a correlation based pulse detection technique for gamma/neutron...
TRANSCRIPT
A Correlation Based Pulse Detection Technique for
Gamma/Neutron DetectorsMuhammad Faisal1), Randolph T. Schiffer1), Marek
Flaska2), Sara A. Pozzi2), David D. Wentzloff1)1) Department of Electrical Engineering and Computer Science2) Department of Nuclear Engineering and Radiological Sciences
University of Michigan, Ann Arbor, Michigan 48109, USA
Symposium on Radiation Measurements and Applications, Ann Arbor, Michigan, USA. May 24‐28, 2010
Wednesday May 26, 2010Acknowledgement: This research was funded by the National Science Foundation and the Domestic Nuclear Detection Office of the
Department of Homeland Security through the Academic Research Initiative Award # CMMI 0938909.
Motivation• Shielded radioactive materials
– Low energy radiation
– Difficult to detect
• Signal processing to detect these pulses
• Efficient online detection required (e.g. airports)
2
Shielded Materials
Detector
Real‐Time Detection
Wireless Integrated Circuits and Systems Group M. Faisal
• Analog : mostly counting pulses
• Digital : Data stored for post‐processing
• Threshold must be lowered to detect lower energy pulses
false detections
Current State‐of‐the‐Art
3
Digitizer UnitV threshold
Data Storage
A/D
A/D
A/D
Wireless Integrated Circuits and Systems Group M. Faisal
PMTs
This Approach
• Data is analyzed prior to storage– Significantly less amount of data is stored
• Improve detector sensitivity• Real‐time pre‐processing
4
Pre‐processing on FPGA
Data StorageA/D
A/D
A/D
Wireless Integrated Circuits and Systems Group M. Faisal
512MB Memory
FPGA• Peak extract• Timing• DSP
PMTs
Equipment
• Commercial board – 250MS/s, 14 bits, 4 channels
– Connects directly to PC (e.g. laptop)
• On‐board FPGA customizablefor pre‐processing of data
5Wireless Integrated Circuits and Systems Group M. Faisal
Real‐time detectionExtract useful information
from pulses (e.g. height, time)
Innovative Integration’s X5‐210M
• Pre‐processing in real time on an FPGA
• Measure of similarity between two signals
Normalized Cross Correlation(1/2)
6
Time
Wireless Integrated Circuits and Systems Group M. Faisal
( )( )[ ] ( )[ ]
( )[ ] ( )[ ]∑∑∑
−−−
−−−=
xx u
x u
tuxtfxf
tuxtfxfuC
22( ) 11 +≤≤− uC
Signal Processing
FPGA
0.0
0.2
0.6
1.0
0.0
0.2
0.6
1.0
Time
Amplify Signal & Attenuate Noise
7
Normalized Cross Correlation(2/2)
Wireless Integrated Circuits and Systems Group M. Faisal
Pulse Template 400 800
0.0
0.4
0.8
1.2
Time
Voltage
Incoming Pulses
400 800Time
C
Correlator Output
20 60 100 140
Voltage
0.0
0.4
0.8
1.2
0.0
0.4
0.8
1.2
Time
88Wireless Integrated Circuits and Systems Group M. Faisal
Template
Capture pulses
Bin pulse data
Average pulses in each bin
Normalize
Average 1500 pulses
Template Measured Pulses
• Template used in the correlator to recognize pulses
• Correlator building block: Multiply and Accumulate
• Available on FPGA as hardware accelerators
Hardware Implementation(1/2)
9Wireless Integrated Circuits and Systems Group M. Faisal
M&A∑
( ) [ ] [ ][ ] [ ]∑∑
∑ ×=
xx
x
TemplateData Incoming
TemplateData IncominguC
22
Normalize
ROM
Hardware Implementation(2/2)
10Wireless Integrated Circuits and Systems Group M. Faisal
( )tKD
NuC⋅
=
M&A
Data
Template
M&A
tK
N
Normalize
( )tKD
NuC⋅
=
M&A
M&AN
D
Normalize
( )tKD
NuC⋅
=
M&A
M&AN
D
Normalize
( )tKD
NuC⋅
=
M&A
M&AN
D
( )2+iC
( )iC( )1+iC
Detect tarrival
Vpk
FPGA. . .
Simulation Results (1/2)
• Pulses detected in presence of noise• E.g. A 5 keVee pulse in 5 keVee noise
11
0 5 10 15 20
0.2
0.4
0.6
0.8Co
rrelation
Energy (keVee)
Noise Level5 keVee Noise Correlation
Cth = 0.245 (3‐σ)
Pulse Correlation
Wireless Integrated Circuits and Systems Group M. Faisal
1.0
Simulation Results (2/2)
• A 5 keVee pulse in presence of 5 keVee noise• Probability of detection:
12Wireless Integrated Circuits and Systems Group M. Faisal
2 4 6 8 10 120
0.2
0.4
0.6
0.8
1.0Prob
ability of D
etection
Energy (keVee)
Probability of Detection vs. Pulse Energy
Amplitude Threshold
CorrelatorBased
18% to 86%
Conclusion
• Proposing an FPGA‐based approach for pulse detection– Real‐time data processing
– Improved sensitivity (≈ 4x @ 5keVee)
– Reduced data storage
13Wireless Integrated Circuits and Systems Group M. Faisal
A Correlation Based Pulse Detection Technique for
Gamma/Neutron DetectorsMuhammad Faisal1), Randolph T. Schiffer1), Marek
Flaska2), Sara A. Pozzi2), David D. Wentzloff1)1) Department of Electrical Engineering and Computer Science2) Department of Nuclear Engineering and Radiological Sciences
University of Michigan, Ann Arbor, Michigan 48109, USA
Symposium on Radiation Measurements and Applications, Ann Arbor, Michigan, USA. May 24‐28, 2010
Wednesday May 26, 2010Acknowledgement: This research was funded by the National Science Foundation and the Domestic Nuclear Detection Office of the
Department of Homeland Security through the Academic Research Initiative Award # CMMI 0938909.