Developing a DSP Core using an FPGA Prototype for Scintillation Detector
Signals
Submitted to:
Communication & Electronics Dept., Al Azhar University
Supervised by:
Prof. Dr. Ahmed Safwat Prof. Dr. Mahmoud Ashour
Dr. Ashraf Aboshosha
Prepared by: Eng. Mahmoud Kamel
Outline
• This core gives us all important features of the scintillation detector signals such as shaping,
counting, pulse height and multichannel analyzing.
• The main purpose of this research work is to de-noise, compress and reconstruct the
scintillation signals by which the processing speed, storage and precision will be improved.
3
Outline
• This core is implemented to apply the forward wavelet transform and interpolation technique. A new contribution of this framework arises from employing the interpolation techniques to reconstruct the signals where the mother wavelet and details are not required.
• Building a Multi-Channel Analyzer of the scintillation detector signals
4
Index of Content
• Scintillation detectors • Importance of scintillation detectors• Data Acquisition System• Proposed digital processing algorithm• Wavelets – Interpolation Technique• Comparative study with the previous techniques• Single channel and multi channel analyzer• Conclusions and future work
5
Scintillation Detector
6
Figure 1 : Schematic diagram of a scintillation detector
Scintillation Detector
• A scintillator is a material that emits light, scintillates, when absorbing radiation.
• The energy can be determined by measuring the pulse height spectrum. This is called spectroscopy.
• A scintillation detector is obtained when a scintillator is coupled to an electronic light sensor such as PMT or photodiode.
7
Importance of Scintillation Detectors• Detection of mixed ionizing fluxes near
nuclear objects.• Radionuclide control of samples and
radiation pollution.• Determination of the type and energy of
high-energy particles and products of their reactions with targets.
• Nuclear medicine (Gamma Camera, PET Tomography, …)
8
Data Acquisition System
9
Figure 2 : The practical data acquisition system of scintillation detector Signals.
(1) Scope , (2) high voltage source, (3) scintillator , (4) power supply
Why FPGA ?
• FPGA incorporates thousands of logic cells linked by programmable switches
• Highly parallel configurable digital signal processor
• A many channel signal processing was required in these detector to obtain a precise signals
• Availability of high-level design entry method• FPGA designs easily changed, recompiled and
low cost
10
FPGA Design Flow of the Solution
Synthesis• Translate Design into Device Specific Primitives• Optimization to Meet Required Area & Performance Constraints
Design Specification
Place & Route• Map Primitives to Specific Locations inside Target Technology with Reference to Area &• Performance Constraints• Specify Routing Resources to Be Used
Design Entry/RTL CodingBehavioral or Structural Description of Design
LEMEM I/O
RTL Simulation• Functional Simulation• Verify Logic Model & Data Flow (No Timing Delays)
11
FPGA Design Flow of the solution
Timing Analysis - Verify Performance Specifications Were Met - Static Timing Analysis
Gate Level Simulation - Timing Simulation - Verify Design Will Work in Target Technology
Program & Test- Program & Test Device on Board
tclk
12
13
Pre-processing Phase 1-Wavelet based Decomposition
2- Interpolation based Reconstruction
Pulse Shaping & Counting
Multichannelanalyzer
Store & Show data
Figure 3: The overall proposed solution
14
Pre Amplifier
Main Amplifier
SCA
MCA
Counter A
B
Figure 4: The proposed solution
Hardware System
15
Figure 5: The FPGA XSC50k-Spartan II and the PC-based parallel interface
Pre-processing Phase
• De-noising • Compression• Reconstruction
16
Effect of Noise onPulse Shaping & Counting
17
Figure 6: Effect of noise on pulse shaping
Wavelets• The wavelet analysis procedure is to adopt a
wavelet prototype function, called an analyzing wavelet or mother wavelet.
• Wavelet transform decompose the original signal into different scales of resolution; these called the approximation and detail coefficients .
18
Wavelet Decomposition Levels
19
H
G
d1
X0 H
G
d2
d3
Figure 7: Three wavelet decomposition levels
A3
Wavelet Families • Haar• Daubechies• Biorthogonal • Coifelt• Symelet• Myer
20
Haar Wavelet Design Pro:• Allows good approximation with a subset of
coefficients.• It can be computed quickly and easily. • Implemented easily by FPGA.
21
Design Block Diagram
22
Figure 8: Design block diagram
Selecting the best Decomposition Level
• The quality of the compressed signals is the main criterion to select the best
decomposition level in terms of Peak Signal to Noise Ratio (PSNR).
• The other similarity measure are Euclidean Distance (ED) , Cross Correlation coefficient (CC) and Mean Square Error(MSE).
23
Decomposition Levels
24
Figure 9: Four approximation coefficients of Haar wavelet transform
Statistics of Four Decomposition Levels
25
LevelPSNRMSEEDCCOne27.85670.169320.47780.9681
Two30.70840.157514.73950.9830
Three31.92580.144312.79900.9866
Four18.95543.256157.06250.7021
Table 1: Statistics of four levels Haar transform
26
Mother Wavelet
PSNRMSEEDCC
Haar31.9260.164312.79900.9866Daubechies32.56560.141811.81390.9890Coiflet32.863
50.132411.38340.9900
Meyer13.5046
11.4230
106.8780.0148
Biorthogonal32.2270.153312.27760.9886Table 2: Similarity measure of constructed and original
signals of the different mother wavelets
Comparison of Different Mother Wavelets
Interpolation
• The Interpolation is a method of constructing new data points within the range of a discrete set of known data points.
• Interpolation is performed by fitting the supplied data with polynomial functions between data points and evaluating the appropriate function at the desired points.
27
Reconstruct Signals Using Interpolation
28
Figure 10: a) Original signals. b) Transformed signal. c) Reconstructed signals
Interpolation Algorithms
• Nearest neighbor interpolation• Linear interpolation• Cubic Hermit Interpolation • Cubic spline interpolation
29
30
MethodPSNRMSEEDCC
Linear29.71920.273116.46690.9782
Cubic Spline31.92580.164312.79900.9866
Nearst27.85010.838028.84620.9307
Cubic Hermit30.60660.222614.89840.9818
Table 3: Statistics of different interpolation techniques
Comparison of Applying Different Interpolation Techniques
Previous Pre-processing Techniques
1. Accumulation Technique 2. Median filter
31
Accumulation Technique
32
Figure 11: Digital processing algorithm of scintillation detector signals
Median Filter• The value of an output sample is determined
by the median of the neighborhood signals.
33Figure 12: Reconstructed signals using Median filter
34
MethodPSNRMSEEDCCAccumulation Tech27.49720.455521.3433 0.9680Median filter30.68560.218614.7856 0.9831Proposed Solution.31.92580.164312.79900.9866
Table 4: Statistics of the preprocessing techniques
Comparison of the Preprocessing Results
35
Figure 13: Pulse shaping after denoising
36
Figure 14: Pulse counting
Multi Channel Analyzer• The MCA system is used to measure the height of
each output pulse and the number of each output pulses simultaneously.
• By performing this operation for all detector events in a given interval the MCA generates a spectrum of the distribution of energy for a measured events with the y axis representing counts and the x axis representing channel value.
37
Multi Channel Analyzer
38Figure 15: Divided original signals into 16 channels
Multi Channel Analyzer
39Figure 16: Energy spectrum with 16 channels
Channel Calibration
• Energy channel values are converted into kilo electron volts with a channel-to-kilo electron volt conversion factor which is determined from a comparison of photo peak energies and channel location close to the energy of interest.
40
Conclusions• One of the most important advantages of this
system is the high compression rate (12.5%) using the interpolated wavelets
• Compared with the accumulation technique and median filtering, the proposed design achieved the best precision
• Capability of constructing MCA from SCA• Coiflet is the best mother wavelet and Cubic
spline is the best interpolation technique. Combining both of them for down and up sapling in wavelets is a new theoretical contribution of this framework
41
Future work
• Applying more complex wavelet filters.• Modifying the proposed architecture to
process more scintillator detectors.• Employing the presented results as a base to
identify radiation type and isotopes.
42