Low Complexity, Reconfigurable Digital Filters and Filter...
Transcript of Low Complexity, Reconfigurable Digital Filters and Filter...
![Page 1: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/1.jpg)
Low Complexity, Reconfigurable Digital
Filters and Filter Banks for Channelization
and Spectrum Sensing in Multi-Standard
Wireless Communication Receivers
Presented by:
Dr. Vinod A. Prasad,
Associate Professor, School of Computer Engineering,
Nanyang Technological University,
Singapore.
Email: [email protected]
Web: www.ntu.edu.sg/home/asvinod
24 June 2013
SINGAPORE
![Page 2: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/2.jpg)
A Word About NTU • NTU ranked 47
th in the most prestigious Quacquarelli
Symonds (QS) World University Rankings 2013.
• Ranked 2nd
in the world among universities below the age
of 50 by QS – 2012/2013.
• Ranked 8
th in the world among universities below the age
of 50 by Times Higher Education - 2013.
• World’s Biggest Engineering University with 1100 faculty
members in Engineering Schools alone.
• Rated as ‘5-star +’ University under QS – Only 7 Universities in the world achieved this
rating (NTU is the first and only Asian University in this league).
•Ranked as the 5th most-cited university with its research output ranked among the top
three universities globally in Engineering by Essential Science Indicators of Thomson
Reuters.
A fast rising
young University
(25 years old)
![Page 3: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/3.jpg)
Outline of today’s presentation
• Overview of Software Defined Radio and Cognitive Radio
• Challenges and Motivations
• Our Research Contributions
• Conclusions
3
SINGAPORE
![Page 4: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/4.jpg)
Introduction: Mobile Wireless
Communication
4
SINGAPORE
GSM
CDMA
WLAN
3G
LTE
Zigbee
Bluetooth
NFC
Voice Communication Data Connectivity
Communication with
peripheral devices
Wireless
Communication
Devices
![Page 5: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/5.jpg)
Introduction: Software Defined Radio
• Software defined radio (SDR) [1, 2] is a technology that has
been proposed as a solution to seamlessly support the
existing and upcoming wireless communication standards.
• Main characteristics:
– Flexible transmitter and receiver (transceiver) architecture,
– Digital Signal processing is able to replace, as much as possible,
analog processing to realize programmable radio functionalities,
– Transceiver where the frequency band and radio channel
bandwidth can be defined by software.
5
SINGAPORE
1. Mitola J., “The software radio architecture,” IEEE Communications Magazine, vol. 33, no. 5, pp.
26-38, May 1995.
2. Buracchini E., “The software radio concept,” IEEE Communications Magazine, vol. 38, no. 9, pp.
138-143, September 2000.
![Page 6: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/6.jpg)
Band
Limiting
Filter
ADC
Digital
Front End
(DFE) LNA
LO1 (Local
Oscillator)
Anti-
Aliasing
Filter
DSP
RF stage IF stage
LO2
π/2
Digital Down Conversion
(DDC) Sample Rate Conversion
(SRC) Channelization
Wideband
Input
Signal
-Fs/2 +Fs/2
Shifted
Signal
-Fs/2 +Fs/2 0Hz
Spectral
Shift
Filtered
Signal
-Fs/2 +Fs/2 0Hz
Decimated
Signal
-Fs/2N +Fs/2N
Introduction: SDR Receiver
6 3. Hentschel T., Henker M., Fettweis G., “The digital front-end of software radio terminals,” IEEE
Personal Communications, vol. 6, no. 4, pp. 40-46, August 1999. SINGAPORE
![Page 7: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/7.jpg)
Digital Front End
fs/2N1
Low Pass
Filter
Down
Sampling
0 fs/2 0
0 fs/2
Wideband
Input
fs/2N2
Digital
Down
Conversion
Band Pass
Filter
Down
Sampling
0 fs/2 0 0 fs/2
fs/2N3 0 fs/2 0
0
Digital
Down
Conversion
High Pass
Filter
Down
Sampling
fs/2
Channelizer
(Filter Bank)
7
SINGAPORE
3. Hentschel T., Henker M., Fettweis G., “The digital front-end of software radio terminals,” IEEE
Personal Communications, vol. 6, no. 4, pp. 40-46, August 1999.
![Page 8: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/8.jpg)
Introduction: Channelization
• Channelization is the process of extraction of single or
more than one channels (frequency bands) of interest
from the wideband input signal [4, 5].
8
SINGAPORE
4. Lee Pucker, “Channelization techniques for software defined radio,” in Proceedings of Spectrum
Signal Processing Inc., Burnaby, B.C, Canada.
5. Hentschel T., “Channelization for software defined base stations,” Annales des
Telecommunications, vol. 57, no. 5-6, pp. 386-420, March 2002.
Block diagram of channelization in SDR receivers
Input SignalChannel Filter/
Filter Bank
Extracted Channels at
Output
It is the most computationally intensive block in the DFE –
Operates at the highest sampling frequency.
![Page 9: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/9.jpg)
• SDR based cognitive radio (CR) targets the opportunistic
usage of the radio frequency spectrum.
• Proposed to solve ‘spectrum scarcity’ due to existing static
spectrum allocation scheme.
• CRs have the ability to sense and detect the current
spectrum utilization, and change their behavioral and
transmission characteristics dynamically so as to achieve
efficient spectrum access.
9
SINGAPORE
6. Mitola J., Maguire G. Q., “Cognitive radio: Making software radios more personal,” IEEE Personal
Commununications, vol. 6, no. 4, pp. 13–18, August 1999.
7. Mitola J., “Cognitive radio for flexible mobile multimedia communications,” IEEE International
Workshop on Mobile Multimedia Communications, MoMuC 1999, pp. 3-10, San Diego, USA, 15-
17 November 1999.
8. Haykin S., “Cognitive radio: Brain-empowered wireless communications,” IEEE Journal on
Selected Areas in Communications, vol. 23, no. 2, pp. 201- 220, February 2005.
Cognitive Radios - Opportunistic Spectrum Utilization
![Page 10: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/10.jpg)
free
free
time
ampl.
free
free
free
free
free
freq.fre
e
free
free
free
standard 1 standard 3standard 2
Spectrum Opportunities for cognitive radios – Time-
varying spectral scenario
Cognitive radios (secondary users/unlicensed users) utilize free
bands (vacant bands /white spaces/holes) for communication in an
opportunistic fashion when primary users (licensed users) are not
using those bands.
![Page 11: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/11.jpg)
Spectrum Sensing
• Spectrum sensing in CRs - the presence and (or) absence
of signals of licensed users (called primary users) is
detected in the wideband input frequency range in order to
allow opportunistic access of the vacant frequency bands
to the unlicensed users (called secondary users or
cognitive radio users) [9].
11
SINGAPORE
9. Ghasemi A., Sousa E.S., “Spectrum sensing in cognitive radio networks: requirements,
challenges and design trade-offs,” IEEE Communications Magazine, vol. 46, no. 4, pp. 32-39,
April 2008.
![Page 12: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/12.jpg)
Introduction: Channelization and
Spectrum Sensing in CRs
• Spectrum sensing is often done prior to channelization.
12
SINGAPORE
10. Yucek T., Arslan H., “A survey of spectrum sensing algorithms for cognitive radio applications,”
IEEE Communications Surveys & Tutorials, vol. 11, no. 1, pp. 116-130, First Quarter 2009.
11. Farhang-Boroujeny B., “Filter bank spectrum sensing for cognitive radios,” IEEE Transactions on
Signal Processing, vol. 56, no. 5, pp. 1801-1811, May 2008.
Input SignalSpectrum Sensing
Channel Filter/Filter Bank
Extracted Channels at
Output
Block diagram of channelization and spectrum sensing in cognitive radios
• Digital filters and filter banks have potential applications in spectrum
sensing and channelization. [10, 11].
![Page 13: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/13.jpg)
Spectrum Sensing and Channelization in Military Radio
time, t1
time, t2
fa1 fb1 fc1 fd1 fe1
fa1 fb1 fc1 fd1 fe1
In military communications, the wideband signal will have time-varying frequency bands scattered across the spectrum. The locations of channels and channel bandwidths will change dynamically.
The channelizer must be dynamically reconfigured to continue reception. Therefore, the center frequencies of the bandpass filter should be dynamically tuned to the new center frequencies with minimum computational overhead (low power) and high speed.
This problem is not adequately addressed in the literature. Also real-time and accurate estimation of spectrum is an open research issue.
![Page 14: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/14.jpg)
14
SINGAPORE
Channelization and Spectrum sensing in CRs
Introduction: Channelization and
Spectrum Sensing in CRs
![Page 15: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/15.jpg)
Filter banks for Spectrum Sensing &
Channelization
15
SINGAPORE
Filter Bank Complexity Group
Delay
Non-
uniform
bandwidth
subbands
Control over
bandwidth and
location of
each subband
Per Channel
approach Very high Low Yes Yes
DFTFB [12] Low Low No No
MPRFB [13] High Low Yes No
FFB [14] Very low Very high No No
• DFTFB: Discrete Fourier transform (DFT) based filter bank.
• MPRFB: Modulated perfect reconstruction filter bank.
• FFB: Fast filter bank.
12. Vaidyanathan P. P., “Multirate digital filters, filter banks, polyphase networks, and applications: a
tutorial,” Proceedings of the IEEE, vol. 78, no. 1, pp. 56-93, January 1990.
13. Abu-Al-Saud W.A., Stuber G. L., “Efficient wideband channelizer for software radio systems using
modulated PR filter banks,” IEEE Transactions on Signal Processing, vol. 52, no. 10, pp. 2807–
2820, October 2004.
14. Lim Y.C., Farhang-Boroujeny B., “Fast filter bank (FFB),” IEEE Transactions on Circuits and
Systems II: Analog and Digital Signal Processing, vol. 39, no. 5, pp. 316-318, May 1992.
![Page 16: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/16.jpg)
Challenges and Motivations
• Channelizer is the most computationally intensive part of the
digital front-end in SDR/CR receivers.
• Channelizers in SDR/CR receivers have stringent area, power
and cost specifications – Mobile handset constraints.
• In multi-standard scenarios, the conventional technique of
switching the operation among distinct receivers for different
standards is not an efficient approach.
• Reliable and fast spectrum sensing techniques with low
implementation complexities are desired in CR receivers.
16
SINGAPORE
![Page 17: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/17.jpg)
Our Selected Contributions
1. Coefficient Decimation Method (CDM) for
Realizing Very Low Complexity Variable
(Reconfigurable) Digital Filters
17
SINGAPORE
![Page 18: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/18.jpg)
Coefficient Decimation Method (CDM) Filter coefficients : h0 h1 h2 h3 h4 h5 h6 h7 h8 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 … hN
CDM-II by M=2 : h0 h2 h4 h6 h8 … hN
•In CDM-I, if denotes Fourier transform (FT) of the modal
(prototype) filter coefficients, then FT of the modified filter
coefficients is [16]
18
SINGAPORE
)( jeH
1
0
2
)(1
)('
M
k
M
kωj
j eHM
eH
16. R. Mahesh, A. P. Vinod, “Coefficient decimation approach for realizing reconfigurable finite
impulse response filters,” IEEE International Symposium on Circuits and Systems, ISCAS 2008,
pp. 81-84, Seattle, USA, 18-21 May 2008.
- Multi-band response with center frequencies at 2piK/M
![Page 19: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/19.jpg)
CDM: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
Frequency response:
19
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 20: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/20.jpg)
CDM: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 0 h10 … hN
Frequency response:
20
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 21: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/21.jpg)
CDM: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 0 h10 … hN
CDM-II by M=2 : h0 h2 h4 h6 h8 h10 … hN
Frequency response:
21
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitu
de (
dB
)
![Page 22: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/22.jpg)
CDM: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=3 : h0 0 0 h3 0 0 h6 0 0 h9 0 … hN
Frequency response:
22
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 23: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/23.jpg)
CDM: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=3 : h0 0 0 h3 0 0 h6 0 0 h9 0 … hN
CDM-II by M=3 : h0 h3 h6 h9 … hN
Frequency response:
23
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitu
de (
dB
)
We can obtain multi-
band (uniform
bandwidth) magnitude
responses as well as
variable bandwidth
frequency responses
from a single fixed-
coefficient LPF using
CDM I and II
respectively.
![Page 24: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/24.jpg)
Contributions
2. CDM based Filter Bank (CDFB)
24
SINGAPORE
![Page 25: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/25.jpg)
25
H(z)
H(z)
FMA(z)
FMA
HC(z) = 1-H(z)
Z-(N-1)/2 -
+
H(zM)
H(zM)
Z-(N-1)M/2
HC (zM)
FMC(z)
FMC
F(z)
F(z)
Frequency Response Masking (FRM) for
Designing Filters with sharp cut-off
SINGAPORE
15. Lim Y.C., “Frequency-response masking approach for the synthesis of sharp linear phase digital
filters,” IEEE Transactions on Circuits and Systems, vol. 33, no. 4, pp. 357- 364, April 1986.
![Page 26: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/26.jpg)
CDM based Filter Bank (CDFB)
26
SINGAPORE
CDFB design procedure
Modal filter
design
CDM operations
Desired subbands at output
Spectral subtraction
Complementary filter operation
Masking filters
17. R. Mahesh, A. P. Vinod, “Low complexity flexible filter banks for uniform and non-uniform
channelisation in software radios using coefficient decimation,” IET Circuits, Devices & Systems,
vol. 5, no. 3, pp. 232-242, May 2011.
![Page 27: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/27.jpg)
CDFB: Design Example
27
SINGAPORE
0.125 0.375 0.625 0.875 1Normalized Frequency (x π rad/sample)
Mag
nit
ud
e
SB1 SB2 SB3 SB4 SB5
Desired subband distribution from 0 to fsamp/2
![Page 28: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/28.jpg)
28
SINGAPORE
Complementary
Delays
Input
Modal Filter
CDM-I by M=2
CDM-I by M=4
Subband 2 (SB2)
Subband 4 (SB4)
Subband 5 (SB5)
+
Subband 1 (SB1)
Subband 3 (SB3)
+
+
+
-
+
-
+ -
CDFB block diagram
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Complementary filter response
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Masking Filter 1
Masking Filter 2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
![Page 29: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/29.jpg)
CDFB vs DFTFB (Xilinx XC40150XV)
29
SINGAPORE
Design DFTFB CDFB
Area (CLBs) Delay Area (CLBs) Delay
FIR filter bank 2352/5184 32.7 1755/5184 20.5
DFT 2350/5184 53.8 0/5184 0
Full Design 4321/5184 51.6 2200/5184 25.4
• CDFB offers area reduction of 49.09% and speed improvement of 50.78%
over DFTFB.
![Page 30: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/30.jpg)
Contributions
3. Decimation-Interpolation and
Masking (DIM) based Channel Filter: Exploiting two degrees of freedom for
enhancing filter programmability
30
SINGAPORE
![Page 31: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/31.jpg)
Mode select for
architectural
reconfigurability
Y1
Y2
FMA(z)
FC(z)
FA(z)
FMC(z)
Y
3
z-1
z-1
h0 h1 h1 h0
Y
z-1
z-1
z-1
z-1
z-1
z-1
Mode select
for varying L
2 2
1 1
1
Mode select
for varying D LUT
Architecture of modal filter.
The reconfigurable channel filter.
• Filter level reconfigurability: by changing I
and D of the modal filter, to obtain D*I
different subbands. (I=interpolation factor,
D=decimation factor for CDM-II)
• Architectural level reconfigurability: by
selecting outputs Y1 and Y2 using Mux-3.
• By exploiting both, we are able to obtain
[D.I + (I+2)] distinct subbands [18].
DIM based Channel Filter
31
SINGAPORE
18. Smitha K. G., A. P. Vinod, “A new low power reconfigurable decimation-interpolation and
masking based filter architecture for channel adaptation in cognitive radio handsets,”
Physical Communication, Elsevier, vol. 2, no. 1–2, pp. 47-57, March–June 2009.
![Page 32: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/32.jpg)
• Specifications of modal filter: passband edge
fp = 0.1 and stopband edge fs = 0.125.
• Following Nyquist criterion, we can note that
the maximum value of D is 4, for normalized
frequency range 0 to 0.5, where 0.5 = fsamp/2.
• I = 4 for illustration.
32
SINGAPORE
Architectural level reconfigurability
DIM based Channel Filter: Design
Example
I D fp fs
[fmp , fms] of
masking
filter
[fmcp , fmcs] of
complementary
masking filter
1 3 1 0.366 0.375 [0.366,0.5] [0.3,0.375]
2 3 2 0.4 0.417 [0.4, 0.5] [0.266, 0.418]
3 3 3 0.433 0.4583 [0.433,0.5] [0.233,0.458]
4 4 1 0.275 0.281 [0.275,0.468] [0.225,0.281]
5 4 2 0.3 0.3125 [0.3,0.438] [0.2,0.3125]
6 4 3 0.325 0.344 [0.325,0.4062] [0.175,0.344]
I D fp fs
[fmp , fms] of
masking filter Nmask
1 1 1 0.1 0.125 [0.1,0.5] 7
2 1 2 0.2 0.25 [0.2,0.5] 10
3 1 3 0.3 0.375 [0.3,0.5] 15
4 1 4 0.4 0.5 [0.4,0.5] 32
5 2 1 0.05 0.0625 [0.05, 0.4375] 8
6 2 2 0.1 0.125 [0.1,0.5] 7
7 2 3 0.15 0.1875 [0.15,0.3125] 19
8 2 4 0.2 0.25 [0.2,0.5] 10
9 3 1 0.033 0.0416 [0.033,0.2917] 12
10 3 2 0.066 0.0833 [0.066,0.25] 17
11 3 3 0.1 0.125 [0.1,0.5] 7
12 3 4 0.133 0.166 [0.133 0.166] 99
13 4 1 0.025 0.0313 [0.025,0.2188] 16
14 4 2 0.05 0.0625 [0.05, 0.4375 8
15 4 3 0.075 0.0938 [0.075,0.3125] 13
16 4 4 0.1 0.125 [0.1,0.5] 6
Filter level reconfigurability
![Page 33: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/33.jpg)
33
SINGAPORE
DIM based Channel Filter: Design
Example
Per Channel
approach
DIM based
channel filter
Gate count 533,451 476,636
Total power (mW) 421.3 396.5
Total delay (ns) 16.71 17.035
• The decimation-interpolation and masking based channel filter
offers gate count reduction of 10.65% and power reduction of
5.89% over the PC approach architecture. The real advantage is
in it’s high frequency response flexibility!
Implementation Results
![Page 34: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/34.jpg)
Contributions
4. CDM-FRM based Filter Bank
(CDM-FRM FB)
34
SINGAPORE
![Page 35: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/35.jpg)
• Stage 1 consists of prototype filter Ha and complementary filter Hc.
• The CDM-II decimation factor DII for prototype filter is varied from DIImin to DIImax.
• Interpolation factor M is fixed.
CDM-FRM FB: Stage 1
Frequency
Frequency
Frequency
DII = DIImin
DII = DIImax
DIImin < DII < DIImax
0 2 4 6 8
0 2 4 6 8
0 2 4 6 8
CDM-II
35
SINGAPORE
19. S. J. Darak, A. P. Vinod and E. M-K. Lai, “A low complexity reconfigurable non-uniform filter bank
for channelization in multi-standard wireless communication receivers,” Journal of Signal
Processing Systems, Springer, vol. 68, no. 1, pp. 95-111, July 2012.
![Page 36: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/36.jpg)
CDM-FRM FB: Stage 1
36
SINGAPORE
Frequency
Frequency
Frequency
DII = DIImin
DII = DIImax
DIImin < DII < DIImax
0 2 4 6 8
0 2 4 6 8
0 2 4 6 8
Frequency
Frequency
Frequency
DII = DIImin
DII = DIImax
DIImin < DII < DIImax
1 3 5 7
1 3 5 7
1 3 5 7
![Page 37: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/37.jpg)
• Stage 2 consists of two banks of fixed-coefficient masking filters, Bank 1 and Bank 2.
• Masking filters extract the desired subband by masking the other subbands.
• The number of masking filters are optimized using CDM-I.
CDM-FRM FB: Stage 2
37
SINGAPORE
CDM-FRM FB
![Page 38: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/38.jpg)
CDM-FRM FB: Stage 2
38
SINGAPORE
• Single fixed-coefficient masking filter
is used to extract bands 0, 4 and 8 as
shown in figure.
• Similarly , masking filters H3(z) and
H4(z) are used to extract subbands
from the complementary response.
0 2 4 6 8
0
0 2 4 6 8
0 8
0 2 4 6 8
0 4 8
Frequency
2 6
H1(z)
H1(z1/2
)
H1(z1/4
)
H2(z)
(a)
(b)
(c)
(d)
(e)
(f)
(h)
Stage 2 - Bank 1
Frequency responses of Stage 2 - Bank 1
H 1 ( z )
H 2 ( z )
( a )
( b )
( d )
( f )
( h )
![Page 39: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/39.jpg)
39
SINGAPORE
Per Channel
approach DFTFB CDM-FRM FB
Number of occupied
slices 14334 11865 10157
Power (mW) 1473 709 690
Post-PAR minimum
period (ns) 11.533 26.549 30.12
• Number of slices occupied by the CDM-FRM FB is 29.14% lower than that
of the PC approach and 14.4% lower than that of the DFTFB.
• Power consumption of the CDM-FRM FB is 53.16% lower than that of the
PC approach and 2.68% lower than that of the DFTFB.
CDM-FRM FB: Implementation Results
![Page 40: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/40.jpg)
Contributions
5. Improved Coefficient Decimation
Method (ICDM) for variable filters and
filter banks with enhanced flexibility
40
SINGAPORE
![Page 41: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/41.jpg)
Recap: Coefficient Decimation Method
(CDM)
Filter coefficients : h0 h1 h2 h3 h4 h5 h6 h7 h8 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 … hN
CDM-II by M=2 : h0 h2 h4 h6 h8 … hN
•In CDM-I, if denotes Fourier transform (FT) of the modal
(prototype) filter coefficients, then FT of the modified filter
coefficients is
41
SINGAPORE
)( jeH
1
0
2
)(1
)('
M
k
M
kωj
j eHM
eH
16. R. Mahesh, A. P. Vinod, “Coefficient decimation approach for realizing reconfigurable finite
impulse response filters,” IEEE International Symposium on Circuits and Systems, ISCAS 2008,
pp. 81-84, Seattle, USA, 18-21 May 2008.
![Page 42: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/42.jpg)
Modified Coefficient Decimation
Method (MCDM)
Filter coefficients : h0 h1 h2 h3 h4 h5 h6 h7 h8 … hN
MCDM-I by M=2 : h0 0 -h2 0 h4 0 -h6 0 h8 … hN
MCDM-II by M=2 : h0 -h2 h4 -h6 h8 … hN
•In MCDM-I, if denotes FT of the modal filter coefficients,
then FT of the modified filter coefficients is [23]
42
SINGAPORE
1
0
)12(
)(1
)('
M
k
M
kωj
j eHM
eH
)( jeH
23. Abhishek Ambede, Smitha K. G., A. P. Vinod, “A modified coefficient decimation method to realize
low complexity FIR filters with enhanced frequency response flexibility and passband resolution,”
2012 35th International Conference on Telecommunications and Signal Processing, TSP 2012,
pp. 658-661, Prague, Czech Republic, 3-4 July 2012.
![Page 43: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/43.jpg)
Improved Coefficient Decimation
Method (ICDM): CDM + MCDM
Filter coefficients : h0 h1 h2 h3 h4 h5 h6 h7 h8 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 … hN
CDM-II by M=2 : h0 h2 h4 h6 h8 … hN
MCDM-I by M=2 : h0 0 -h2 0 h4 0 -h6 0 h8 … hN
MCDM-II by M=2 : h0 -h2 h4 -h6 h8 … hN
43
SINGAPORE
24. Abhishek Ambede, Smitha K. G., A. P. Vinod, “An improved coefficient decimation based
reconfigurable low complexity FIR channel filter for cognitive radios,” 2012 12th International
Symposium on Communications and Information Technologies, ISCIT 2012, pp.22-27, Gold Coast,
Australia, 2-5 October 2012
![Page 44: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/44.jpg)
ICDM
• ICDM = CDM + MCDM
• ICDM-I = CDM-I + MCDM-I
– Different multi-band frequency responses with a center frequency
resolution of π/M can be obtained, where M = decimation factor.
• ICDM-II = CDM-II + MCDM-II
– Variable lowpass and highpass frequency responses can be
obtained.
44
SINGAPORE
![Page 45: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/45.jpg)
Contributions
6. ICDM-I based Channel Filter
45
SINGAPORE
![Page 46: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/46.jpg)
ICDM-I: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
Frequency response:
46
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 47: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/47.jpg)
ICDM-I: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 0 h10 … hN
Frequency response:
47
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 48: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/48.jpg)
ICDM-I: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 0 h10 … hN
MCDM-I by M=2 : h0 0 -h2 0 h4 0 -h6 0 h8 0 -h10 … hN
Frequency response:
48
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 49: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/49.jpg)
ICDM-I: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=3 : h0 0 0 h3 0 0 h6 0 0 h9 0 … hN
Frequency response:
49
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 50: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/50.jpg)
ICDM-I: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=3 : h0 0 0 h3 0 0 h6 0 0 h9 0 … hN
MCDM-I by M=3 : h0 0 0 -h3 0 0 h6 0 0 -h9 0 … hN
Frequency response:
50
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
![Page 51: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/51.jpg)
ICDM-I: Illustrative Example
51
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itud
e (
dB
)
CDM-I by
M=4
MCDM-I by
M=2
![Page 52: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/52.jpg)
ICDM-I vs CDM-I
CDM-I ICDM-I
Modal filter order required Higher Lower
No. of coefficient multiplications
(Implementation complexity) Higher Lower
Center frequency resolution of
subbands (flexibility) 2π/M π/M
Stopband attenuation performance
(for same order modal filter) Lower SA Greater SA
52
SINGAPORE
![Page 53: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/53.jpg)
Contributions
7. ICDM-I based FB
53
SINGAPORE
![Page 54: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/54.jpg)
ICDM-I based FB
54
SINGAPORE
ICDM-I based FB design procedure
Modal filter
design
ICDM-I operations
Uniform subbands at output
Spectral subtraction
Complementary filter operation
Masking filters
25. Abhishek Ambede, Smitha K. G., A. P. Vinod, “A New Low Complexity Uniform Filter Bank based
on the Improved Coefficient Decimation Method,” Radioengineering Journal, vol. 22, no. 1, pp. 34-
43, April 2013.
![Page 55: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/55.jpg)
ICDM-I based FB: Design Example
55
SINGAPORE
0.125 0.375 0.625 0.875 1Normalized Frequency (x π rad/sample)
Mag
nit
ud
e
SB1 SB2 SB3 SB4 SB5
Desired subband distribution from 0 to fsamp/2
![Page 56: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/56.jpg)
56
SINGAPORE ICDM-I based FB block diagram
Complementary
Delays
Input
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Modal Filter
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
CDM-I using M=2
MCDM-I using M=2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Subband 2 (SB2)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Subband 4 (SB4)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Subband 5 (SB5)
+
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Subband 1 (SB1)
Subband 3 (SB3)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Masking
Filter 1
+
+
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Complementary filter response
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
Subbands 2 and 4Masking
Filter 2
+
-
+
-
+ -
![Page 57: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/57.jpg)
ICDM-I based FB: Implementation
Results
DFTFB CDFB ICDM-I based FB
Number of occupied slices 20433 7032 5943
Power (mW) 499 268 206
Post-PAR minimum period
(ns) 4.344 16.172 15.617
57
SINGAPORE
• Number of slices occupied by the proposed ICDM-I based FB is 70.92%
lower than that of the DFTFB and 15.49% lower than that of the CDFB.
• Power consumption of the proposed ICDM-I based FB is 58.72% lower
than that of the DFTFB and 23.13% lower than that of the CDFB.
![Page 58: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/58.jpg)
Contributions
8. ICDM-II based Channel Filter and FB
58
SINGAPORE
![Page 59: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/59.jpg)
ICDM-II: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
Frequency response:
59
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itu
de
(d
B)
![Page 60: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/60.jpg)
ICDM-II: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
CDM-I by M=2 : h0 0 h2 0 h4 0 h6 0 h8 0 h10 … hN
CDM-II by M=2 : h0 h2 h4 h6 h8 h10 … hN
Frequency response:
60
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
![Page 61: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/61.jpg)
ICDM-II: Illustrative Example
Filter coefficients: h0 h1 h2 h3 h4 h5 h6 h7 h8 h9 h10 … hN
MCDM-I by M=4 : h0 0 0 0 -h4 0 0 0 h8 0 0 … hN
MCDM-II by M=4: h0 -h4 h8 … hN
Frequency response:
61
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itu
de
(d
B)
![Page 62: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/62.jpg)
ICDM-II: Spectral Subtraction
62
SINGAPORE
(a) Frequency response after CDM-II by M=2 (b) Frequency response of modal filter
Resultant
frequency
response
[(a) - (b)]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Magnitude (
dB
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itu
de
(d
B)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itu
de
(d
B)
![Page 63: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/63.jpg)
ICDM-II: Spectral Subtraction
63
SINGAPORE
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itu
de
(d
B)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itu
de
(d
B)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Normalized Frequency ( rad/sample)
Ma
gn
itu
de
(d
B)
(c) Frequency response after MCDM-II by M=4 (d) Frequency response after MCDM-II by M=1
Resultant
frequency
response
[(c) - (d)]
![Page 64: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/64.jpg)
ICDM-II based Channel Filter and FB
64
SINGAPORE
ICDM-II based channel filter/ FB design procedure
Modal filter
design
ICDM-II operations
Uniform and non-uniform
subbands at output
Spectral subtraction
26. Abhishek Ambede, Smitha K. G., A. P. Vinod, “A low complexity uniform and non-uniform digital
filter bank based on an improved coefficient decimation method for multi-standard communication
channelizers,” Circuits, Systems, and Signal Processing (CSSP), Springer, DOI: 10.1007/s00034-
012-9532-9, Published online in December 2012.
![Page 65: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/65.jpg)
0 1 2 3 4 5 6 7 8 9 10-10
0
10
20
30
40
50
60
70
80
90
100
Frequency (MHz)
Ma
gn
itud
e (
dB
)
MCDM-II output frequency responsesCDM-II output frequency responses
0 1 2 3 4 5 6 7 8 9 10-10
0
10
20
30
40
50
60
70
80
90
100
Frequency (MHz)
Ma
gn
itud
e (
dB
)
ICDM-II based Channel Filter: Design
Example
65
SINGAPORE
Modal
Filter
ICDM-II by
M=1 to M=5
![Page 66: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/66.jpg)
0 1 2 3 4 5 6 7 8 9 10-10
0
10
20
30
40
50
60
70
80
90
100
Frequency (MHz)
Ma
gn
itud
e (
dB
)
Bluetooth: MCDM-IIby M=1
Zigbee: CDM-II by M=4
ICDM-II based Channel Filter: Design
Example
66
SINGAPORE
Input signal during time interval t1 – t2 Input signal during time interval t2 – t3
ICDM-II
frequency
responses
required to
extract above
channels
0 1 2 3 4 5 6 7 8 9 10-10
0
10
20
30
40
50
60
70
80
90
100
Frequency (MHz)
Ma
gn
itu
de
(d
B)
4MHz bandwidth Zigbee channel
0 1 2 3 4 5 6 7 8 9 10-20
0
20
40
60
80
100
Frequency (MHz)
Ma
gn
itu
de
(d
B)
1MHz bandwidth Bluetooth channel
![Page 67: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/67.jpg)
ICDM-II based Channel Filter: Design
Example
CDM-II based
channel filter
ICDM-II based
channel filter
Largest decimation factor
required Mmax 10 5
Worst case TBW
increment
10 times that of the
modal filter’s TBW
5 times that of the
modal filter’s TBW
Worst case SA reduction 10 times that of modal
filter’s SA
5 times that of modal
filter’s SA
67
SINGAPORE
• The order of the modal filter in the proposed ICDM-II based channel
filter is 57.14% lower than that of the modal filter in the CDM-II based
channel filter.
24. Abhishek Ambede, Smitha K. G., A. P. Vinod, “An improved coefficient decimation based
reconfigurable low complexity FIR channel filter for cognitive radios,” 2012 12th International
Symposium on Communications and Information Technologies, ISCIT 2012, pp.22-27, Gold Coast,
Australia, 2-5 October 2012
![Page 68: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/68.jpg)
ICDM-II based FB: Design Example
68
SINGAPORE
Frequency responses obtained by performing ICDM-II using M=1 to M=6
0 1 2 3 4 5 6 7 8 9 10 11 12-80
-70
-60
-50
-40
-30
-20
-10
0
10
Frequency (MHz)
Ma
gn
itud
e (
dB
)
CD
M-I
I us
ing
M=
1
CD
M-I
I us
ing
M=
2
CD
M-I
I us
ing
M=
3
CD
M-I
I us
ing
M=
4
CD
M-I
I us
ing
M=
5
CD
M-I
I us
ing
M=
6
MC
DM
-II
usin
g M
=6
MC
DM
-II
usin
g M
=5
MC
DM
-II
usin
g M
=4
MC
DM
-II
usin
g M
=3
MC
DM
-II
usin
g M
=2
MC
DM
-II
usin
g M
=1
![Page 69: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/69.jpg)
ICDM-II based FB: Design Example
69
SINGAPORE
Input signal containing
WCDMA, Zigbee and
Bluetooth channels
ICDM-II frequency
responses required to
extract above channels
0 1 2 3 4 5 6 7 8 9 10 11 12
0
20
40
60
80
100
120
Frequency (MHz)
Ma
gn
itu
de
(d
B)
5 MHz bandwidth WCDMA channel
1 MHz bandwidth Bluetooth channels
4 MHz bandwidth Zigbee channel
WCDMA ZigbeeBluetooth
CH 1Bluetooth
CH 2
0 1 2 3 4 5 6 7 8 9 10 11 12-80
-60
-40
-20
0
20
40
Frequency (MHz)
Ma
gn
itu
de
(d
B)
Frequency response obtained after CDM-II with M=5 (for extracting WCDMA channel)
Frequency response obtained after MCDM-II with M=5 (for extracting Zigbee channel, Bluetooth CH 1)
Frequency response obtained after MCDM-II with M=6 (for extracting Bluetooth CH 1)
Frequency response obtained after MCDM-II with M=1 (for extracting Zigbee channel, Bluetooth CH 2)
![Page 70: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/70.jpg)
ICDM-II based FB: Design Example
70
SINGAPORE
27. Mengda Lin, A. P. Vinod, Chong Meng Samson See, “Progressive decimation filter banks for
variable resolution spectrum sensing in cognitive radios,” IEEE 17th International Conference on
Telecommunications, ICT 2010, pp. 857-863, Doha, Qatar, 4-7 April 2010.
PDFB
[27] ICDM-II based FB
Largest decimation factor
required Mmax 12 6
Worst case TBW
increment
12 times that of the
modal filter’s TBW
6 times that of the
modal filter’s TBW
Worst case SA reduction 12 times that of
modal filter’s SA
6 times that of modal
filter’s SA
• The order of the modal filter in the proposed ICDM-II based FB is
95.67% lower than that of the modal filter in the CDM-II based PDFB.
![Page 71: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/71.jpg)
Conclusions
• A coefficient decimation method (CDM) and an improved
coefficient decimation method (ICDM) to obtain different
lowpass, highpass and multiband frequency responses
using a single lowpass prototype filter have been
proposed.
• Low complexity channel filters and filter banks (FBs) based
on FRM, CDM and ICDM have been proposed for uniform
as well as non-uniform multi-standard channelization.
71
SINGAPORE
![Page 72: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/72.jpg)
Conclusions
• The proposed techniques are characterized by low
complexities and high flexibilities when compared with the
other methods in literature.
• Proposed filter banks also have potential applications in
spectrum sensing and other areas (example: biomedical
signal processing).
72
SINGAPORE
![Page 73: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/73.jpg)
Call for Papers – International Symposium on Electronic
System Design (ISED 2013), 12-13 December, Singapore (IEEE Computer Society Sponsored)
SYMPOSIUM TRACKS:
• Analog/Mixed-Signal System Design
• Digital System Design and Validation
• Embedded System Design
• Emerging Technology and System Design
• Power Aware System Design
• Software System and Application Design
• Wireless/Wired Communication Systems
Submission deadline: 15 July 2013
Website: http://ised.seedsnet.org/ 73
![Page 74: Low Complexity, Reconfigurable Digital Filters and Filter ...ljilja/cnl/guests/Filter_Banks_for_SDR_and_CR.pdf · • Software [defined radio (SDR) 1, 2] is a technology that has](https://reader034.fdocuments.us/reader034/viewer/2022051810/60189ef42895f020b946e2fa/html5/thumbnails/74.jpg)
Thank you
74
SINGAPORE