new directions in channelized receivers and transmitters€¦ · version of edwin armstrong’s...
TRANSCRIPT
New Directions in Channelized Receivers and Transmitters
fred harris
2-December 2011
1
Motivation For Using Multirate Filters
2
Processing Task: Obtain Digital Samples
of Complex Envelope Residing at Frequency fC
Analog Digital
Multi-Channel FDM Input Signal
Single-Channel Base banded Output Signal
Receiver
Input Spectrum
ffC
Selec ted Narrow Band Signal
3
See!
Utah
4
First Generation DSP Receiver
f
f
f
f
Low Pass Filter
fs
-jte
1 2 3
3
4
4
2
1
AnalogSignal Processing
5
Signal Conditioning for DSP Receiver
f
f
ffs/2-fs/2
1
2
3
.... ....
Low Pass Filter
fs1 2 3 4
6
Replicate Analog Processing in DSP
Low Pass Filter
fs/M
M:1
-jne
5 6
f
f
f
f
fs/2
fs/2
fs/2
fs/M
-fs/2
-fs/2
-fs/2
-fs/M
-fs fs
3
4
5
6
....
....
....
....
....
....
....
....
Low Pass Filter
fs1 2 3 4
~ ~~ ~
IgnoringGood Advice!
7
ADC
M:1
s(t) s(n)
h(n)
r(n) r(nM)s(n)
e -j n
e -j n
LOCLK
LOWPASS FILTER
•Fundamental Operations•Select Frequency•Limit Bandwidth•Select Sample Rate
Digital Down Converter (DDC)
8
Spectral Description ofFundamental Operations
0
0
0
0
fs/2
fs/2
fs/2
fs/M
-fs/2
-fs/2
-fs/2
-fs/M
f
f
f
f
CHANNEL OF INTEREST
TRANSLATED SPECTRUM
FILTERED SPECTRUM
SPECTRAL REPLICATES AT DOWN-SAMPLED RATE
OUTPUT DIGITAL FILTER RESPONSE
INPUT ANALOG FILTER RESPONSE
.... ....
9
A Shell Game:Rearrange the Players!
Keep Your Eye on the Pea!
10
Signal and Filter are at Different FrequenciesWhich One to Move?
0
0
0
0
f
f
f
f
CHANNEL OF INTEREST
TRANSLATED SPECTRUM
TRANSLATED FILTER
FILTER RESPONSE
SecondOption
FirstOption
11
Down Sample Complex Digital IF
Band Pass Filter
fs/M
M:1
-jMnjn
ee
5 6
f
f
f
f
fs/2
fs/2
fs/M
fs/M
-fs/2
-fs/2
-fs/M
-fs/M
-fs
fs
3
4
5
6
....
....
....
....
....
....
....
....
Low Pass Filter
fs1 2 3 4
~
~
~
~
No Spectral
Image
12
Fundamental Operation with Rearrange Operators
ADCM:1
s(t)
h(n)
s(n) r(n)r(n) r(nM)
e -j ne j n
e j n
LOCLK
BANDPASS FILTER
Up-Convert Filter, Filter Signal at IF, Down Convert Output of Filter
13
Equivalency Theorem
0
0
0 0
0 0
( )
( ) ( ) * ( )( ) ( )
( ) ( )
{ ( )* ( ) }
j n
j n k
k
j n j k
k
j n j n
r n s n e h ks n k e h k
e s n k h k e
e s n h n e
Down-Convert Signal at Input to Low-Pass Filter
Down-Convert Signal at Output of Band-Pass Filter
Up-Convert Low Pass FilterTo Become Complex
Band-Pass Filter14
Signal Flow Description of Equivalency Theorem
Not Finished: Moving Down Converter from Input to Output Replaces 2-Multipliers (Complex Scalar) with 4-Multipliers (Complex Product)
15
Interchange Down Converter and Resampler
ADCM:1
s(t)
h(n)
s(n) r(nM)r(n)r(nM)
e -j Mne j n
e j n e j Mn
LOCLK
BANDPASS FILTER
Only Down Convert the Samples we Intend to Keep!Let the Resampler Alias the Center Frequency to Baseband
0 Aliases to M0 Mod(2)0 Aliases to 0 if M0 = k2
16
SPECTRAL DESCRIPTION REORDERED FUNDAMENTAL OPERATION
0
0
0
0
fs/2
fs/2
fs/M
fs/M
-fs/2
-fs/2
-fs/M
-fs/M
f
f
f
f
CHANNEL OF INTEREST
TRANSLATED FILTERFILTERED SPECTRUM
ALIASED REPLICATES AT DOWN-SAMPLED RATE
TRANSLATED REPLICATES AT DOWN-SAMPLED RATE
INPUT ANALOG FILTER RESPONSE
....
....
....
....
17
Successive Transformations Turn Sampled Data Version of Edwin Armstrong’s Heterodyne Receiver
to Tuned Radio Frequency (TRF) Receiverand then to Aliased TRF Receiver.
DigitalBand-Pass
M-to-1
e -j kn
H(Ze )-j k
DigitalLow-Pass
M-to-1
H(Z)
e-j kn
DigitalBand-Pass
M-to-1
H(Ze )-jk
k
k
M θ = k 2π2πor θ = kM
Any Multiple of Output Sample Rate Aliases to Baseband
Armstrong Nyquist Equivalency Theorem
DigitalBand-Pass
M-to-1
-j Mkne
H(Ze )-jk
18
Let’s Keep Rearranging the Players!
19
Linear Systems Commute and are Associative
x(n)
x(n)
(n)
y(n)
h(n)
w(n)
w(n)
r(n)
r(n)
R(f)= H(f)G(f)
h(n)
H(f)
h(n)
g(n)
G(f)
g(n)
f
f
f
x(n)
x(n)
(n)
v(n)
g(n)
w(n)
w(n)
r(n)
r(n)
R(f)= G(f)H(f)
h(n)
H(f)
h(n)
g(n)
G(f)
g(n)
f
f
f
20
Linear systems Are Associative
X
X
Y W
W
L1 L2
L = L L3 1 2
X
X
Y W
W
L1 L2
L = L L3 1 2
Filter Resample
Resampled Filter
21
In Case you Couldn’t Wait to See the Proof
22
2
1 1
2 2
1 2 1 2
3 3 2 2
3 3 2 2
3 3
2 2
1
2
2 1
2 3
2
3
3
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
where ( ) ( ) ( )
k
k
k k
k k
k k
k
k
y n x n k h k
dy n y n k g k
x n k k h k g k
x n k h k k g k
x n k h k k g k
x n k f k
f n h n k g k
Filter and Output Resampler can Commuteto Input Resampler and Resampled Filter
X(Z)
X(Z)
X(Z)
Y(Z)
Y(Z)
y(n)
y(n)
Y(Z )M
Y(Z )M
Y(Z )M
Z-rX(Z )
M
y(nM)
y(nM)
y(nM)x(nM+ r)
x(n)
x(n)
x(n)
H(Z)
M:1
M:1
M:1
Z H(Z )-r M
r
Z H(Z)-r
r
r
r
23
Coefficient Assignment of Low-Pass Polyphase Partition
Extract Delays To First Non-Zero Coefficient
For M-to-1 resample start at Index r and Increment by MFor 3-to-1 resample start at index r and increment by 3
C0
C0
C0
C3
C3
C3
C4
C4
C4
C5
C5
C5
C6
C6
C6
C7
C7
C7
C8
C8
C8
C9
C9
C9
C10
C10
C10
C11
C11
C11
C1
C1
C1
C2
C2
C2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
00
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
This mapping from 1-D to 2-D is used by Cooley-Tukey FFT. Polyphase Filters and CT-FFT are kissing cousins!
24
Polyphase Partition of Low Pass Filter
1-Path to M-Path Transformation
H( Z) x(n) y(n) y(nM)M:1
1
0( ) ( )
Nn
nH Z h n Z
1 1
( )
0 0( ) ( )
M Nr nM
r nH Z h r nM Z
1 1
0 0( ) ( )
M Nr nM
r nH Z Z h r nM Z
Z
Z
Z
Z
-1
-2
-(M-2)
-(M-1)
H ( Z )0M
H ( Z )1M
H ( Z )2M
H ( Z )M-2M
H ( Z )M-1M
....
.... ....
x(n) y(n) y(nM)M:1
M-Path Partition Supports M-to-1 Down Sample Also Supports Rational Ratio M-to-Q and M-to-Q/P Down Sample!
25
Noble Identity: Commute M-units of Delay
followed by M-to-1 Down SampleM Delays
1 Delay
M Delays
1 Delay
Input Clock, T
Output Clock, MT
Z Z-M -1
M:1 M:1M-Units of Delay at Input Rate Same as 1-Unit of Delay at Output Rate
H(Z ) H(Z )-M -1
M:1 M:1
Interchange Filters and Resampler: Place Resampler at Input
Rather Than at Output of Filter
Z
Z
Z
Z
-1
-2
-(M-2)
-(M-1)
H ( Z )0
H ( Z )1
H ( Z )2
H ( Z )M-2
H ( Z )M-1
....
.... ....
x(n) y(nM,k)
M:1
M:1
M:1
M:1
M:1
Replace Delays with CommutatorPerform Path Operations Sequentially
H ( Z )0
H ( Z )1
H ( Z )2
H ( Z )M-2
H ( Z )M-1
.... ....
x(n) y(nM)
8-tap
Coeffic ient BankSelect
IFSTAGE
RFSTAGE
IFSTAGE
DDS
DDS
S-P
P-S
Clock
Clock
CARRIER PLL
TIMING PLL
DAC
ADC
DIGITALLOW-PASS
DIGITALLOW-PASS
DIGITALLOW-PASS
DIGITALLOW-PASS
DIGITALLOW-PASS
DIGITALLOW-PASS
DIGITALLOW-PASS
DIGITALLOW-PASS
Shape &Upsample
Matched Filter
Interpola te
Dec ima te
I-Q Table
Detect
cos( n)
cos( n)
sin( n)
sin( n)
Modulator
Demodulator
f
Analog Analog
Analog
Osc illator
PA
RF Carrier
GainControl
Osc illator
CarrierWaveform
VGALNA
RF Carrier
IFChannel Channel
Bits
Bits
Transmitter Process, Up-Sample and Up-ConvertReceiver Process, Down Convert and Down-Sample
Modulator Raises Sample Rate &Applies Heterodyne at High Output Sample Rate!
De-Modulator Applies Heterodynes at High Input Rate & then Reduces
Conventional and Ubiquitous DDC
DDS
DIGITALLOW-PASS
DIGITALLOW-PASS
M:1
M:1
DDS
CIC
CIC
M:1 2:1
2:1
2:1
2:1M:1
Scale Factor
CIC Correc tion
CIC Correc tion
HalfBand
HalfBand
HalfBand
HalfBand
z z zz-1z-1z-1 -1 -1 -1- - -
zz-1 -1-
zz-1 -1-
M:1Integra tors Derivative Filters
Convert Two Parallel Paths into M Sequential Paths for each Path
DDS
DIGITALLOW-PASS
DIGITALLOW-PASS
M:1
M:1
……
……
……
0
0
1
1
M-1
M-1
DDS
DDS
8-tap
8-tap
Coeffic ient
Replace CIC with Cascade 2-to-1 Half Band FIR Filters
Filter Number 1 2 3 4 5 6 7 8 9 10 Total
Number Taps 3 3 3 3 7 7 7 7 11 19 70
Operations Per Filter
2‐A2‐Shifts
2‐A2‐Shifts
2‐A2‐Shifts
2‐A2‐Shifts
4‐A2‐Mult
4‐A2‐Mult
4‐A2‐Mult
4‐A2‐Mult
6‐A3‐Mult
10‐A5‐Mult
___
Adds Ref to Input
2 2/2 2/4 2/8 4/16 4/32 4/64 4/128 6/256 10/512 4.26
Mult Ref to Input
0 0 0 0 2/16 2/32 2/64 2/128 3/256 5/512 0.27
2 4 6 8 10 12 14 16 18 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1-80-60-40-20
0
5 10 15 20 25 30 35 40 45 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2-80-60-40-20
0
10 20 30 40 50 60 70 80 90 -4 -3 -2 -1 0 1 2 3 4-80-60-40-20
0
20 40 60 80 100 120 140 160 180 200 -8 -6 -4 -2 0 2 4 6 8-80-60-40-20
0
50 100 150 200 250 300 350 400 15 10 5 0 5 10 15-80-60-40-20
0
Impulse and Frequency Response of Last Stage Referred to Earlier Stages
Replace CIC with Cascade 2-to-1 Half Band Linear Phase IIR Filters
Filter Number 1 2 3 4 5 6 7 8 9 10 Total
Number Taps 1 1 1 1 3 3 3 3 3 4 23
Operations Per Filter
3‐A1‐Mult
3‐A1‐Mult
3‐A1‐Mult
3‐A1‐Mult
7‐A3‐Mult
7‐A3‐Mult
7‐A3‐Mult
7‐A3‐Mult
7‐A3‐Mult
9‐A4‐Mult
___
Adds Ref to Input
3/2 3/4 3/8 3/16 7/32 7/64 7/128 7/256 7/512 9/1024 3.25
Mult Ref to Input
1/2 1/4 1/8 1/16 3/32 3/64 3/128 3/256 3/512 4/1024 1.12
Impulse and Frequency Response of Last Stage Referred to Earlier Stages
5 10 15 20 25 30 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1-80-60-40-20
0
10 20 30 40 50 60 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2-80-60-40-20
0
20 40 60 80 100 120 -4 -3 -2 -1 0 1 2 3 4-80-60-40-20
0
50 100 150 200 -8 -6 -4 -2 0 2 4 6 8-80-60-40-20
0
100 150 200 250 300 350 400 450 15 10 5 0 5 10 15-80-60-40-20
0
0 2 4 6 8 10 12 14 16 18 20
Impulse Response
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Frequency Response
0 8 0 6 0 4 0 2 0 0 2 0 4 0 6 0 8 1
Group Delay
Impulse, Frequency, & Group Delay Response of 2-PathLinear Phase, Recursive Half-Band Filter
0 0
h1 h1
0 0
h5 h5
0 0 0
h7h7
0
h9 h9
01
h3h3
-
-0
0
1
1
Low-Pass
Low-Pass
High-Pass
High-Pass
Half Band Filter: h(n)
h(2n)
h(2n+ 1)
2-to-1
2-to-1 Resampling 2-Path Polyphase Filter and Digital Down-Converter
f-0.5 0.50 0.25-0.25
LP
HP
2-Point DFT
o-1Resampling 4-Path Down-Sample Polyphase Filter and 4-Point IFFT Extracts Signal Component
From One-of-Four Selected Nyquist Zones
Half Band FiltersCentered on
Cardinal DirectionsEach Reduces
BW 2-to-1 and Reduces
Sample Rate 2-to-1
0
2
1
3
Low-PassFilter
Pos FreqHilbert Filter
High-PassFilter
Neg FreqHilbert Filter
)
1
1
1
1H (Z )0
2
H (Z )12
Z H (Z )-1
22
Z H (Z )-1
32
4-Point IFFT
4-Path, 2-to-1 Down-Samplewith 4 Possible Trivial Phase Shifters
4-Path Polyphase FilterPath-0 Not Used
2.5-Multiplies per Input
4-Phase Rotatorsfsk/4: {c0 c1 c2 c3}fs0/4: {1 1 1 1}fs1/4: {1 j -1 -j}fs2/4: {1 -1 1 -1}fs3/4: {1 -j -1 j}
2-to-1Down-Sample
0
0
h1
h1
0
0
h5
h5
0
0 0
h13
h13
0
h9
h9
0
1
h17
h17
-
-
-
-
-
-
-
-
- -
-
--
-
-
-
-
-
-
-
c0
c 1
c 2
c 3
Four Bands Centered on the Cardinal Directions
Bands Centeredon 0 and 180 (DC and fS/2)
Alias To DC When Down-Sampled
2-to-1
Bands Centeredon +90 and -90 (+fS/4 and –fS/4)
Alias To fs/2 When Down-Sampled
2-to-1
Spectra: Four Half Band Filters on Unit Circle Showing Alias Free Pass, Transition, and Aliased Bands
Alias FreePass Band
Alias FreePass Band
TransitionBandwidth
TransitionBandwidth
Folded BandwidthDue to 2-to-1 Down Sample
Low-PassBand-0
Positive Freq-Pass Band-1
Negative Freq-PassBand-3
High-PassBand-2
Any NarrowbandSignal Must Reside
in One of the 4 Alias Free Band Intervals.
The Alias FreeBand Intervals Overlap!
Pole-Zero Diagrams of Four Nyquist Zone Filters
-2 -1 0 1 2
-2
-1
0
1
2
-2 -1 0 1 2
-2
-1
0
1
2
-2 -1 0 1 2
-2
-1
0
1
2
-2 -1 0 1 2
-2
-1
0
1
2
Frequency Responses of Four Nyquist Zone Filters
-0.5 0 0.5
-80
-60
-40
-20
0
Frequency Response
Log
Mag
(dB
)
-0.5 0 0.5
-80
-60
-40
-20
0
Frequency Response
-80
-60
-40
-20
0
Log
Mag
(dB
)
-80
-60
-40
-20
0
ctra of Signal Aliased to Different Sampled Data Frequencies in Successive 2-to-1 Sample Rate Reductions.
Most Efficient MultistageHalf-Band Digital Down-Converter
Channelizer 4-Path 2-to-1 Down Sample
Channelizer 4-Path 2-to-1 Down Sample
Channelizer 4-Path 2-to-1 Down Sample
Channelizer 4-Path 2-to-1 Down Sample
......
......
Filter
Channel Select DDS
0 5 10 15 20
0
0.5
1
Impulse Response: Half Band Filter 2 2 22.5 [1 ]2 4 8
1 1 12.5 2.5 [1 ]2 4 8
2.5 3 7.5
N
RL I-Q I-Q I-Q
Spectrum of Input Signal and Zoom to Spectral Segment
0 5 10 15 20 25 30 35 40 45100
-50
0
Frequency / MHz
Received Signal Spectrum
17.9 17.92 17.94 17.96 17.98 18 18.02 18.04 18.06 18.08 18.1100
-50
0
F / MH
Zoom-in
Spectra: Last Four Stages Processing Chain. Dotted Line Indicates Center Frequency of
Desired Spectral Component
0.2 -0.1 0 0.1 0.2 0.3Frequency/MHz
Stage 6
-0.15 -0.1 -0.05 0 0.05 0.1 0.15-100
-80
-60
-40
-20
0
Frequency/MHz
Stage 7
-0.05 0 0.05-100
-80
-60
-40
-20
0
Frequency/MHz
Stage 8
-0.04 -0.02 0 0.02 0.04-100
-80
-60
-40
-20
0
Frequency/MHz
Stage 9
+0.56= -0.440.283
-0.86= 0.14
ampled Data Frequency Locations on Successive Aliases
-1 -0.5 0 0.5 1
-1
-0.5
0
0.5
1
-1 -0.5 0 0.5 1
-1
-0.5
0
0.5
1
-1
-0.5
0
0.5
1
-1
-0.5
0
0.5
1
pectrum at Input and Output of Final Heterodyne and Filter Stage
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04-100
-50
0
Frequency / MHz
Mag
nitu
de /
dB
Post Processing--Input Signal Spectrum (ch # 600)
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04-100
-50
0
Frequency / MHz
Mag
nitu
de /
dB
Post Processing--Down Convert (ch # 600)
-100
-50
0
Mag
nitu
de /
dB
Post Processing--Filtering (ch # 600)
A 375-to-1 down-sample: 90 MHz to 240 kHz with a 30 kHz output BW 80 dB dynamic range.Require 6 CIC stages. The gain of each stage is 375: Gain of 6 stages becomes (375)6 or 2.8 .1015 or 52 bits growth in the CIC integrators. With 16-bit input data integrator bit width is 16+52 or 68. Six integrators in both I & Q paths would be circulating 816 bits per input sample which if converted to the 16-bit width required of the arithmetic in the half-band filters proves to be same number of bits to manipulate 48 arithmetic operations per input sample.
Number of operations for the I-Q half band filter chain is on the order of 8-multiply and 16 adds per input sample which represents a workload 1/6 of the CIC chain. The efficient cascade CIC filter chain can be replaced with an even more efficient cascade four-path half band filter chain.
Linear Phase IIR Filter
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
8
0
2Group Delay: Two Path, 4-Coefficient, Linear Phase 2-Path Filter
0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.59
8
1Group Delay: Detail
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.50
0
0
Frequency response
Most Efficient MultistageHalf-Band Digital Down-Converter
Channelizer 4-Path 2-to-1 Down Sample
Channelizer 4-Path 2-to-1 Down Sample
Channelizer 4-Path 2-to-1 Down Sample
Channelizer 4-Path 2-to-1 Down Sample
......
......
Filter
Channel Select DDS
2 2 22.0 [1 ]2 4 8
1 1 12.0 2.0 [1 ]2 4 8
2.0 3 6.0
N
RL I-Q I-Q I-Q
5 10 15 20 25 30
Impulse Response, Two-Path, 4-Coefficient, Linear Phase IIR
Polyphase Partition of Band Pass Filter1-Path to M-Path Transformation
21 1
0 0
( ) ( )M Nj k r nMM
r n
G Z e Z h r nM Z
1 1
0 0( ) ( ) ( ) ( ) ( )k k k
N Nj n j jn n
n nG Z h n e Z h n e Z H e Z
1 1
( ) ( )
0 0( ) ( ) k
M Nj r nM r nM
r nG Z h r nM e Z
1 1
0 0( ) ( )k k
M Nj j nMr nM
r nG Z e Z h r nM e Z
k
k
θ = k 2π2πθ = kM
Modulation Theorem of Z-Transform
Polyphase Band Pass Filter and M-to-1 Resampler
Z
Z
Z
Z
-1
-2
-(M-2)
-(M-1)
H ( Z )0M
H ( Z )1M
H ( Z )2M
H ( Z )M-2M
H ( Z )M-1M
....
.... ....
x(n) y(n) y(nM)M:1e
e
e
e
e
j k0
j k1
j k2
j k(M-2)
j k(M-1)
2
2
2
2
2
M
M
M
M
M
Apply Noble Identity to Polyphase Partition
Z
Z
Z
Z
-1
-2
-(M-2)
-(M-1)
H ( Z )0
H ( Z )1
H ( Z )2
H ( Z )M-2
H ( Z )M-1
....
.... ....
x(n) y(nM,k)
M:1
M:1
M:1
M:1
M:1
e
e
e
e
e
j k0
j k1
j k2
j k(M-2)
j k(M-1)
2
2
2
2
2
M
M
M
M
M
We Reduce Sample Rate M-to-1 Prior to Reducing Bandwidth
(Nyquist is Raising His Eyebrows!)
We Intentionally Alias the Spectrum.(Were you Paying Attention
in school when they discussed the importance of anti-aliasing filters?)
M-fold Aliasing! M-Unknowns!M-Paths supply M-EquationsWe can the separate Aliases!
Move Phase Spinners to
Output of Polyphase Filter Paths
Z
Z
Z
Z
-1
-2
-(M-2)
-(M-1)
H ( Z )0
H ( Z )1
H ( Z )2
H ( Z )M-2
H ( Z )M-1
....
.... ....
x(n) y(nM,k)
M:1
M:1
M:1
M:1
M:1
e
e
e
e
e
j k0
j k1
j k2
j k(M-2)
j k(M-1)
2
2
2
2
2
M
M
M
M
M
W t Ph S i f f l ibl
Polyphase Partition with Commutator Replacing the “r” Delays in the “r-th” Path
H ( Z )0
H ( Z )1
H ( Z )2
H ( Z )M-2
H ( Z )M-1
.... ....
x(n) y(nM,k)e
e
e
e
e
j k0
j k1
j k2
j k(M-2)
j k(M-1)
2
2
2
2
2
M
M
M
M
M
Note: We don’t assignPhase Spinners to Select Desired Center FrequencyTill after Down SamplingAnd Path Processing
This Means thatThe Processing for every Channelis the same till the Phase Spinner
No longer LTI, Filter now hasM-Different Impulse Responses!Now LTV or PTV Filter.
Armstrong to Tuned RF with Alias Down Conversion to Polyphase Receiver
DigitalBand-Pass
M-to-1
H(Ze )-jk
DigitalLow-Pass
M-to-1
H(Z)
e-j kn
Rather than selecting center frequency at input and reduce sample rate at output, we reverse the order, reduce sample rate at input and select center frequency at output. We perform arithmetic operations at low output rate rather than at high input rate!
M-Path DigitalPolyphase
M-to-1
H(Z)r
e-j 2
M rk
Down Sample 6-to-1
n n-1 n-2 n-3 n-4 n-5 n-6 n-7 n-8 n-9
n n-1 n-2 n-3 n-4 n-5 n-6 n-7 n-8
n n-1 n-2 n-3 n-4 n-5 n-6 n-7 n-8 n+ 6 n+ 5 n+ 4 n+ 3 n+ 2 n+ 1
n-10 n-11 n-12 n-13 n-14
— — — — — —
Polyphase Partition 1-D filter becomes2-D M-Path Filter
n
n-1
n-2
n-3
n-4
n-5
n
n-1
n-2
n-3
n-4
n-5
n
n-1
n-2
n-3
n-4
n-5
n-12
n-13
n-14
n-15
n-16
n-17
n-12
n-13
n-14
n-15
n-16
n-17
n-12
n-13
n-14
n-15
n-16
n-17
n+ 6
n+ 5
n+ 4
n+ 3
n+ 2
n+ 1
n-6
n-7
n-8
n-9
n-10
n-11
n-6
n-7
n-8
n-9
n-10
n-11
n-6
n-7
n-8
n-9
n-10
n-11
——————
——————
Reorder Filter and Resample
ADC
s(t)
h(r+ nM)
h(M-1+ nM)
h(0+ nM)
h(1+ nM)
s(n)
r(nM)
r(nM,k)
CLK
BANDPASS FILTERPOLYPHASE PARTITION
LOWPASS FILTERPOLYPHASE PARTITION
PHASE ROTATORSALIASED HETERODYNE
ej r kM
... ...
.....
.....
.. this is verystuff....
Phase and Gain Response
(3-Versions of Filter)
Prototype Filter,
Polyphase Filter Prior to Resampling,
Polyphase Filter after Resampling
Impulse Response and Frequency Response of Prototype Low Pass FIR Filter
Impulse Response of 6-Path PolyphasePartition Prior to 6-to-1 Resampling
Frequency Response of 6-Path PolyphasePartition Prior to 6-to-1 Resampling
Phase Response of 6-Path PolyphasePartition Prior to 6-to-1 Resampling
Overlay Phase Response of 6-Path Polyphase Partition Prior to 6-to-1 Resampling
0 Nyquist
Zone +1 Nyquist
Zone +2 Nyquist
Zone
-1 Nyquist
Zone
-2 Nyquist
Zone
Phase Aligned 2/6
PhaseShifts
-2/6PhaseShifts
-2/3PhaseShifts
2/3PhaseShifts
De-Trended Overlay Phase Response: 6-Path Partition Prior to 6-to-1 Resampling
3-D Paddle-Wheel Phase Profiles, 6-Path Partition Prior to 6-to-1 Resampling
Overlay 3-D Paddle-Wheel Phase Profiles, 6-Path Partition Prior to 6-to-1 Resampling
Overlay 3-D Paddle-Wheel Phase Profiles, Showing Phase Shifts in +1 Nyquist Zone
Overlay 3-D Paddle-Wheel Phase Profiles, Phase Shifted to Align Phases in +1 Nyquist Zone
PolyChanDemo
Single Channel Armstrong and Multirate Aliased Polyphase Receiver
H(Z )
H (Z )
y(nM,k)
y(nM,k)y (nM)
y(n,k)
e
M-to-1
M-to-1
1
1
2
2
2 2
M M
x(n)
x(n) x(n)
e M
M
j nk
j nk
2
2
rr
Standard DDC
Polyphase DDC
2-Polyphase Filters
1-Polyphase Filter
ide Input Down Conversion to Output of Filter Where it anishes Due to Down Sampling. Rotators in Filter Factor Out and are Applied to Path Outputs Rather than to
Coefficients. Advantage: Real sequence is made complex at output of
Filter Rather than at Input to Filter
……
……
……
0
0
1
1
M-1
M-1
DDS
H (Z )
H (Z )
H (Z )
H (Z )
0
1
2
M-1
......
x(n) y(nM,k)
e Mj 1k2
e Mj 0k2
e Mj 2k2
e Mj (M-1)k2
Bad Mismatch: Sample Rate Large Compared to Transition Bandwidth
Nyquist Rate for Filter is 200 kHz + 200 kHz = 400 kHz or fs/50
200 kHz
80 dB
0.1 dB
200 kHz
20 MHz
f-6 dB/Octave
400 TapFIR Filter
20 MHz InputSample Ra te
20 MHz Output Sample Rate
Polyphase Partition of Low‐Pass Filter
… … …0
1
2
49
48
50-to-1
400 Taps
8 Taps
20 kHz
20 MHz
400 kHz
400 kHz
PolyphaseLow Pass Filter
Cascade Polyphase Filter Down‐Sampling and Up‐Sampling
… …… ………
0 0
1 1
2 2
49 49
48 48
50-to-1 1-to-50
400 Taps 400 Taps20 MHz 20 MHz
20 MHz 20 MHz
400 kHz
400 kHz
PolyphaseLow Pass Filter
PolyphaseLow Pass Filter
8-taps 8-taps
8-tap 8-tap
20 MHz 20 MHz400 kHz
Coeffic ient Bank
Coeffic ient BankSelec t Selec t
Efficient Polyphase Filter Implementation
400 TapFIR Filter
20 MHz InputSample Ra te
20 MHz Output Sample Rate
16‐Ops/Input
60‐Ops/Input
Two Processing in Boxes:ow can you tell which is which from outside box?
400-TapLowpass Filter
20MHz
20MHz
20MHz
400 kHz20MHz8-Tap Filter
8-Tap Filter
Coeffic ient Bank
Coeffic ient Bank
State MachineSelec tSelec t
White Box
White Box
Polyphase Partition of Low‐Pass Filter
H (Z ) H (Z )
H (Z ) H (Z )
H (Z ) H (Z )
H (Z ) H (Z )
0 0
1 1
2 2
M-1 M-1
......
......
x(n)y(nM)
y(n)
f f f
Polyphase Partition of Band Pass Filter
H (Z ) H (Z )
H (Z ) H (Z )
H (Z ) H (Z )
H (Z ) H (Z )
0 0
1 1
2 2
M-1 M-1
......
......
x(n)y(nM,k)
y(n,k)
e Mj 1k2
e Mj 1k2
e Mj 0k2
e Mj 0k2
e Mj 2k2
e Mj 2k2
e Mj (M-1)k2
e Mj (M-1)k2
f f f
olyphase Partition of Two Band Pass Filters
H (Z ) H (Z )
H (Z ) H (Z )
H (Z ) H (Z )
H (Z ) H (Z )
0 0
1 1
2 2
M-1 M-1
......
......
y(nM,k )1
y(nM,k )2
y(n,k )+ y(n+ k )1 2
e Mj rk12 e Mj rk2
2
e Mj rk12 e Mj rk2
2
f
Workload for Multiple M-Path Filters
• 1-Channel M-to-1 Down Sample• 1-Filter and M Complex Phase Rotators
• 2-Channels M-to-1 Down Sample• 1-Filter and 2M Complex Phase Rotators
• K-Channels M-to-1 Down Sample• 1-Filter and kM Complex Phase Rotators
• M-channels M-to-1 Down Sample (use FFT)• 1-Filter and [log2(M)/2]M Complex Phase Rotators
hen k > Log2(M)/2 Build all channels and discard the channels you don’t need!M=16, Log2(16)/2 = 2: thus if you want 2 or more, Build them all!M=128, Log2(128)/2 = 3.5: thus if you want 4 or more, Build them all!M=1024, Log2(1024)/2 = 5: thus if you want 5 or more, Build them all!
M-Channel Channelizer: Resampled M-Path Narrowband Filter Channels Alias to Baseband: Phase Aligned Sums Separate Aliases: rk Performed at Low Output Rate Rather Than at High Input Rate.
One Input Filter Services M-Output Channels
fs
h (n)0
h (n)2
h (n)= h(r+ nM)r
Polyphase Pa rtition
h (n)M-2
h (n)1
h (n)3
h (n)M-1
FDM TDM
M-PNT IFFT
..... ..........
.....
.. this is verystuff....
Dual Channel Armstrong and Multirate Aliased Polyphase Receiver
H(Z )
H(Z )
H (Z )
y(nM,k )
y(nM,k )
y(nM,k )
y(nM k )
y (nM)
y(n,k )
y(n,k )
e
e
M-to-1
M-to-1
M-to-1
1
1
1
2
2
2
2
2
2
2
M M
x(n)
x(n) x(n)
e
e
M
M
M
M
j nk
j nk
j nk
j nk
2
2
2
2
rr
Standard DDC
Polyphase DDC
1
1
1
1
1
2
2
2
2
2
2-Polyphase Filters
2-Polyphase Filters
1-Polyphase Filter
Up-sampling by Zero Packing and Filtering
Spectra Of Input, of Zero-Packed, and of Low-Pass Filtered Zero-Packed Signal
y(m)
x(m)
x(n)
0
0
0
6
6
3
3
3
9
9
1
1
2
2
1
7
7
4
4
4
10
10
20
20
2
8
8
5
5
5
25
25
15
15
m
n
m
fs
fs
fs
f
f
f
5
5
....
....
....
....
....
....
FilterRejec tedSpec trum
Rejec tedSpec trum
Spectra Of Input, of Zero-Packed, and of Band Pass Filtered Zero-Pack Signal
m)
m)
n)
0
0
63
3
91 2
1
74
4
10 20
2
85
5
2515
m
n
m
fs
fs
fs
f
f
f
5
5
.... .... ....
FilterRejec tedSpec trum
Rejec tedSpec trum
1
0( ) ( )
Nn
nH Z h n Z
C0 C3 C4 C5 C6 C7 C8 C9 C12C10 C13C11 C14C1 C2
1-to-5
11( )
0 0( ) ( )
NM M
r nM
r nH Z h r nM Z
C0
C3
C4
C5
C6
C7
C8
C12
C9
C13
C10
C14
C11C1
C2
0
0
0 0
0
0
0
0
0
0 0
0
0
0
00
0
00
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 0
0
0
0
0
0 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1-to-5
Polyphase Partition of Resampling Filter
11
0 0
( ) ( )
NM M
r nM
r n
H Z Z h r nM Z
C0
C3
C4
C5
C6
C7
C8
C12
C9
C13
C10
C14
C11C1
C2
0 0 0
0 0
0
0
0 0
0
0
0
00
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0 0
0
0
0
0
0 0
0
0
0
0
0
0
00
0
0
0
0
0
0
0
0
0
0
0
1-to-5
C0
C3
C4
C5
C6
C7
C8
C12
C9
C13
C10
C14
C11C1
C2
0 0 0
0 0
0
0
0 0
0
0
0
00
0
0
0
00
0
0
0
0
0
0
0
0
0 0
0
0
0
0
0 0
0
0
0
0
0
0
0 0
0
0
0
0
0
0
0
1-to-5
Factor Delays and Rearrange
M Delays
1 Delay
M Delays
1 Delay
Input Clock, T
Output Clock, MT
Z
H(Z )
Z
H(Z )
-M
-M
-1
-1
M:1
M:1
M:1
M:1x(n) y(m) y(m)x(n)
Noble Identity: Interchange M-Delays with M-to-1 Resample
C0
C3
C4
C5
C6
C7
C8
C12
C9
C13
C10
C14
C11C1
C2
0
0
0
0
0
0
0
0
0
0
1-to-5
1-to-5
1-to-5
1-to-5
1-to-5
C0
C3
C4
C5
C6
C7
C8
C12
C9
C13
C10
C14
C11C1
C2
Interchange Filter and Resampler
Replace Up Samplers, Delays, and Summerwith M-Port Output
Commutator
21
0
( ) ( )N j k n nM
n
G Z h n e Z
1 21 ( ) ( )
0 0
12 21
0 0
121
0 0
( ) ( )
( ) ( )
( ) ( )
NM M j r nM k r nMM
r n
NM Mj r k j nMr nMM M
r nN
M Mj r kr nMM
r n
G Z h r nM e Z
G Z Z e h r nM e Z
G Z Z e h r nM Z
Low-Pass Replaced by Band-Pass
Spin The Delays, Don’t Touch the M-Path Partitioned Weights
Low-Pass to Band-Pass1-to-M Up-Sampling Filter
M-Path, M-Channel Channelizer: Spinners are in IFFT
......
fs
fBW
fsM
f
F0 F1 F2 FM-2 FM-1
M-Point IFFT Supplies Phase Spinners to FormUp Converters to all Multiples of Input Sample Rate
All Output Channels Centered on Multiples of Input Sample Rate
Example: Multiples of 6-MHz
h (5n)0 h (5n)0
h (5n)2 h (5n)2
h (5n)4 h (5n)4
h (5n)1 h (5n)1
h (5n)3 h (5n)3
x(n)y(m,k) y(m,k)
e e
e
j j
j n
2 25 5kr kr
Heterodyne Input Signal a Small Frequency Offset from DC: Channelizer Aliases DC to Channel Center and offset signal from DC is
Offset from Channel Center
Input Offset Frequency
Input Offset Observed at Output
y(m)
x(m)
x(n)
x(n)
0 31 42 5
m
n
n
m
fs
fs
fs
fs
f
f
f
f
5
5
FilterRejec tedSpec trum
Rejec tedSpec trum
Input Spectrum
Shifted Spectrum
Up-Sampled Spectrum
Distorted Spectrum
y(m)
x(m)
x(n)
x(n)
0 31 42 5
m
n
n
m
fs
fs
fs
fs
f
f
f
f
5
5
2 FiltersRejec tedSpec trum
Rejec tedSpec trum
Input Spectrum
Shifted Spectrum
Up-Sampled Spectrum
Two Filter Spectra
M-Channel Polyphase Channelizer:
M-path Filter and M-point FFT
fs
h (n)0
h (n)2
h (n)= h(r+ nM)r
Polyphase Partition
h (n)M-2
h (n)1
h (n)3
h (n)M-1
FDM TDM
M-PNT FFT
..... ..........
Critically SampledfS=fC
NyquistSampling
fS=2fC
QRT NyquistFilterfS=2fC
QRT NyquistFilterfS=4fC
NyquistSampling
fS=4fC
Various Filter-Channelizer Configurations
f
f
f
f
f
f
f
f
f
f
fs
fs
fs
fs
fs
f = fs
f = fs
f = fs
f = fs
f = fs
f = fsBW
f = fsBW
f = fsBW
f = 2fsBW
f = fsBW
f = fsBW
f = fsBW
f = fsBW
ch(k-1)
ch(k-1)
ch(k-1)
ch(k-1)
ch(k-1)
ch(k+ 1)
ch(k+ 1)
ch(k+ 1)
ch(k+ 1)
ch(k+ 1)
ch(k)
ch(k)
ch(k)
ch(k)
ch(k)
0.8
2.0
2.0
4.0
1.0
1.0
1.0
1.0
2.0
0.1 dB
0.1 dB
3.0 or 6.0 dB
6.0 dB
0.1 dB
f f
BW BW
Sample Rate= 2 Channel SpacingSample Rate = Channel SpacingChannel Spac ing Channel Spac ing
AliasedTransition Bandwidth
Filter Sampled at Rate to Avoid Band Edge Aliasing
Prototype Low‐Pass Filter for 120 Channel Channelizer
(M 1)
H ( Z )02
H ( Z )12
H ( Z )22
H ( Z )2
H ( Z )2
H ( Z )2
2
.... ....
y(n )y(n) M2
Z
Z
Z
Z
Z
-1
-2
....
....
....
.... ....
....
M
MM
M
MM
2
22
2
22
M2
:1
-( - 1)
-( 1)+ ( 1)+
( - 1)
-( )( )
M‐Path Polyphase Filter andM/2‐to‐1 Down Sampling
(M 1)
H ( Z )02
H ( Z )12
H ( Z )22
H ( Z )2
H ( Z )2
H ( Z )2
.... ....
y(n )M2
Z
Z
Z
Z
Z
-1
-2
....
....
....
.... ....
....
M
M
MM
M
MM
2
2
22
2
22
:1
M2
:1
M2
:1
M2
:1
M2
:1
M2
:1
M2
:1
-( -1)
-( 1)+ ( 1)+
( -1)
-( )( )
Use Noble Identity to PullM/2‐to‐1 Resampler Through Path Filter
Path Filters:Polynomials in
ZMConverted toPolynomials in
Z2
H ( Z )02
H ( Z )12
H ( Z )22
H ( Z )2
H ( Z )2
H ( Z )2
H ( Z )2
.... ....
y(n )M2
Z
Z
Z
Z
Z
Z
-1
-1
-1
-1
-1
-2
....
....
....
....
....
M
M
M
M
M
M
2
2
2
2
2
:1
M2
:1
M2
:1
M2
:1
M2
:1
M2
:1
M2
:1
-( -1)
-( -1)
( 1)+
( -1)
( )
Use Noble Identity to Pull M/2‐to‐1 ResamplerThrough Delays in Lower Half of Paths
Z
Z
Z
-1
-1
-1
H ( Z )0
H ( Z )1
H ( Z )2
H ( Z )( -1)
H ( Z )( )
H ( Z )M-2
H ( Z )M 1
....
....
........
....
x(n)
y(n )M2
M2
M2
2
2
2
2
2
2
2
Replace Delays and M/2‐to‐1 Resamplers withDual Input M/2 Path Commutator
H (Z )k2
Z H (Z )-1 2(k+ M/2)
h(k)
h(k+M/2)
h(k+M)
h(k+3 )
h(k+2M)
h(k+5 )
h(k+3M)
h(k+7 )
h(k+4M)
h(k+9 )M2
M2
M2
M2
Fold Unit Delays in lower half Filter PathsInto Filter Polynomials in Z2
fs
f
fs
fs
fs
fs
fs
fs
fs
fs
M M
M
M
M
M
M
M
M
2
2
-2
-2
-
-
M-to-1 Resample
M/2-to-1 Resample
.....
.....
.....
.....
f
f
M‐to‐1 Down Sample Aliases Multiples ofOutput Sample Rate to DC
M/2‐to‐1 Down Sample Aliases Odd Multiples ofOutput Sample Rate to Half sample Rate
0 5T T
1.5T
T
T
0.5T
0.5T
0mT
(m+1)T
fs
2 fs/M M Path PolyPhaseFilter in Z
M-PNT IFFT
.....
flg=0
flg=0
flg=1
flg=1
Circular Buffer
2
Circular Buffer Between Polyphase Filter and IFFTAligns Shifting Input Origin with IFFT’s Origin
0 01 12 2
3
M/2-1
M-1
M/2M/2+ 1
M-2M-1
M MM
....
......
..
….
M-P
ath
Polyp
hase
Filte
r
M-P
oint
IFFT
FDM
TDM
M-P
ath
Inpu
t Dat
a Bu
ffer
C
ircul
ar O
utpu
t Buf
fer
State Engine
M/2‐to‐1 Analysis Channelizer
001122
3
M/2-1
M-1
M/2
M/2+ 1
M-2M-1
MM M
....
......
..
….
M-P
ath
Polyp
hase
Filte
r
M-P
oint
IFFT
FDM
TDM
M-P
ath
Inpu
t Dat
a Bu
ffer
C
ircul
ar O
utpu
t Buf
fer
State Engine
1‐to‐M/2 Synthesis Channelizer
....
.... .... .... ....
....
....
MPoint IFFT
MPoint IFFT
MPathFilter
MPathFilter
nM/2 Point Shift M-PointCircular Buffer
nM/2 Point Shift M-PointCircular Buffer
xx
xx
Frequency Domain Filtering With Cascade M/2‐to‐1 Analysis and 1‐to‐M/2 Synthesis Channelizers
Impulse response and Frequency Response25‐Enabled Ports: 2.4 MHz Bandwidth
Impulse Response and Frequency Response40‐Enabled Ports: 3.9 MHz Bandwidth
Mixed, Arbitrary BandwidthChannelizers
Mixed Bandwidth Signals presented to Channel Synthesizer
ompose Broadband Signals Using Short Analysis rs and Present Components to Synthesizer
....
....
....
....
....
....
....
....
....
....
....
....
Sele
ctor
Input Channel Configuration
MM MMNN
Anal. 0
Anal. N-1
Bw0P0
PN-1BWN-1
2-to-M, M-Channel Synthesis Channelizer2fs Mfs
01
2
M/2-1
M/2
M/2+ 1
M-1
....
….
M-P
ath
Polyp
hase
Filte
r
M-P
oint
IFFT
M-P
ath
Inpu
t Dat
a Bu
ffer
C
ircul
ar O
utpu
t Buf
fer
State Engine
AnalysisChannleizers
0-MHz Input Signal Partitioned into ive 10-MHz Sub Channels: fS=20 MHz
Multiple Partitioned Input Bands Presented to Synthesizer
Assembled Multiple BW Channels in Single Synthesis Channelizer
Reassemble Decomposed Broadband Signals Using Short Synthesis Filters formed by Multiple Channel Analysis Channelizer
01
2
M/2-1M/2
M/2+ 1
M-1
....
....
....
....
....
....
....
....
....
....
....
....
....
….
M-P
ath
Polyp
hase
Filte
r
M-P
oint
IFFT
Sele
ctor
M-P
ath
Inpu
t Dat
a Bu
ffer
C
ircul
ar O
utpu
t Buf
fer
State Engine
M M M M NN
Synth. 0
Synth. N-1
Bw0P0
PN-1BWN-1
M/2-to-1, M-Channel Channelizer
fs
Partitioned spectral Components from ingle Multi-Channel Analyzer
-10 0 10-40
040
FLTR 0
-10 0 10-40
040
FLTR 1
-10 0 10-40
040
FLTR 2
-10 0 10-40
040
FLTR 3
-10 0 10-40
040
FLTR 4
-10 0 10-40
040
FLTR 5
-10 0 10-40
040
FLTR 6
-10 0 10-40
040
FLTR 7
-10 0 10-40
040
FLTR 8
-10 0 10-40
040
FLTR 9
-10 0 10-40
040
FLTR 10
-10 0 10-40
040
FLTR 11
-10 0 10-40
040
FLTR 12
-10 0 10-40
040
FLTR 13
-10 0 10-40
040
FLTR 14
-10 0 10-40
040
FLTR 15
-10 0 10-40
040
FLTR 16
-10 0 10-40
040
FLTR 17
-10 0 10-40
040
FLTR 18
-10 0 10-40
040
FLTR 19
-10 0 10-40
040
FLTR 20
-10 0 10-40
040
FLTR 21
-10 0 10-40
040
FLTR 22
-10 0 10-40
040
FLTR 23
-10 0 10-40
040
FLTR 24
-10 0 10-40
040
FLTR 25
-10 0 10-40
040
FLTR 26
-10 0 10-40
040
FLTR 27
-10 0 10-40
040
FLTR 28
-10 0 10-40
040
FLTR 29
-10 0 10-40
040
FLTR 30
-10 0 10-40
040
FLTR 31
-10 0 10-40
040
FLTR 32
-10 0 10-40
040
FLTR 33
-10 0 10-40
040
FLTR 34
-10 0 10-40
040
FLTR 35
-10 0 10-40
040
FLTR 36
-10 0 10-40
040
FLTR 37
-10 0 10-40
040
FLTR 38
-10 0 10-40
040
FLTR 39
-400
40
FLTR 40
-400
40
FLTR 41
-400
40
FLTR 42
-400
40
FLTR 43
-400
40
FLTR 44
-400
40
FLTR 45
-400
40
FLTR 46
-400
40
FLTR 47
Reassembled Wide band Channels rom Short Synthesis Channelizers
-20 0 20-40
0
40
Sig 3
Mag
nitu
de (d
B)
-20 0 20-40
0
40
Sig 6
-20 0 20-40
0
40
Sig 8
-20 0 20-40
0
40
Sig 4
Frequency (MHz)
Mag
nitu
de (d
B)
-20 0 20-40
0
40
Sig 5
Frequency (MHz)-20 0 20
-40
0
40
Sig 7
Frequency (MHz)
-30 -20 -10 0 10 20 30-40
0
40
Sig 1
F (MH )
Mag
nitu
de (d
B)
-30 -20 -10 0 10 20 30-40
0
40
Sig 2
F (MH )
Signal Fidelity Preserved under Multiple Sub-Channel Disassembly and Reassembly
....
.... .... ....
....
....
MPoint IFFT
MPoint IFFT
MPathFilter
MPathFilter
nM/2 Point Shift M-PointCircular Buffer
nM/2 Point Shift M-PointCircular Buffer
Cascade M/2‐to‐1 Analysis and 1‐to‐M/2 Synthesis Channelizers
Frequency Domain Filtering and Spectral Shuffle
HERE POINTED HAIR BOSS. THIS REPORT EXPLAINS HOW A SMALL FREQUENCY OFFSET AT THE INPUT SAMPLE RATE IS CONVERTED TO THE SAME FREQUENCY OFFSET FROM THE CHANNEL CENTER FREQUENCY AT THE HIGH OUTPUT SAMPLE RATE.
Suspicions Confirmed!
Is There any Doubt???