relecture JMF

jfriedt
1 parent b43d41ac2d
Showing 2 changed files with 80 additions and 29 deletions Side-by-side Diff
ifcs2018_journal.tex
ifcs2018_journal_reponse.tex
@@ -120,9 +120,10 @@
 signals, High Level Synthesis (HLS) languages \cite{kasbah2008multigrid} are not considered but
 the problem is tackled at the Very-high-speed-integrated-circuit Hardware Description Language
 (VHDL) level.
-Since latency is not an issue in a openloop phase noise characterization instrument, the large
+{\color{red}Since latency is not an issue in a openloop phase noise characterization instrument, 
+the large
 numbre of taps in the FIR, as opposed to the shorter Infinite Impulse Response (IIR) filter,
-is not considered as an issue as would be in a closed loop system.
+is not considered as an issue as would be in a closed loop system.}  % r2.4
  
 The coefficients are classically expressed as floating point values. However, this binary
 number representation is not efficient for fast arithmetic computation by an FPGA. Instead,
@@ -143,8 +144,9 @@
 relation between number of fiter taps and quantization, Fig. \ref{float_vs_int} exhibits
 a 128-coefficient FIR bandpass filter designed using floating point numbers (blue). Upon
 quantization on 6~bit integers, 60 of the 128~coefficients in the beginning and end of the
-taps become null, making the large number of coefficients irrelevant and allowing to save
-processing resource by shrinking the filter length. This tradeoff aimed at minimizing resources
+taps become null, {\color{red}making the large number of coefficients irrelevant: processing 
+resources % r1.1
+are hence saved by shrinking the filter length.} This tradeoff aimed at minimizing resources
 to reach a given rejection level, or maximizing out of band rejection for a given computational
 resource, will drive the investigation on cascading filters designed with varying tap resolution
 and tap length, as will be shown in the next section. Indeed, our development strategy closely
  
@@ -182,22 +184,24 @@
  
 Addressing only two operations allows for demonstrating the methodology but should not be
 considered as a limitation of the framework which can be extended to assembling any number
-of skeleton blocks as long as perfomance and resource occupation can be determined. Hence,
-in this paper we will apply our methodology on simple DSP chains: a white noise input signal
-is generated using a Pseudo-Random Number (PRN) generator or thanks at a radiofrequency-grade
-Analog to Digital Converter (ADC) loaded by a 50~$\Omega$ resistor. Once samples have been
+of skeleton blocks as long as perfomance and resource occupation can be determined. {\color{red}
+Hence,
+in this paper we will apply our methodology on simple DSP chains: a white noise input signal % r1.2
+is generated using a Pseudo-Random Number (PRN) generator or by sampling a wideband (125~MS/s)
+14-bit Analog to Digital Converter (ADC) loaded by a 50~$\Omega$ resistor.} Once samples have been
 digitized at a rate of 125~MS/s, filtering is applied to qualify the processing block performance --
 practically meeting the radiofrequency frontend requirement of noise and bandwidth reduction
 by filtering and decimating. Finally, bursts of filtered samples are stored for post-processing,
 allowing to assess either filter rejection for a given resource usage, or validating the rejection
 when implementing a solution minimizing resource occupation.
  
-The first step of our approach is to model the DSP chain and since we just optimize
-the filtering, we have not modeling the PRN generator or the ADC. The filtering can be
-done by two ways. The first one we use only one FIR filter with lot of coefficients
-to rejection the noise, we called this approach a monolithic approach. And the second one
-we select different FIR filters with less coefficients the monolithic filter and we cascaded
-it to filtering the signal.
+{\color{red}
+The first step of our approach is to model the DSP chain. Since we aim at only optimizing % r1.3
+the filtering part of the signal processing chain, we have not included the PRN generator or the 
+ADC in the model: the input data size and rate are considered fixed and defined by the hardware. 
+The filtering can be done in two ways, either by considering a single monolithic FIR filter
+requiring many coefficients to reach the targeted noise rejection ratio, or by 
+cascading multiple FIR filters, each with fewer coefficients than found in the monolithic filter.}
  
 After each filter we leave the possibility of shifting the filtered data to consume
 less resources. Hence in the case of cascaded filter, we define a stage as a filter
@@ -241,7 +245,11 @@
 transfer function.
 Comparing the performance between FIRs requires however defining a unique criterion. As shown in figure~\ref{fig:fir_mag},
 the FIR magnitude exhibits two parts: we focus here on the transitions width and the rejection rather than on the
-bandpass ripples as emphasized in \cite{lim_1988,lim_1996}.
+bandpass ripples as emphasized in \cite{lim_1988,lim_1996}. {\color{red}Throughout this demonstration,
+we arbitrarily set a bandpass of 40\% of the Nyquist frequency and a bandstop from 60\%
+of the Nyquist frequency to the end of the band, as would be typically selected to prevent
+aliasing before decimating the dataflow by 2. The method is however generalized to any filter
+shape as long as it is defined from the initial modelling steps.}
  
 \begin{figure}
 \begin{center}
@@ -83,14 +83,32 @@
  
 {\bf
 On page 2,  "...allowing to save processing resource..." could be improved.       % r1.1
+}
  
+The sentence was split and now reads ``number of coefficients irrelevant: processing 
+resources are hence saved by shrinking the filter length.''
+
+{\bf
 On page 2, "... or thanks at a radiofrequency-grade..." isn't at all clear what   % r1.2
-the author meant.
+the author meant.}
  
-One page 2, the whole paragraph "The first step of our approach is to model..."   % r1.3
+Grammatical error: this sentence now reads ``or by sampling a wideband (125~MS/s)
+Analog to Digital Converter (ADC) loaded by a 50~$\Omega$ resistor.''
+
+{\bf 
+On page 2, the whole paragraph "The first step of our approach is to model..."   % r1.3
 could be improved.
 }
  
+Indeed this paragraph has be written again and now reads as\\
+``The first step of our approach is to model the DSP chain. Since we aim at only optimizing
+the filtering part of the signal processing chain, we have not included the PRN generator or the
+ADC in the model: the input data size and rate are considered fixed and defined by the hardware. 
+The filtering can be done in two ways, either by considering a single monolithic FIR filter
+requiring many coefficients to reach the targeted noise rejection ratio, or by
+cascading multiple FIR filters, each with fewer coefficients than found in the monolithic filter.
+''
+
 {\bf
 I appreciate that the authors attempted and document two optimizations: that      % r1.4 - en attente des résultats
 of maximum rejection ratio at fixed silicon area, as well as minimum silicon
  
  
@@ -178,17 +196,36 @@
 than FIR. This is not true in general. The sentence should be reconsidered.
 }
  
-J'aurais du dire ``lag'' au lieu de ``impulse response'' je pense
-AH: Je ne comprends pas trop ce qui ne va pas ici
+We have not stated that the IIR has a shorter impulse response but a shorter lag.
+Indeed while a typical FIR filter will have 32 to 128~coefficients, few IIR filters
+have more than 5~coefficients. Hence, while a FIR requires 128 inputs before providing
+the first output, an IIR will start providing outputs only 5 time steps after the initial
+input starts feeding the IIR. Hence, the issue we address here is lag and not impulse
+response. We aimed at making this sentence clearer by stating that ``Since latency is not an issue 
+in a openloop phase noise characterization instrument, the large
+numbre of taps in the FIR, as opposed to the shorter Infinite Impulse Response (IIR) filter,
+is not considered as an issue as would be in a closed loop system in which lag aims at being
+minimized to avoid oscillation conditions.
+''
  
 {\bf
 - Fig. 4: the Author should motivate in the text why it has been chosen           % r2.5
 this transition bandwidth and if it is a typical requirement for phase-noise
 metrology.
 }
-AH: Je ne sais pas comment justifier ça. Je dois dire que comme ça on peut éventuellement
-décimer par deux le flux ?
  
+The purpose of the paper is to demonstrate how a given filter shape can be achieved by
+minimizing varous resource criteria. Indeed the stopband and bandpass boundaries can
+be questioned: we have selected this filter shape as a typical anti-aliasing filter considering
+the the dataflow is to be halved. Hence, selecting a cutoff frequency of 40\% the initial
+Nyquist frequency prevents noise from reaching baseband after decimating the dataflow by a
+factor of 2. Such ideas are now stated explicitly in the text as ``Throughout this demonstration,
+we arbitrarily set a bandpass of 40\% of the Nyquist frequency and a bandstop from 60\%
+of the Nyquist frequency to the end of the band, as would be typically selected to prevent
+aliasing before decimating the dataflow by 2. The method is however generalized to any filter
+shape as long as it is defined from the initial modelling steps: Fig. \ref{fig:rejection_pyramid}
+as described below is indeed unique for each filter shape.''
+
 {\bf
 - The impact of the coefficient resolution is discussed. What about the           % r2.6 - fait
 resolution of the data stream? Is it fixed? If so, which value has been
  
@@ -196,13 +233,15 @@
 coefficient resolution?
 }
  
-Pr\'eciser que le flux de donn\'ees en entr\'ees est de r\'esolution fixe
+We have now stated in the beginning of the document that ``we have not included the PRN generator 
+or the ADC in the model: the input data size and rate are considered fixed and defined by the 
+hardware.'' so indeed the input datastream resolution is considered as a given.
  
 {\bf
 - Page 3, line 47: the initial criterion can be omitted and, consequently,        % r2.7  - fait
 Fig. 5 can be removed.
-- Page 3, line 55: “maximum rejection” is not compatible with fig. 4.             % r2.8  - fait
-It should be “minimum”
+- Page 3, line 55: ``maximum rejection'' is not compatible with fig. 4.             % r2.8  - fait
+It should be ``minimum''
 }
 AH: Je ne suis pas d'accord, le critère n'est pas le min de la rejection mais le max
 de la magnitude. J'ai corrigé en ce sens.
@@ -214,7 +253,12 @@
 - Page 4, line 10: how $p$ is chosen? Which is the criterion used to choose       % r2.12 - fait
 these particular configurations? Are they chosen automatically?
 - Page 4, line 31: how does the delta function transform model from non-linear    % r2.13 - fait
-and non-quadratic to a quadratic?
+and non-quadratic to a quadratic?}
+
+JMF : il faudra mettre une phrase qui explique, ca en lisant cette reponse dans l'article
+je ne comprends pas comment ca repond a la question
+
+{\bf
 - Captions of figure and tables are too minimal.                                  % r2.14
 - Figures can be grouped: fig. 10-12 can be grouped as three subplots (a, b, c)   % r2.15 - fait
 of a single figure. Same for fig. 13-16.
@@ -226,10 +270,9 @@
 differences among the curves. I suggest to reduce the noise below 1 dBpk-pk.
 }
  
-Comment as tu fait tes spectres Arthur ? Si tu as fait une FFT sur e.g. 2048 points
-mais que tu as des jeux de donnees de e.g. 10000 points, on peut faire des moyennes
-sur les sequences successives. Au pire si pas possible, une moyenne glissante sur
-chaque spectre pour affiner les traits ?
+Indeed averaging had been omitted during post-processing and figure generation: we 
+are grateful to the reviewer for emphasizing this point which has now been corrected. All spectra
+now exhibit sub-dBpk-pl line thickness.
  
 %In conclusion, my opinion is that the methodology presented in the Manuscript
 %deserve to be published, provided that the criterion is changed according
...	...	@@ -120,9 +120,10 @@
120	120	signals, High Level Synthesis (HLS) languages \cite{kasbah2008multigrid} are not considered but
121	121	the problem is tackled at the Very-high-speed-integrated-circuit Hardware Description Language
122	122	(VHDL) level.
123		-Since latency is not an issue in a openloop phase noise characterization instrument, the large
	123	+{\color{red}Since latency is not an issue in a openloop phase noise characterization instrument,
	124	+the large
124	125	numbre of taps in the FIR, as opposed to the shorter Infinite Impulse Response (IIR) filter,
125		-is not considered as an issue as would be in a closed loop system.
	126	+is not considered as an issue as would be in a closed loop system.} % r2.4
126	127
127	128	The coefficients are classically expressed as floating point values. However, this binary
128	129	number representation is not efficient for fast arithmetic computation by an FPGA. Instead,
...	...	@@ -143,8 +144,9 @@
143	144	relation between number of fiter taps and quantization, Fig. \ref{float_vs_int} exhibits
144	145	a 128-coefficient FIR bandpass filter designed using floating point numbers (blue). Upon
145	146	quantization on 6~bit integers, 60 of the 128~coefficients in the beginning and end of the
146		-taps become null, making the large number of coefficients irrelevant and allowing to save
147		-processing resource by shrinking the filter length. This tradeoff aimed at minimizing resources
	147	+taps become null, {\color{red}making the large number of coefficients irrelevant: processing
	148	+resources % r1.1
	149	+are hence saved by shrinking the filter length.} This tradeoff aimed at minimizing resources
148	150	to reach a given rejection level, or maximizing out of band rejection for a given computational
149	151	resource, will drive the investigation on cascading filters designed with varying tap resolution
150	152	and tap length, as will be shown in the next section. Indeed, our development strategy closely
151	153
...	...	@@ -182,22 +184,24 @@
182	184
183	185	Addressing only two operations allows for demonstrating the methodology but should not be
184	186	considered as a limitation of the framework which can be extended to assembling any number
185		-of skeleton blocks as long as perfomance and resource occupation can be determined. Hence,
186		-in this paper we will apply our methodology on simple DSP chains: a white noise input signal
187		-is generated using a Pseudo-Random Number (PRN) generator or thanks at a radiofrequency-grade
188		-Analog to Digital Converter (ADC) loaded by a 50~$\Omega$ resistor. Once samples have been
	187	+of skeleton blocks as long as perfomance and resource occupation can be determined. {\color{red}
	188	+Hence,
	189	+in this paper we will apply our methodology on simple DSP chains: a white noise input signal % r1.2
	190	+is generated using a Pseudo-Random Number (PRN) generator or by sampling a wideband (125~MS/s)
	191	+14-bit Analog to Digital Converter (ADC) loaded by a 50~$\Omega$ resistor.} Once samples have been
189	192	digitized at a rate of 125~MS/s, filtering is applied to qualify the processing block performance --
190	193	practically meeting the radiofrequency frontend requirement of noise and bandwidth reduction
191	194	by filtering and decimating. Finally, bursts of filtered samples are stored for post-processing,
192	195	allowing to assess either filter rejection for a given resource usage, or validating the rejection
193	196	when implementing a solution minimizing resource occupation.
194	197
195		-The first step of our approach is to model the DSP chain and since we just optimize
196		-the filtering, we have not modeling the PRN generator or the ADC. The filtering can be
197		-done by two ways. The first one we use only one FIR filter with lot of coefficients
198		-to rejection the noise, we called this approach a monolithic approach. And the second one
199		-we select different FIR filters with less coefficients the monolithic filter and we cascaded
200		-it to filtering the signal.
	198	+{\color{red}
	199	+The first step of our approach is to model the DSP chain. Since we aim at only optimizing % r1.3
	200	+the filtering part of the signal processing chain, we have not included the PRN generator or the
	201	+ADC in the model: the input data size and rate are considered fixed and defined by the hardware.
	202	+The filtering can be done in two ways, either by considering a single monolithic FIR filter
	203	+requiring many coefficients to reach the targeted noise rejection ratio, or by
	204	+cascading multiple FIR filters, each with fewer coefficients than found in the monolithic filter.}
201	205
202	206	After each filter we leave the possibility of shifting the filtered data to consume
203	207	less resources. Hence in the case of cascaded filter, we define a stage as a filter
...	...	@@ -241,7 +245,11 @@
241	245	transfer function.
242	246	Comparing the performance between FIRs requires however defining a unique criterion. As shown in figure~\ref{fig:fir_mag},
243	247	the FIR magnitude exhibits two parts: we focus here on the transitions width and the rejection rather than on the
244		-bandpass ripples as emphasized in \cite{lim_1988,lim_1996}.
	248	+bandpass ripples as emphasized in \cite{lim_1988,lim_1996}. {\color{red}Throughout this demonstration,
	249	+we arbitrarily set a bandpass of 40\% of the Nyquist frequency and a bandstop from 60\%
	250	+of the Nyquist frequency to the end of the band, as would be typically selected to prevent
	251	+aliasing before decimating the dataflow by 2. The method is however generalized to any filter
	252	+shape as long as it is defined from the initial modelling steps.}
245	253
246	254	\begin{figure}
247	255	\begin{center}
...	...	@@ -83,14 +83,32 @@
83	83
84	84	{\bf
85	85	On page 2, "...allowing to save processing resource..." could be improved. % r1.1
	86	+}
86	87
	88	+The sentence was split and now reads ``number of coefficients irrelevant: processing
	89	+resources are hence saved by shrinking the filter length.''
	90	+
	91	+{\bf
87	92	On page 2, "... or thanks at a radiofrequency-grade..." isn't at all clear what % r1.2
88		-the author meant.
	93	+the author meant.}
89	94
90		-One page 2, the whole paragraph "The first step of our approach is to model..." % r1.3
	95	+Grammatical error: this sentence now reads ``or by sampling a wideband (125~MS/s)
	96	+Analog to Digital Converter (ADC) loaded by a 50~$\Omega$ resistor.''
	97	+
	98	+{\bf
	99	+On page 2, the whole paragraph "The first step of our approach is to model..." % r1.3
91	100	could be improved.
92	101	}
93	102
	103	+Indeed this paragraph has be written again and now reads as\\
	104	+``The first step of our approach is to model the DSP chain. Since we aim at only optimizing
	105	+the filtering part of the signal processing chain, we have not included the PRN generator or the
	106	+ADC in the model: the input data size and rate are considered fixed and defined by the hardware.
	107	+The filtering can be done in two ways, either by considering a single monolithic FIR filter
	108	+requiring many coefficients to reach the targeted noise rejection ratio, or by
	109	+cascading multiple FIR filters, each with fewer coefficients than found in the monolithic filter.
	110	+''
	111	+
94	112	{\bf
95	113	I appreciate that the authors attempted and document two optimizations: that % r1.4 - en attente des résultats
96	114	of maximum rejection ratio at fixed silicon area, as well as minimum silicon
97	115
98	116
...	...	@@ -178,17 +196,36 @@
178	196	than FIR. This is not true in general. The sentence should be reconsidered.
179	197	}
180	198
181		-J'aurais du dire ``lag'' au lieu de ``impulse response'' je pense
182		-AH: Je ne comprends pas trop ce qui ne va pas ici
	199	+We have not stated that the IIR has a shorter impulse response but a shorter lag.
	200	+Indeed while a typical FIR filter will have 32 to 128~coefficients, few IIR filters
	201	+have more than 5~coefficients. Hence, while a FIR requires 128 inputs before providing
	202	+the first output, an IIR will start providing outputs only 5 time steps after the initial
	203	+input starts feeding the IIR. Hence, the issue we address here is lag and not impulse
	204	+response. We aimed at making this sentence clearer by stating that ``Since latency is not an issue
	205	+in a openloop phase noise characterization instrument, the large
	206	+numbre of taps in the FIR, as opposed to the shorter Infinite Impulse Response (IIR) filter,
	207	+is not considered as an issue as would be in a closed loop system in which lag aims at being
	208	+minimized to avoid oscillation conditions.
	209	+''
183	210
184	211	{\bf
185	212	- Fig. 4: the Author should motivate in the text why it has been chosen % r2.5
186	213	this transition bandwidth and if it is a typical requirement for phase-noise
187	214	metrology.
188	215	}
189		-AH: Je ne sais pas comment justifier ça. Je dois dire que comme ça on peut éventuellement
190		-décimer par deux le flux ?
191	216
	217	+The purpose of the paper is to demonstrate how a given filter shape can be achieved by
	218	+minimizing varous resource criteria. Indeed the stopband and bandpass boundaries can
	219	+be questioned: we have selected this filter shape as a typical anti-aliasing filter considering
	220	+the the dataflow is to be halved. Hence, selecting a cutoff frequency of 40\% the initial
	221	+Nyquist frequency prevents noise from reaching baseband after decimating the dataflow by a
	222	+factor of 2. Such ideas are now stated explicitly in the text as ``Throughout this demonstration,
	223	+we arbitrarily set a bandpass of 40\% of the Nyquist frequency and a bandstop from 60\%
	224	+of the Nyquist frequency to the end of the band, as would be typically selected to prevent
	225	+aliasing before decimating the dataflow by 2. The method is however generalized to any filter
	226	+shape as long as it is defined from the initial modelling steps: Fig. \ref{fig:rejection_pyramid}
	227	+as described below is indeed unique for each filter shape.''
	228	+
192	229	{\bf
193	230	- The impact of the coefficient resolution is discussed. What about the % r2.6 - fait
194	231	resolution of the data stream? Is it fixed? If so, which value has been
195	232
...	...	@@ -196,13 +233,15 @@
196	233	coefficient resolution?
197	234	}
198	235
199		-Pr\'eciser que le flux de donn\'ees en entr\'ees est de r\'esolution fixe
	236	+We have now stated in the beginning of the document that ``we have not included the PRN generator
	237	+or the ADC in the model: the input data size and rate are considered fixed and defined by the
	238	+hardware.'' so indeed the input datastream resolution is considered as a given.
200	239
201	240	{\bf
202	241	- Page 3, line 47: the initial criterion can be omitted and, consequently, % r2.7 - fait
203	242	Fig. 5 can be removed.
204		-- Page 3, line 55: “maximum rejection” is not compatible with fig. 4. % r2.8 - fait
205		-It should be “minimum”
	243	+- Page 3, line 55: ``maximum rejection'' is not compatible with fig. 4. % r2.8 - fait
	244	+It should be ``minimum''
206	245	}
207	246	AH: Je ne suis pas d'accord, le critère n'est pas le min de la rejection mais le max
208	247	de la magnitude. J'ai corrigé en ce sens.
...	...	@@ -214,7 +253,12 @@
214	253	- Page 4, line 10: how $p$ is chosen? Which is the criterion used to choose % r2.12 - fait
215	254	these particular configurations? Are they chosen automatically?
216	255	- Page 4, line 31: how does the delta function transform model from non-linear % r2.13 - fait
217		-and non-quadratic to a quadratic?
	256	+and non-quadratic to a quadratic?}
	257	+
	258	+JMF : il faudra mettre une phrase qui explique, ca en lisant cette reponse dans l'article
	259	+je ne comprends pas comment ca repond a la question
	260	+
	261	+{\bf
218	262	- Captions of figure and tables are too minimal. % r2.14
219	263	- Figures can be grouped: fig. 10-12 can be grouped as three subplots (a, b, c) % r2.15 - fait
220	264	of a single figure. Same for fig. 13-16.
...	...	@@ -226,10 +270,9 @@
226	270	differences among the curves. I suggest to reduce the noise below 1 dBpk-pk.
227	271	}
228	272
229		-Comment as tu fait tes spectres Arthur ? Si tu as fait une FFT sur e.g. 2048 points
230		-mais que tu as des jeux de donnees de e.g. 10000 points, on peut faire des moyennes
231		-sur les sequences successives. Au pire si pas possible, une moyenne glissante sur
232		-chaque spectre pour affiner les traits ?
	273	+Indeed averaging had been omitted during post-processing and figure generation: we
	274	+are grateful to the reviewer for emphasizing this point which has now been corrected. All spectra
	275	+now exhibit sub-dBpk-pl line thickness.
233	276
234	277	%In conclusion, my opinion is that the methodology presented in the Manuscript
235	278	%deserve to be published, provided that the criterion is changed according