1
Broad Agency Announcement
Cognitive radio Low-energy signal Analysis Sensor
ICs (CLASIC)
Microsystems Technology Office
DARPA-BAA-10-77
8/31/2010
2
Table of Contents
Part I: Overview Information ...........................................................................................3
Part II: Full Text of Announcement ...................................................................................4
Sec. 1: FUNDING OPPORTUNITY DESCRIPTION .........................................4
Sec. 2: AWARD INFORMATION .....................................................................11
Sec. 3: ELIGIBILITY INFORMATION .............................................................12
A. Eligible Applicants...............................................................................12
B. Cost Sharing / Matching ......................................................................13
C. Other Eligibility Criteria ......................................................................14
Sec. 4: APPLICATION AND SUBMISSION INFORMATION .......................14
A. Address to Request Application Package ............................................14
B. Content and Form of Application Submission .....................................14
1. Security and Proprietary Issues .......................................................14
2. Abstract and Proposal Information .................................................16
3. Proposal Abstract Format ................................................................18
4. Full Proposal Format .......................................................................19
5. Volume I: Technical and Management Proposal ...........................19
6. Volume II: Cost Proposal ...............................................................22
7. Submission Dates and Times ..........................................................25
8. Intergovernmental Review ..............................................................25
9. Funding Restrictions .......................................................................25
Sec. 5: APPLICATION REVIEW INFORMATION .........................................25
A. Evaluation Criteria ...............................................................................25
B. Review and Selection Process .............................................................27
Sec. 6: AWARD ADMINISTRATION INFORMATION .................................28
A. Award Notices .....................................................................................28
B. Administrative and National Policy Requirements ..............................28
1. Meeting and Travel Requirements ..................................................28
2. Human Use ......................................................................................28
3. Animal Use .....................................................................................29
4. Publication Approval ......................................................................30
5. Export Control .................................................................................31
6. Subcontracting ................................................................................32
7. Electronic and Information Technology .........................................33
8. Employment Eligibility Verification ..............................................33
C. Reporting .............................................................................................33
D. Electronic Systems ...............................................................................33
Sec. 7: AGENCY CONTACTS ..........................................................................34
Sec. 8: OTHER INFORMATION .......................................................................34
A. Intellectual Property .............................................................................34
B. Other Transactions..............................................................................37
Sec. 9: Attachment 1. ..........................................................................................39
3
Part I: Overview Information
• Federal Agency Name – Defense Advanced Research Projects Agency
(DARPA), Microsystems Technology Office (MTO)
• Funding Opportunity Title – Cognitive Radio Low-energy signal Analysis
Sensor ICs (CLASIC)
• Announcement Type – Initial Announcement
• Funding Opportunity Number – DARPA-BAA-10-77
• Catalog of Federal Domestic Assistance Numbers (CFDA) – 12.910
Research and Technology Development
• Dates
o Posting Date: August 31, 2010
o Proposal Abstract Due Date: 4:00:00 p.m. Eastern Time, October 1,
2010
o Proposal Due Date: 4:00:00 p.m. Eastern Time, December 10, 2010
• Concise description of the funding opportunity - The goal of CLASIC is to
enable monolithic, high performance, ultra high energy efficiency, signal
recognition integrated circuits (ICs) for next-generation military microsystems in
areas such as cognitive communications, radar and electronic warfare.
• Anticipated individual awards – Multiple awards are anticipated.
• Types of instruments that may be awarded - Procurement contract, grant,
cooperative agreement or other transaction.
• Any cost sharing requirements - None
• Agency contact -
o Dr. Sanjay Raman
The BAA Coordinator for this effort can be reached at, fax: (703) 248-
8070, electronic mail: DARPA-BAA-10-77@darpa.mil.
DARPA/MTO
ATTN: DARPA-BAA-10-77
3701 North Fairfax Drive
Arlington, VA 22203-1714
4
Part II: Full Text of Announcement
Sec. 1: FUNDING OPPORTUNITY DESCRIPTION
The Defense Advanced Research Projects Agency often selects its research efforts
through the Broad Agency Announcement (BAA) process. The BAA will appear first on
the FedBizOpps website, http://www.fedbizopps.gov/, and Grants.gov website at
http://www.grants.gov/. The following information is for those wishing to respond to the
BAA.
DARPA is soliciting innovative research and development (R&D) proposals in the area of
Cognitive radio Low-energy signal Analysis Sensor Integrated Circuits (CLASIC), a
thrust within the DARPA Adaptive RF Technology (ART) program. The goal of
CLASIC is to enable monolithic, high performance, ultra high energy efficiency, signal
recognition integrated circuits (ICs) for next-generation military microsystems in areas such
as cognitive communications, radar and electronic warfare. A cognitive system is aware of
its external environment and internal states, such as the electromagnetic/signal
environment in the case of cognitive RF systems, and can autonomously decide and
adjust its behavior to optimize quality of service or other operational objectives. Signal
parameters of interest include modulation schemes, signal constellations, multiple access
or hopping schemes, channel utilization and demodulated symbols. DARPA envisions
that this goal will be achieved by investigating and developing novel RF, analog and mixed-
signal integrated circuit architectures and design techniques.
The scaling of integrated circuit technologies has resulted in transistor unity current gain
cutoff frequencies (fT) surpassing 400 GHz. In combination with specialized processor
architectures, this has resulted in digital processor speeds approaching 100,000 MIPS and
processing efficiencies surpassing 2 MIPS/mW with silicon-based integrated circuits.
However, the waveform processing requirements of emerging military cognitive radio
systems have resulted in ADC, DSP and algorithm complexity levels pushing beyond what
is realizable in low-power hand-held form factors. Therefore, a major challenge in
communications integrated circuit design is achieving ultra-high levels of performance and
energy efficiency for signal (waveform) recognition processing (i.e., high performance
cognitive radio signal sensor on a chip).
Significant technical obstacles to be overcome in CLASIC include the development of
energy-efficient, analog and/or mixed-signal processing techniques for separating and
analyzing mixtures of complex signals. These techniques may require: blind source
separation using RF adaptive recursive and transversal filters; analog implementations of
fast Fourier and wavelet transforms; and efficient implementations of signal feature
extraction and classification algorithms (e.g., cyclostationary signal feature extractors and
classifiers, etc.) in analog/neuromorphic processing blocks.
Proposed research should investigate innovative approaches that enable revolutionary
advances in integrated circuit design, architectures or algorithms. Specifically excluded
is research that primarily results in evolutionary improvements to the existing state of
practice.
5
Background
Over the past several years, DARPA has initiated programs addressing a number of
important technical challenges in RF/mixed-signal microelectronics, including design of
silicon CMOS/silicon-germanium BiCMOS RF/microwave integrated circuits;
linearization; self-healing integrated circuits to combat variability, environment and
ageing; and reducing power and thermal dissipation. Meanwhile, in the Cognitive Radio
space, a number of recent DARPA programs such as Analog Spectral Processors (ASP),
Wolfpack, NeXt-Generation Communications (XG) and Chip-Scale Mechanical
Spectrum Analyzers (CSSA) have implemented energy detection-based sensing
functionalities for characterization of spectral occupancy. With CLASIC, DARPA seeks
to develop new technologies that can realize blind waveform sensing functions of
cognitive military radios with unprecedented low energy performance. Waveform
parameters to be sensed include modulation schemes, signal constellations, multiple
access or hopping schemes, and channel utilization. In addition, DARPA is also
interested in approaches that can provide demodulated symbol streams.
Traditionally, signal recognition functions have been implemented in digital electronics,
including FPGAs and DSPs. Figure 1 shows the architecture of a conventional signal
recognition system, where RF signals are down-converted, digitized and correlated
against a library of known signals in order to determine waveform parameters. In
commercial wireless communications systems, where a known set of established
communications standards are involved, this process can be done quite efficiently.
However, in military systems, excessive amounts of energy are typically expended in the
waveform correlation process due to the potentially large numbers of waveforms of
interest, as well as significant power for the A/D conversion process in the case of wide
analysis bandwidths of interest. Alternative classification approaches, such as those based
on cyclostationary statistic computation and analysis, likewise require large amounts of
energy because of the high computational complexity of the underlying algorithms.
Figure 1. A conventional “digitize, search, correlate” signal recognition approach
While the technological advances in digital electronics have enabled increasing processor
power efficiencies in MIPS/mW, the high computational requirements in MIPS for
6
advanced waveform recognition functions result in physically large, power hungry
processing systems. Figure 2 shows this trend as related to processing requirements of
several waveforms of interest and an estimated processing requirement for blind source
separation.
Figure 2. Processing requirements of emerging cognitive radio techniques outpace
increasing power efficiencies of General Purpose Processors (GPP), Field-
Programmable Gate Arrays (FPGAs), Digital Signal Processors (DSPs) and
Graphics Processing Units (GPUs).
Efficient signal processing techniques for separating and decoding mixtures of complex
signals are needed. DARPA envisions that, in combination with architectural innovations,
analog signal processing techniques may provide significant improvements in effective
MIPS/mW with respect to state-of-art DSP/FPGA techniques; however, other approaches
that can reach the energy consumption goals are of interest as well. Figure 3 shows the
expected improvement over the state-of-art as applicable to recognition of an example
narrowband BPSK signal with SNR of -8 dB in a 500 MHz bandwidth. For the SOA
signal recognizer and other possible digital processors, use of an 11.7-ENOB COTS
analog-to-digital converter was assumed.
100 mW
1 W
10 W
100 W 1 kW
0.00001
0.0001
0.001
0.01
0.1
1
10
0.1 1 10 100 1000 10000 100000 1000000
M
IP
S
/
m
W
MIPS
Single Core GPP
Multi Core GPP
FPGA
DSP
GPU
GSM SINCGARS WNW OFDM
Blind Source
Separation of
25 10 Mbps
Signals
7
Figure 3: CLASIC goal; recognition of an example narrowband BPSK signal with
SNR of -8 dB in a 500 MHz analysis bandwidth1.
Technical Areas
DARPA seeks innovative proposals for research and development (R&D) of technologies
that will lead to revolutionary decreases in energy consumption needed to separate and
analyze arbitrary mixtures of signals. Figure 4 shows a conceptual block diagram of a
CLASIC demonstration radio platform, including an integrated, highly energy efficient
signal analyzer. Symbol estimation, although of interest, is not a required functional
component of the processor. It is expected that proposers will need to address both
technical areas described below. Proposers should include in their proposal a radio front-
end architecture that will couple to their CLASIC processor or a teaming strategy that
will enable field demonstration of the processor in the final phase of the thrust.
Specifically excluded is research and development that primarily results in evolutionary
improvements to the existing state of practice, regardless of the chosen analyzer
architecture or approach.
1 The curves for probability of correct classification vs. energy are derived from results presented in
Spooner et al, Automatic Radio-Frequency Environment Analysis, , IEEE 2000
30
40
50
60
70
80
90
100
0.01 0.1 1 10 100 1000 10000
Pr
ob
ab
ili
ty
o
f C
or
re
ct
C
la
ss
ifi
ca
tio
n
[%
]
Energy [J]
400x More Energy Efficient
Goal = 95 %
SOA Signal
Recognizer
ER = 700 Joules
CLASIC
ER = 0.25 Joules
SOA COTS Digital
Processor
ER > 100 Joules
8
Figure 4: Conceptual block diagram of the CLASIC Processor platform.
The technical areas of interest are as follows:
Technical Area One: Low-Energy Blind Signal Separation and Parameter Extraction
The goal of this area is to demonstrate (in fabricated hardware) the ability to accurately
separate “mixtures” of signals (waveforms) present in the analysis bandwidth with low
energy performance to reduce overall power requirements for next-generation cognitive
radio applications. The signal sensor should have the capability to separate multiple,
emitters with arbitrary signal parameters such as carrier frequency, symbol rate,
modulation type, multiple access scheme, etc., including overlapping (co-channel)
signals. Signal bandwidths of interest are both narrowband (e.g., 12.5 kHz) to wideband
(up to 500 MHz). Further, the signal sensor should have the capability to accurately (with
high probability of correct classification and low probability of false classification)
extract parameters from multiple emitters with randomly chosen emitter parameters such
as carrier frequency, symbol rate, modulation type, multiple access scheme, etc.,
including overlapping (co-channel) signals. Table 1 describes example parameters of
interest; however, this list should not be construed as exhaustive and proposals will be
evaluated on their ability to extract as many different parameter combinations as possible.
Table 1. Example Signal Parameters of Interest
Signal Parameter Examples
Modulation Scheme
FM, AM, USB, LSB, CW,
ASK, FSK, PSK, CPM, QAM
and OFDM
Digital Modulation Constellation Size 2, 4, 8, 16, 32, 64, 128, 256
Spread Spectrum Technique Frequency hopping and direct sequence
Multiple Access Scheme CDMA, TDMA and FDMA
Signal Mixture
Measurements
M1
M2
Mk
S1
S2
Sm
P1 P2 Pn
Parameters Of
Separated Signals
B
lin
d
So
ur
ce
Se
pa
ra
tio
n
Sy
m
bo
l
Es
tim
at
io
n
Feature Extraction /
Signal Recognition
CLASIC
Signal Mixture
Symbols of
Selected
Separated
Signals LNA
LO
9
CLASIC technology should be able to identify signal parameters for waveforms that are
not known a priori; however, example military waveforms of interest include
SINCGARS, SRW, WNW, HAVE QUICK, LINK-11, EPLRS, DAMA SATCOM, and
DSCS, among others. Commercial waveforms of interest include GSM, UMTS, LTE,
WiMAX, 802.11 standards, FM & AM, TV, Bluetooth, etc. Limitations in the relative
signal strengths of signals in the mixture should be studied and evaluated. Methods that
can classify signals in the negative SNR regime are also of interest to DARPA.
Technical Area Two: Integration with RF Front-End
Proposers should include in their proposal a radio front-end architecture that will couple
to their CLASIC processor, or a teaming strategy that will enable field demonstration of
the processor in the final phase of the thrust. It should also be noted that approaches that
optimize the overall energy consumption by co-design of the RF front-end with the
CLASIC processor circuitry are also considered within the scope of this solicitation.
However, in cases where innovative co-design of the RF front-end and the signal
recognition processing is employed, the proposers should clearly and credibly explain
how efficiencies gained in the RF front-end are consistent with the performance goals of
CLASIC discussed below.
Another area that is not explicitly identified as a separate technical area, but is integral to
all the above technical areas of interest, is the test and measurement (T&M) techniques
required to evaluate performance of the developed integrated circuits. These will likely
include capabilities such as generation of numerous signals of different types, various
higher order modulations, etc. Therefore, proposers should provide detailed information
on how they plan to characterize the circuits developed under CLASIC.
CLASIC Program Structure
DARPA anticipates that in order to achieve the end goal of signal identification as a
sensing function for low-energy waveform-agile cognitive radios, CLASIC performers
will need to structure their efforts as described below. DARPA also anticipates that this
program will be conducted over approximately four years; however, the length of each
program phase should be proposed based on the approach and the level of effort needed.
The aggressiveness of the proposed schedule will be considered in the evaluation of the
proposals. Intermediate technical milestones that demonstrate progress over the proposed
duration should be included in the proposal. These milestones must define a credible
trajectory towards achieving the program goals described below.
Signal recognition processor design and simulation: Performers are expected to focus on
algorithm development and optimization, processor architecture definition and behavioral
simulation, design and simulation of constituent function blocks, and transistor-level
simulation of overall processor design including parasitics and variability. The
performers should clearly and credibly demonstrate that their design will be able to meet
the end-of-thrust goals (described in the next section). A Critical Design Review (CDR)
of performers' designs will be conducted by a government review team at the end of each
10