig. 95: False alarms in consecutive Doppler bin and range gate clearance algorithm. Doppler spreading and range migration found. ... 127

Fig. 96: Final detection matrix result after the post-detection processing to eliminate false alarms. ... 128

Fig. 97: Rough radar scanning along the equidistant angles of the radar aperture, for 4 antennas. ... 130

Fig. 98: First angle under evaluation in beamforming. ... 131

Fig. 99: Beamforming using the radar data cube 𝐿𝑐 𝑥 𝑁𝑣 𝑥 𝑁 and a weighting vector of certain angle, besides new radar data cube structure with the angles dimension. ... 132

Fig. 100: Vector beamforming using the radar data cube 𝐿𝑐 𝑥 𝑁𝑣 𝑥 𝑁 and a weighting vector of certain angle, besides a new radar data cube structure with the angles dimension. ... 133

Fig. 101: Angles’ radar data cube completed for all the angles of the rough beamforming ... 134

Fig. 102: Beamforming completed for the first angle of the rough beamforming. ... 134

Fig. 103: CFAR detector along each matrix linked with an angle... 135

Fig. 104: Consecutive angles' radar data matrix analysis. ... 135

Fig. 105: Target causing a double detection in different angles. ... 136

Fig. 106: Location of the fine beamforming processing part in the radar system. ... 137

Fig. 107: Fine beamforming around a rough beamforming wide lobe linked to a target in a detected angle. ... 138

Fig. 108: Data sets selection from the radar data cube using the CFAR detection coordinates. ... 138

Fig. 109: Beamforming evaluation in different angles around the angle where the detection 1 was located with the rough beamforming. ... 139

Fig. 110: Block diagram of the transmitter part of the PMCW radar system, SISO and MIMO configurations. ... 142

(14)

Fig. 111: Block diagram of the receiver side of the PMCW radar system. ... 143

Fig. 112: Simulation 1, detection in the range – Doppler domain 3D representation of the radar data cube. ... 149

Fig. 113: Simulation1, detection showed in the range profile of the radar data cube taken before the FFT processing. ... 150

Fig. 114: Simulation 1, detection in the Doppler profile of the radar data cube taken after the FFT processing. ... 151

Fig. 115: Simulation 1, detection cuts in the Doppler domain cuts and the range domain cuts where there have been detections. ... 151

Fig. 116: Simulation 1, Detection matrixes; after CFAR detection(left) and after post-processing algorithm(right). ... 154

Fig. 117: Simulation 2, detections of the simulation 2 showed in the detection matrixes. ... 158

Fig. 118: Simulation 2, detections showed in Range and Doppler cuts. ... 158

Fig. 119: Simulation 2, detections of the targets of the simulation 2 in Range and Doppler profiles. ... 159

Fig. 120: Simulation 2, detections over the system noise floor for 𝑃𝑓𝑎 = 10 − 4(left) and 𝑃𝑓𝑎 = 10 − 2 (right). ... 160

Fig. 121: Simulation 2, detection matrixes of the detections for 𝑃𝑓𝑎 = 10 − 4(left) 𝑃𝑓𝑎 = 10 − 2 (right). ... 161

Fig. 122: Simulation 2, Detection profiles of the detections in Simulation 2 for 𝑃𝑓𝑎 = 10 − 4 (up) and 𝑃𝑓𝑎 = 10 − 2 (down). ... 162

Fig. 123: Simulation 2, detection profiles of the targets with RCS=-20 dBsm, with 𝑃𝑓𝑎 = 10 − 6 (up) and 𝑃𝑓𝑎 = 10 − 4 (down). ... 163

Fig. 124: Simulation 3, detection profile of a CA CFAR algorithm for 𝑁𝑡𝑐=2 (up), 𝑁𝑡𝑐 =10 (middle) and 𝑁𝑡𝑐=16 (down). .. 164

Fig. 125: Simulation 3, Doppler profile with a target causing range migration. ... 165

Fig. 127: Simulation 3, detection matrix of a range migrated detection with 0 guard cells (left) and 1 guard cell (right) ... 166

Fig. 126: Simulation 3, detection profiles varying the number of guard cells, from 0(up) to 8(down). ... 166

Fig. 128: Simulation 3, CA algorithm applied for two contiguous detections with 𝑁𝑡𝑐= 12 (left) and 𝑁𝑡𝑐= 20 (right). ... 167

Fig. 129: Simulation 3, GOCA algorithm applied for two contiguous detections with 𝑁𝑡𝑐= 12 (left) and 𝑁𝑡𝑐= 20 (right). .... 167

Fig. 130: Simulation 3, SOCA algorithm applied for two contiguous detections with N_tc= 12 (left) and N_tc= 20 (right). .. 168

Fig. 131: Simulation 3, OS algorithm applied for two contiguous detections with 𝑁𝑡𝑐= 12 (left) and 𝑁𝑡𝑐= 20 (right). ... 169

Fig. 132: Simulation 4, targets in the range – Doppler domain 3D representation of the radar data cube. ... 172

Fig. 133: Simulation 4; range profile of the received signal after the pulse compression. ... 173

Fig. 134: Simulation 4, range profile after M accumulations and MIMO, before the FFT processing. ... 173

Fig. 135: Simulation 4, Doppler cuts of the radar data cube after the FFT processing. ... 174

Fig. 136: Simulation 4, Doppler cut of the radar data cube after the rough beamforming processing. ... 175

Fig. 137: Simulation 5, range profile after pulse compression. ... 178

Fig. 138: Simulation 5, range profile after the pulse compression and M accumulation. ... 178

Fig. 139: Simulation 5, Doppler profile of the radar data cube after the FFT. ... 179

Fig. 140: Simulation 5, range profile of the radar data cube after the FFT processing. ... 180

Fig. 141: Simulation 5, 3D representation of the radar data cube data after the rough beamforming processing. ... 180

Fig. 142: Simulation 5, range profile of the radar data cube after the rough beamforming processing. ... 181

Fig. 143: Simulation 5, CFAR profile cuts of the detections. ... 182

Fig. 144: Simulation 5, detection matrixes. ... 183

Fig. 145: Simulation 6, Doppler profile of the radar data cube after the FFT. ... 185

Fig. 146: Simulation 6, Doppler profile of the radar data cube after the rough beamforming. ... 185

Fig. 147: Simulation 6, targets in the range – Doppler domain 3D representation of the radar data cube. ... 186

Fig. 148: Simulation 6, range profile of the radar data cube after the rough beamforming processing. ... 186

Fig. 149: Simulation 6, targets in the range – Doppler domain 3D back representation of the radar data cube. ... 187

(15)

Fig. 152: Simulation 7, angle of arrival cut for the Target 1. ... 193

Fig. 155: Simulation 7, Doppler cuts for the Target 1, angle 12 degrees. ... 195

Fig. 157: Simulation 7, Doppler cuts for the Target 3, angle -36 degrees. ... 196

Fig. 158: Simulation 7, target 3 in the range – Doppler domain 3D for the angle = -36 degrees. ... 197

Fig. 162: Simulation 7, range profile for the angle 30. ... 200

Fig. 163: Simulation 8, range profile for the angle 30. ... 201

Fig. 164: Simulation 7, Bartlett beamforming algorithm applied for the detected targets. ... 202

Fig. 165: Simulation 7, Capon beamforming algorithm applied for the detected targets. ... 203

Fig. 166: Simulation 7, MUSIC beamforming algorithm applied for the detected targets. ... 203

Fig. 167: Simulation 8, Bartlett beamforming algorithm applied for the detected targets. ... 204

Fig. 168: Simulation 8, Capon beamforming algorithm applied for the detected targets. ... 205

Fig. 169: Simulation 8, MUSIC beamforming algorithm applied for the detected targets. ... 205

Fig. 170: Simulation 9, CFAR profile cuts for the detections. ... 209

Fig. 172: Simulation 9, CFAR profile cuts for the detections with range resolution decreased... 213

Fig. 173: Simulation 9, detection matrixes for the range resolution decreased. ... 214

Fig. 174: Simulation 9, detection matrixes for a decreased velocity resolution. ... 218

Fig. 175: Simulation 9, CFAR profile cuts for the detections with unambiguous velocity decreased. ... 221

Fig. 176: Simulation 9, detection matrixes for unambigous velocity decreased. ... 222

Fig. 177: Simulation 9, detection matrixes for increased unambiguous speed. ... 224

Fig. 178: Simulation 10, CFAR profile cuts for the detections. ... 227

(16)

List of Tables

Table 1: Summary of the main PMCW radar system parameters. ... 76

Table 2: Final Values of the Parameters after recalculation. ... 81

Table 3: Some of the functionalities that are included in the digital front end and back end of the radar system. ... 87

Table 4: Matlab chain parts and main functionalities. ... 94

Table 5: Simulation 1, targets scenario initialization parameters. ... 146

Table 6: Simulation 1, main design parameters chosen for the radar system under simulation. ... 146

Table 7: Simulation 1, theoretical PMCW link budget of the targets. ... 147

Table 8: Simulation 1, summary of the main PMCW system parameters. ... 148

Table 9: Simulation 1, CFAR Performance. ... 153

Table 10: Simulation 1, detected targets coordinates and interpretation. ... 153

Table 11: Simulation 1, radar system performance at different stages of simulation, theoretical and simulated. ... 155

Table 12: Simulation 2, targets scenario initialization parameters. ... 156

Table 14: Simulation 2, values of the CFAR parameters chosen. ... 157

Table 15: Simulation 2, CFAR performance. ... 157

Table 18: Simulation 3, new targets location to show the variation in training cells effect. ... 163

Table 20: Simulation 4, selectable parameters of the MIMO stage. ... 171

Table 22: Simulation 5, simulated targets initialization parameters. ... 176

Table 24: Simulation 5, MIMO configuration. ... 176

Table 31: Simulation 6, radar system performance at different stages of the system. ... 184

Table 33: Simulation 6, CFAR detected targets coordinates and interpretation. ... 188

Table 39: Simulation 7, CFAR detected targets coordinates and interpretation. ... 198

(17)

Table 43: Variable PMCW system design parameters. ... 206

Table 44: Simulation 9, simulated targets initialization parameters ... 206

Table 45: Simulation 9, decreased unambiguous range variation... 207

Table 46: Simulation 9, summary of main PMCW system parameters for decreased unambiguous range. ... 208

Table 47: Simulation 9, theoretical PMCW link budget for the targets scenario. ... 208

Table 50: Simulation 9, increased unambiguous range. ... 210

Table 51: Simulation 9, summary of main PMCW system parameters for increased unambiguous range. ... 211

Table 52: Simulation 9, decreased range resolution. ... 212

Table 53: Simulation 9, summary of the main PMCW system parameters for decreased range resolution. ... 212

Table 54: Simulation 9, detected targets coordinates and interpretation for a decreased range resolution. ... 213

Table 55: Simulation 9, increased range resolution. ... 214

Table 56: Simulation 9, summary of the main PMCW system parameters for increased range resolution. ... 215

Table 57: Simulation 9, detected targets coordinates and interpretation for a increased range resolution. ... 215

Table 58: Simulation 9, decreased velocity resolution. ... 216

Table 59: Simulation 9, summary of the main PMCW system parameters for decreased velocity resolution... 216

Table 60: Simulation 9, detected targets coordinates and interpretation for decreased velocity resolution... 217

Table 61: Simulation 9, increased velocity resolution. ... 218

Table 62: Simulation 9, summary of the main PMCW system parameters for increased velocity resolution. ... 219

Table 63: Simulation 9, detected targets coordinates and interpretation for increased velocity resolution. ... 219

Table 64: Simulation 9, decreased unambiguous speed. ... 220

Table 65: Simulation 9, summary of the main PMCW system parameters for decreased unambiguous velocity. ... 220

Table 66: Simulation 9, detected targets coordinates and interpretation for decreased unambiguous velocity. ... 221

Table 67: Simulation 9, increased unambiguous velocity. ... 222

Table 68: Simulation 9, summary of main PMCW system parameters for increased unambiguous velocity. ... 223

Table 69: Simulation 9, detected targets coordinates and interpretation for increased unambiguous velocity. ... 223

(18)

(19)

1 Introduction

In this chapter, an historical overview of the radar technology from its beginnings through its evolution is presented. The chapter also reviews the basic concepts of radar; the main functions and characteristics, the main classification groups in which the different radar systems are included, the frequency bands where the different types of radar work and some applications. A background of automotive radars is also presented; the technology motivation and development during the last years, the technologies used with their advantages and drawbacks besides future needs. An introduction to the mmWave radar that this thesis focuses in is also provided, besides a description of the aim of the thesis. Finally, in the last part of the chapter an outline of the thesis is included, where the content that each chapter develops is described.

1.1 Historical Overview

The origin of radar cannot be attributed to any particular person or country in particular. The development of the technology was pushed by certain circumstances that lead to its appearance. The

‘discovery’ was boosted by several factors and actors which contribution summed up to the materialization of the final product. However there are some milestones and historical moments that are worth mentioning since they added significant and relevant tools and inventions to the process.

In 1865, a Scottish physicist named James Clerk Maxwell, introduced a relevant and very important theory for the future development and growth of wireless technologies included in the radar. ‘A Dynamical Theory of the Electromagnetic Field’ was the third article that Maxwell published regarding electromagnetism [1]. This paper is of vital importance since it introduced the famous set of equations shown in Fig. 1, today referred as ‘the four Maxwell’s equations’; that describe how the magnetic and electric fields are related to each other. The basic demonstration of this theory was the fact that the magnetic and electric fields are moving through the space as waves and with a constant speed.

∇ ∙ 𝐸 = ^𝜌

𝜀0

∇ ∙ 𝐵 = 0 ∇ × E = −^𝜕𝐵_𝜕𝑡

∇ × 𝐵 = 𝜇₀𝐽 + 𝜇₀𝜀₀^𝜕𝐸

𝜕𝑡

Fig. 1: Maxwell electromagnetism equations.

(20)

These Maxwell’s papers gave birth to the use of electromagnetism as we know it today. Thanks to these foundations was possible for the next vital actor to appear. A further improvement of electromagnetism theory was developed by Heinrich Rudolf Hertz, a German physicist, who in the late 19^th century demonstrated that the electromagnetic waves are reflected by metallic surfaces [2]. He studied the response to electromagnetic radiation, soon to be called radio waves, of different conductors of dielectrics, demonstrating in fact what was theorized by the Maxwell’s equations.

Another important milestone was achieved by Guglielmo Marconi. After conducting some experiments, consisting on tests transmitting waves throughout the countryside, he indicated the potential use of the technology in order to help ships avoid coastline points in bad weather conditions.

The achievement of his long distance electromagnetic transmissions [3] and his presented ideas to apply the technology were the key kick off for a wide interest in the development of the technologies that will later come out as the so called radar.

In 1904, Christian Hülsmeyer started to give demonstrations of a device characterized by the use of the echos produced by radio waves [4]. The invention was called ’telemobiloscope’(Fig. 2) and consisted on a gap where a signal was generated using a dipole antenna with a reflector, as the transmitter.

The reflected echoes were taken by a similar receiver coherent antenna and when a detection happened a bell ringed. The device was very useful for sea traffic control thanks to its capable range capability covering up to 3000 meters. The device did not provide range information yet, just the warning of a nearby object, but it was the first practical invention applying the work of G. Marconi.

A parallel invention of significant importance happened in 1921. The Magnetron emerged as an efficient transmitter invented by the Albert Wallace Hull [5]. The discovery took place when A. Hull was researching the movement of electrons under the influence of a magnetic field and he realized that there was a possibility of controlling the electron current by varying the produced magnetic field.

Fig. 2: Telemobiloscope [1].

(21)

Almost at the same time as Hull’s discovery, Albert Hoyt Taylor and Leo C. Young were conducting transmitting experiments in the US Naval Aircraft Radio Laboratory, when they realized they had an interference in the system, which was actually a detected wooden vessel. Later in 1930, Lawrence A.

Hyland decided to use an analogous arrangement with purpose of detecting an airplane. This directed to a patent proposition [6] to use the technique to detect boats and aircrafts for what it would now be called CW radar (Continuous Wave Radar).

By that time, the invention could detect the existence of an object, but could not determine its velocity, nor its location. Leo C. Young suggested the use of pulsing techniques to overcome this limitation.

Robert Morris Page implemented Young’s idea and in 1934 his invention was tested to detect a plane.

Even though the results were poor they were good enough to show the potential of the technology and resulted in the world first real radar device.

In 1936 William C. Hahn and George Metcalf developed an important component which will be useful for the radars, as an oscillator and amplifier, the Klystron. During this time period important efforts and investigations were put in place to further develop the technology, mainly in countries such as United Kingdom, Germany, US, URSS and Japan.

The British, directly threaten and facing possible attacks coming by air, developed the famous Chain Home surveillance radar network that remained working until the end of the World War II. In 1939, after numerous rudimentary tests and developments in United Kingdom, the engineers John Turton Randall and Henry Albert Howard Boot constructed a radar using a multicavity magnetron [7]. The device was light and high powered due to its short wavelength and suitable to be mounted on aircrafts to effectively German submarines during night or with bad visibility conditions.

After the beginning of the Second World War (WWII), in 1940, the US and United Kingdom started a cooperation to further upgrade the radar technology. Until then, most of the existing radar systems were functioning at HF (High Frequency) and VHF (Very High Frequency) wavelengths, but with the progresses of the microwave magnetron by the British and the creation of the Radiation Laboratory in the MIT (Massachusetts Institute of Technology), the radar based on microwave frequencies was settled and remained dominant until nowadays.

The development of the radar technology was guided by the war defense’s needs, while currently military applications are still the main market and reason for development, the use of radar has been widely extended to a number of varied civilian applications. The most known might be the traffic radars, used to measure the speed of cars, or the meteorological radars with which the weather forecasts are shown in TV news. Another radar very important for nowadays functioning of the transport is the air traffic control radar, used to control the aircraft traffic over the congested air space.

Aviation also uses radars for determining speed and avoiding bad weather conditions. One of the late developments for which radar is being used for and where many efforts and investment are being put, is in the scope of the automotive industry, for aid driving and autonomous cars. Ultimately, it is foreseen that their miniaturization, radars will be used in small devices making up the Internet of

(22)

Things. It is precisely in these latest applications in the medium term and automotive applications in the short term, where the radar system presented in the following document chapters is intended to be used in.

1.2 Radar in Brief

The word Radar stands for RAdio Detecting and Ranging, and it was given by Samuel M. Tucker and F. Furth in 1940 [8]. It describes the electronic systems that are able to detect the presence and speed of one or several objects, by means of electromagnetic waves. The fundamentals of radar can be seen as an analogy to the reflection caused by a sound wave. When a shout is generated a series of echoes can be received back as a result of reflections from objects surrounding the origin. Moreover, the distance of the reflecting object can be easily computed, by knowing the speed of the sound and the coefficient of the medium through which the sound is being propagated.

Radar, instead of sound waves, uses electromagnetic waves. Radar can be defined as an electromagnetic sensing device for locating and derive further information regarding the detected object, through processing and analysis of the information contained in the signals coming back from the reflecting objects. The summarized radar operation, shown in Fig. 3, is outlined as follows:

- Initially the radar transmitter, by means of an antenna, radiates energy in space, in the shape of electromagnetic waves

- Through space, electromagnetic waves might come across with an object, called target. Depending on the reflectivity of the target, a portion of the energy will be reflected back.

- The electromagnetic energy hitting the target, is reflected in multiple directions.

One of the directions comes to be the same direction of the initially radiated electromagnetic wave. This energy is received by the radar receiver antenna.

- After an amplification and with the assistance of convenient signal processing a decision has to be made in whether there is a detection or not in the signal received.

- After the decision is made, other information about the target can be extracted for further analysis and processing.

(23)

The basic operation principles of radar systems are based on the physics of electromagnetism mechanisms. Basically:

- The reflection of electromagnetic waves: the energy is reflected when the wave front come across with an electrically conductive surface.

- Electromagnetic waves move through the space at a constant velocity, roughly the speed of light, 𝑐 = 3 ∙ 10⁸ 𝑚/𝑠. Due to the fact that the speed is constant, the distance between the reflecting targets and the signal source can be measured.

- The waves travel in a straight line, unless slight deviations due to atmospheric conditions, meaning that with special antenna arrangements and processing, the energy can be centered into a desired direction. So that way, also the direction of the detected targets can be measured.

Radars main functionalities are; the detection, ranging, speed measurement and angle location:

Detection and Ranging: It is one of the main features of a radar and describes its ability to determine the distance from a target. As mentioned before, this is done by measuring the time that the electromagnetic wave reflected back to the receiver takes from the moment it has left the transmitter.

Depending on the distance at which the target is located and other parameters (such as the pulse duration), the precision of the range can be as good as few centimeters.

To be able to measure the time difference, there has to be included a tool that will allow the receiver to differentiate the wanted signal from clutter and/or other undesired detections. This timing label can be thought in several ways: from a determined length pulse, a phase modulation or a frequency modulation. The main parameter for a good ranging precision will be the bandwidth, as it will be explained in Chapter 2.

Fig. 3: Radar basic operation principle.

Radar Antenna

Echo from Target

Transmitted Signal

(24)

Velocity: Radars can also be used to obtain the information concerning the speed of a moving target.

This information can be gathered from the distance variation over certain duration of time. Another used method is by means of the changes in the measurement of the Doppler frequency. The Doppler frequency accuracy will vary depending on the time of observation; meaning that an accurate frequency Doppler measurement will require long integration time. This concept will be explained in more detail in Chapter 2.

Angle of arrival: Another relevant parameter that will be used to improve the characteristics of the presented radar system is the angle. Using the angle information embedded in the reflected waves will allow us to determine the direction to a target by doing a sweep along the sight of the antenna. Further explanation will be provided in the Chapter 2 of the present document.

Even though there is not an official taxonomy of the existing radar technologies, this paragraph will present the main groups and their characteristics:

Monostatic/Bistatic: Firstly, there can be distinguished different radar architectures, the monostatic and bistatic. The difference between them can be found in the design of the transmitter and receiver antennas sides of the system. While the bistatic radars are characterized by a design where the transmitter and receiver antennas are separated spatially (Fig. 4 [9]), the monostatic architecture is based on the transmission and reception of the signal in the same antenna, without any spatial separation. This is usually performed with a circulator, as shown in Fig. 5 [9].

The radar systems can also be classified according to the way they radiate the electromagnetic waves, basically in the pulse radars or continuous wave radars (CW).

Pulse radar: This type of radar radiates multiple and repetitive short and high energy pulses modulated in order to be able to also obtain the speed information out of the Doppler shifts. The

The radars can also be classified according to the way they radiate electromagnetic waves, in the pulse radars and continuous wave radars (CW).

Fig. 4: Scheme of a bistatic radar design. Fig. 5: Scheme of a monostatic radar design.

(25)

Pulse radar: This type of radar radiates multiple and repetitive short and high energy pulses modulated in order to be able to also obtain the speed information out of the Doppler shifts. The method is method is characterized by the shortness of its transmission pulses, which are usually in the order of the microseconds. After a pulse has been transmitted there is a long period of lack of transmission (T >> pulse length). This period is typically named ‘receiving time’, since the length of this silent time interval will be used to interpret the time between the transmitted pulse and reflected one, thus, giving the range unambiguously. In this type of radar is important to consider the control of the timing since the correct calculation of the range depends on the time accuracy calculation.

The pulse is going to be targeting one particular direction, defined by the directivity of the antenna, at each given moment. The distance from the target can be easily measured in an oscilloscope, evaluating the time between the two observed pulses in the screen (as far as there have not been any other transmitted pulses in the meantime). The traveling of the pulse is a two way trip, thus, the time measured must be divided by two in order to obtain the one way time interval that the electromagnetic wave needed to reach the object. Leading to be the following simple equation [33]:

𝑅 =𝑐_0∙𝑡

2 (1)

Derived from,

𝑣 =^𝑠

𝑡 and 𝑡 =^2∙𝑅

𝑐0 (2)

Where

R = distance

𝐶₀= vacuum speed of light v = the speed

t = time interval

Therefore, once obtained the half round trip time t/2, simply by a multiplication by the speed of light the distance is easily computable. Fig. 6 shows the block diagram of a pulsed radar system, in this case with a monostatic architecture.

(26)

A wave signal is generated with the local oscillator (LO) with certain constant frequency 𝑓₀ and, during transmission, radiated from the antenna, then, the received RF signal is processed at an intermediate frequency IF by mixing it with the LO frequency. The name “pulsed” is given because of the modulation induced by the switch in series with the antenna.

The radiated signal can be defined according to the following waveform formula [37]:

𝑠(𝑡) = 𝐴(𝑡) ∙ sin [2𝜋𝑓(𝑡) ∙ 𝜑(𝑡)] (3)

Where A(t) is the square wave signal shape varying according to the time, will be logical one in the short time when there is a transmission and zero otherwise (in echoes’ waiting mode) and 𝑠𝑖𝑛 [2𝜋𝑓(𝑡) ∙ 𝜑(𝑡)] is the carrier sine wave generated by the local oscillator. The pulsed radars were basically designed having in mind long distances applications.

Continuous Wave Radars (CW): This radar technology uses a continuous sinus wave at high frequency. It also uses the Doppler frequency shift in order to detect non-standing targets or their relative speed. The spectrum of standing targets would look centered in 𝑓₀, if the targets are moving this will be shifted by 𝑓_𝑑 (Doppler frequency). This characteristic makes the CW radars highly accurate when estimating the speed. In order to prevent the transmission disruption of the continuous radiation, two separate antennas need to be used. Thus, all the CW radars are bistatic.

There can be distinguished two main types of CW radar systems depending on the modulation they use; Frequency Modulated CW radars (FMCW) and Phase Modulated CW (PMCW) radars. A deeper comparison between FMCW and PMCW radars is given in Chapter 2.

Fig. 6: Block diagram of a monostatic architecture pulsed radar system.

(27)

- FMCW Radars: This type of modulated radar uses frequency modulation in the source. A VCO (Voltage Controlled Oscillator) generates the signal with up and down chirps with a period of Tp and 50% duty cycle. The signal is continuously radiated, and a small part of it, through the couples, is mixed with the echoes received, as shown in Fig. 7.

- PMCW radars: This type is also called phase-coded radar[10]. In contradiction with the pulsed radar systems (energy radiated in a short time interval), in this type of modulated radar a method of pulse compression can be used to generate a spread spectrum signal which will span the energy over a long period of time. For this purpose the phase modulation can be achieved with pseudo-random binary sequences, M-sequences or any other codes that will reduce the peak power of the system. As the radar system type presented in the document is going to be this type, a more detailed explanation of this method is given in Chapter 2.

Radars can be operated at a wide range of frequencies. Varying from frequencies as low as few MHz to as high as some hundred GHz, inside, what is called, the millimeter-wave region. A frequency range where radar systems can be found working covers from the 5 MHz up to the 95 GHz bands, as can be observed in Fig. 8. The used frequencies in radar vary in a wide frequency range. Thus, it is expected that the technologies associated are different and their use is linked to very different applications with very different capabilities and characteristics. Depending on the application, one frequency of operation is preferred to another. In Fig. 9(modified from [11]), an attenuation graph depending on the operating frequency can be found with a mark in the carrier frequency of the radar under study.

Fig. 7: Block diagram of the architecture of a FMCW radar.

(28)

As a first thought; the high frequency bands show an obvious advantage in applications that need small antennas; due to the small wavelength a more efficient antenna is possible. The short wavelengths permit high gain antennas with a convenient small size. But it is also easier to achieve better accuracy in range and placement of the target due to the wider band used in such frequency’s bands. By the other hand, low frequencies work better for applications with long distances goals since is more feasible to get high power antennas and the electromagnetic waves are less attenuated while propagating in the air. Also, high frequency radars will be more affected by weather conditions (water absorption peak) than a low frequency band radar, in long ranges. Following, an overview of the radar operating bands and applications:

- HF / VHF (< 300 MHz) – A/B bands: nowadays the technologies that use these frequency bands are called Over The Horizon radars (OTH). The main positive characteristics of this band are; the immunity to weather conditions, the long range capability out to 3500 km (since it is easier to obtain high power transmitters) and the low attenuation they suffer. However, their drawbacks are their low efficiency. Mainly useful for applications over the oceans. The low part of this band takes advantage on the ionosphere reflections to reach longer distances.

- UHF (300 MHz to 1 GHz) – C band: characterized by low or medium accuracy and resolution, but still very seldom weather conditions affection. It is a good frequency band to early detection of aircrafts and Airborne Moving Target Indication (AMTI). It is also good for tracking missiles and satellites. At the upper part of this frequency band there can be found the so-called ‘wind profiler radars’ that are capable of measuring the direction and velocity of the wind. Another application of this frequency band is the Ground Penetrating Radars (GPR), which usually extends further than the UHF band to get better accuracy. These ranges are appropriate to locate big objects under the ground.

- L band (1 – 2 GHz): It is commonly used for radars operating in ranges up to 400 km for air surveillance applications. The weather phenomenon starts to be slightly more noticeable in this range, and the more the frequency increases the more the rain effect will be significant. This band is also used for missile surveillance and low orbit satellite tracking.

- S band (2 – 4 GHz): used for short range surveillance, up to 120 km. This band is where regional airspace surveillance radars for aviation are located. It has medium accuracy and it can be said that is a band of compromises, relatively long range coverage with relative accuracy. Another radar application that uses this band is the surveillance aircraft AWACS (Airborne Warning and Control System).

- C band (4 – 8 GHz): Used for short range surveillance and long range tracking with good accuracy. It takes properties from the C and X bands and is highly sensitive to bad weather. Some examples of applications working in this band are: the weather radars to locate rains and, surveillance of military battlefield.

(29)

- X band (8 – 12 GHz): Meant to short range surveillance in good weather conditions. It is well known to be used for military applications, mainly for airborne radars, and it is also common for police speed meters. They are usually very convenient due to the size of the antennas so they are very popular for applications where weight and mobility are the main factors to take into account.

- Ku / Ka bands (12 – 40 GHz): Mainly for short-range tracking. Since the higher the frequency the more is the atmosphere absorption, this band is good for applications where weather conditions are not taken into account. This band is attractive when it comes to applications that require very little installation space and don’t need long ranges. An application working in this range is the Surface Radar in the major airports.

- V / W / mm-Wave bands (40 – 100+ GHz): This range is clearly limited to short or very short ranges, depending on the weather conditions, and mainly work besides smart antennas which are able to point their very thin beams. In this wide band there is particular interest around the 96 GHz zone, since there is a relative minimum of attenuation. The effects and technology of the mm-wave technology are different than those of the microwave radar technology, and both are more limiting in the mm-wave radar technology. The use of this band has lately become more popular because of many advantages. The main one, is the availability of a wide band of frequencies, there is a lot of unused space, so that the radar systems to be developed in this region can have a wide bandwidth which will give higher range resolution and narrower beams with smaller antennas. It is also very well considered for its robustness, for example, for military applications where countermeasures are expected. Nowadays there are big investments in mmwave radar for automotive applications due to its small size and very good values of accuracy and resolution. Usually these automotive systems operate in the range from 73 to 81 GHz. Further explanations and development on this precise application, technology and frequency band will be provided throughout the present document. The working frequency of the radar system here presented is going to be in this band, with working frequencies located in the mm-wave band.

L S C X Ku K Ka U V / mmWave

Microwave Bands mmWave Bands

Radar

Fig. 8: Spectrum with the location of the traditional microwave bands and the mmWave bands.

(30)

1.3 Proposed Radar System

The radar system described in this thesis has its main application goal in the automotive industry.

Concretely, the specifications were established in order to focus in the so called short range radar (SRR) systems [12], which reaches distances around 30 meters in the direct proximity of the vehicle, as can be seen in Fig. 10.

There have been developments of automotive radar systems in the band of the 24 GHz with technology built on SiGe (Silicon Germaium) blocks [13], [14]. Other frequencies have also been explored (under 10 GHz or over 100 GHz) but their role is insignificant. Being that the market pursues for higher integration achievements with lower power consumption, which cannot be achieved with these technologies and frequency bands. The use of the mmWave band is more challenging and complex from the technical point of view, but it provides higher performance possibilities. A key characteristic for an easy integration is the sensor size, which is determined by the antenna aperture.

Fig. 9: Attenuation in the atmosphere of electromagnetic waves according to their working frequency.

(31)

At high frequency the antenna size is allowed to be small keeping a good angular resolution. In order to achieve the same performance in the 24 GHz band, a three times larger aperture would be required [12]. Furthermore, the mmWave band is not affected by tough atmospheric phenomena as rain, snow, sand storms, lightning, etc. Nevertheless, their small form factor means very low transmitted energy and narrow beams, which provide interference immunity. The 79 GHz band [77 – 81 GHz] has arisen as the standard initiative to locate the worldwide vehicular radar frequencies and it will be worldwide used for SRR [15].

The first disclosed use of a radar system based on millimeter-wave technology for automotive purposes dates back in the 1970s [12]. Nonetheless, their use did not go further than the building of prototypes. The main reasons for its market failure were the large size and high production costs that the systems represented by that time. It was not until 1998 when the first mmWave was ready for use.

By 2003 the majority of car makers had a radar option based on this technology [16].

Nowadays the mmWave band radar market production is experiencing a rapid growth in terms of volume, with a consequent drop of the prizes, and allowing, for example, driver assistance systems available for middle class vehicles [17].

SRR 77-81 GHz Range < 30 m Resolution < 10 cm

LRR 76-77 GHz Range < 200 m Resolution > 1 m

Fig. 10: Radar short and long range types with the mmWave frequencies and capabilities.

(32)

Since the market and development of these systems is still undergoing and the technical background behind its development is complex, there is an important assortment of different market solutions.

Despite these differences, all the available sensors in the market count on the FMCW modulation technique [18]; for example the LRR3 sensor of Bosch [19], the ARS 300 by Conti [20] or the Denso sensor [21]. However, the implementation of the FMCW technique faces challenges due basically to the big amount of energy that the synthesizer needs and the limited performance of the frequency slope. The target resolution is important and in FMCW this resolution is linked to the sweep bandwidth of the transmitting signal which requires a VCO with a big tuning range. Besides the difficulties to realize the system for big bandwidths, VCO’s linearity also limits the accuracy of the measurements.

One alternative to FMCW is the PMCW technique using pseudo-noise signals; it is easier to realize and its accuracy does not depend on the linearity of the transmitter signal. The resolution depends on the bandwidth of the modulation which is given by the clock frequency of the code generator. There have been some published studies, in [22], [23], [24], and implementations with SiGe technology [25]

and [13].

In the last years, the functions implemented in cars with radar technology solutions have been mainly comfort services, such as: warning systems for a near collision, assistance in lane changes, blind spots monitors, park assistance, etc. There also exist some advances in active safety, however in the immediate future; the 79 GHz frequency band will add high-resolution radar performance, therefore allowing the increase of the safety for car occupants and pedestrians, also improving the comfort and services onboard for the driver and companions.

In order to be able to develop and apply these safety systems with assurance, there are improvements that need to be implemented. The systems need to be able to differentiate better the objects in the road in front and nearby the car. The development and exploitation of the 79 GHz band brings the opportunity of using systems with a much broader bandwidth comparing to the existing solutions, thus increasing the capability of the radar systems. It will permit the implementation of proactive safety characteristics. Using the technologies associated will allow a better resolution and accuracy to analyze the objects on the road. The wide bandwidth together with the investigation in signal processing will allow functionalities such as autonomous emergency braking against collisions or pedestrians.

Signal Processing for mmWave MIMO Radar

FACULTY OF ENGINEERING AND SUSTAINABLE DEVELOPMENT