jfriedt / IFCS2018 article

Download as
Browse Code ยป

Commit bad78fb7c777e4f94c7f209404857097c824a97f

Authored by jfriedt
1 parent 08e32d5a93
Exists in master

corrections

Showing 2 changed files with 62 additions and 22 deletions Side-by-side Diff

... ... @@ -143,9 +143,9 @@
143 143 \begin{center}
144 144 \begin{tabular}{|c|cccc|}\hline
145 145 FIR & BlockRAM & LookUpTables & DSP & rejection (dB)\\\hline\hline
146   -1 (monolithic) & 1 & 4064 & 40 & -71.78 \\
147   -5 & 5 & 12332 & 0 & -216.58 \\
148   -10 & 10 & 12717 & 0 & -251.01 \\\hline\hline
  146 +1 (monolithic) & 1 & 4064 & 40 & -72 \\
  147 +5 & 5 & 12332 & 0 & -217 \\
  148 +10 & 10 & 12717 & 0 & -251 \\\hline\hline
149 149 Zynq 7010 & 60 & 17600 & 80 & \\\hline
150 150 \end{tabular}
151 151 \end{center}
... ... @@ -34,12 +34,13 @@
34 34 % Title
35 35 \begin{center}
36 36 \textbf{{\scshape
37   - \Large\color{OliveGreen}
  37 + \LARGE\color{OliveGreen}
38 38 Filter optimization for real time digital processing of radiofrequency signals: application
39 39 to oscillator metrology
40 40 \\}}
41 41 \end{center}
42 42  
  43 +\vspace{-0.7cm}
43 44 % Authors
44 45 \begin{center}
45 46 \addalignedblock{0.18\textwidth}{flushleft}{%
46 47  
47 48  
48 49  
... ... @@ -61,14 +62,33 @@
61 62 \end{center}
62 63  
63 64 % First part
  65 +\vspace{-.71cm}
64 66 \newsection{Digital signal processing of ultrastable clock signals}
65 67  
66   -Je ne sais pas trop quoi dire ici
  68 +\vspace{-.21cm}
  69 +\begin{itemize}[leftmargin=*]
  70 +\setlength{\itemsep}{0pt}%
  71 +\setlength{\parskip}{0pt}%
  72 +\item
  73 +{\bf Digital phase noise characterization}: flexibility (software defined local
  74 +oscillator), stability (no long term drift), reconfigurabilty
  75 +$\Rightarrow$ {\bf software defined radio} oscillator phase noise
  76 +characterization
  77 +\item analog to digital conversion of radiofrequency signal, software
  78 +defined local oscillator, mixer and {\bf low pass filter}
  79 +\item low pass filter uses most resources and introduces latency (phase delay
  80 +in feedback loop): needs to be optimized
  81 +\end{itemize}
67 82  
  83 +\vspace{-.21cm}
  84 +\hrule{\hfill}
68 85 % Second part
69   -\newsection{Filter design}
70   -\begin{itemize}
71   - \item How to implementing filter:\\
  86 +\vspace{-.71cm}
  87 +\newsection{Filter design and implementation strategy:}
  88 +%\begin{itemize}[leftmargin=*]
  89 +%\setlength{\itemsep}{0pt}%
  90 +%\setlength{\parskip}{0pt}%
  91 +\vspace{-.41cm}
72 92 \addblock{0.48\textwidth}{
73 93 \begin{enumerate}[noitemsep,nolistsep]
74 94 \item \textbf{Classical way:}\\
75 95  
76 96  
77 97  
78 98  
... ... @@ -92,21 +112,28 @@
92 112 \end{itemize}
93 113 \end{enumerate}
94 114 }
95   - \item The 2\textsuperscript{nd} way could be considered as an optimization problem:
  115 + The 2\textsuperscript{nd} way could be considered as an optimization problem:
96 116 \begin{itemize}[noitemsep,nolistsep]
97   - \item One or many performance criteria (rejection, noise, throughput...)
98   - \item Limited resources (on FPGA)
  117 + \item One or many {\bf performance criteria} (rejection, noise,
  118 +throughput...)
  119 + \item Limited {\bf resources} (on FPGA)
99 120 \end{itemize}
100   - \item Translation into a Mixed-Integer Linear Programming (MILP) with GLPK solver
101   - \item 3 degrees of freedom:
  121 + Translation into a Mixed-Integer Linear Programming (MILP) with GLPK solver
  122 + 3 degrees of freedom:
  123 +
  124 +\vspace{.1cm}
  125 +\hfill
  126 +\parbox{.60\linewidth}{
102 127 \begin{enumerate}[noitemsep,nolistsep]
103 128 \item The size of chain filters
104 129 \item The number of coefficients for each filter $i$: $N_i$
105 130 \item The number of bits for each coefficients and for each filter $i$: $c_i$
106 131 \end{enumerate}
107   -\end{itemize}
108   -\vspace{-0.5cm}
  132 +}
  133 +%\end{itemize}
  134 +\vspace{-1.0cm}
109 135 \newsection{Filter selection}
  136 +\vspace{-0.3cm}
110 137 \begin{itemize}[noitemsep,nolistsep]
111 138 \item For select the filter design we need to evaluate the rejection like:
112 139 \begin{enumerate}[noitemsep,nolistsep]
113 140  
114 141  
115 142  
116 143  
117 144  
... ... @@ -140,20 +167,33 @@
140 167 \includegraphics[width=0.95\textwidth]{images/fir-mono-vs-fir-series-noise-fixe-jmf.pdf}
141 168 \captionof{figure}{Custom criterion}
142 169 \end{minipage}
143   - \item For the rejection: the last configuration is better than the first but it's worst than monolithic filter
144   - \item For the resources consumption: the last better than the single filter
  170 + \item {\bf Rejection}: the last configuration is better than the first but worse
  171 +than the monolithic filter
  172 + \item Resources {\bf consumption}: last filter is better than the single monolithic filter
  173 +(monolithic does not fit in available resources)
  174 +\vspace{-.33cm}
145 175 \begin{center}
146 176 \begin{tabular}{|c|ccccc|}\hline
147 177 FIR & BlockRAM36 & BlockRAM18 & LookUpTables & DSP & rejection (dB)\\\hline\hline
148   - 1 (monolithic) & 1 & 0 & {\color{Red}76183} & 220 & -162.19 \\
149   - 5 & 0 & 5 & {\color{Green}18597} & 220 & -160.06 \\
150   - 10 & 0 & 8 & {\color{Green}24729} & 220 & -161.30 \\\hline\hline
  178 + 1 (monolithic) & 1 & 0 & {\color{Red}76183} & 220 & -162 \\
  179 + 5 & 0 & 5 & {\color{Green}18597} & 220 & -160 \\
  180 + 10 & 0 & 8 & {\color{Green}24729} & 220 & -161 \\\hline\hline
151 181 \textbf{Zynq 7020} & \textbf{140} & \textbf{280} & \textbf{53200} & \textbf{220} & \\\hline
152 182 \end{tabular}
153   - \captionof{table}{Resources consumption when we use the configuration with the custom criterion}
  183 +% \captionof{table}{Resources consumption when we use the configuration with the custom criterion}
154 184 \end{center}
155   - \item With a serie of filters we able to reach the rejection level because we consume less resources than the traditional filter
  185 + \item Series of filters: targetd rejection level (-160~dB) reached since less
  186 +resources are needed than with a monolithic filter
156 187 \end{itemize}
  188 +\hrule{\hfill}
157 189  
  190 +\vspace{-.71cm}
  191 +\newsection{Conclusion}
  192 +
  193 +\vspace{-.21cm}
  194 +\noindent
  195 +FIR filter implementation in an FPGA as an optimization problem: best
  196 +results with cascaded filters with increasing number of coefficients
  197 +and resolution
158 198 \end{document}