1. Grand Challenges in Achieving Dynamic Spectrum Access
2006 Technology and Policy for Advanced Spectrum
Exploiting
Grand Challenges in Achieving Dynamic Spectrum Access
Adapt,
Morph,
Proliferate
Wireless Network After Next
Preston Marshall
Defense Advanced Research Projects Agency
Advanced Technology Office
Radio-Isotope
Micro-power
Sources
One of DARPA’s top 5 programs!
Defense News, June 3, 2002
Wolfpack
Photonic
Avionics
Polarization
Modulation
Rotation
Connectionless
Networks
Rescue
Radio
Opinions expressed are those of the authors, and do not represent the position of DARPA, the Department of Defense, or the United States Government

2. Why Just Put Dynamic Spectrum in Static Network Concepts?
The Next Challenges
Dynamic Spectrum is on It’s Way to Accomplishment
Key to Adoption is the Widest Possible Range of Benefit
Better Spectrum Management is Important, ….
But May not be the Sole or Most Effective Argument for its Adoption
Need to Research Network Capability Enabled by Dynamic Spectrum
How Can We Exploit Truly Dynamic Networks?
Understand Physical Layer Bandwidth Sharing
Create (and Tear Down) Topologies Every Second
Why Just Put Dynamic Spectrum in
Static Network Concepts?

3. Some Research Topics Enabled by Dynamic Spectrum (Partial)
Adaptive Topology
Just in Time Connectivity
Don’t Do Topology Discovery – Do Topology Creation
Non-Linear Effects Mitigation
Look at Total Energy – Not Just One Frequency
Adaptive Spectral Bandwidth, Modes, ….
Optimization of Networks with No Fixed Bandwidth, But some Interactions due to Shared Spectrum Resource
…. Oh, and update them every second!
How Can Dynamic Spectrum and MAC Layer Operation Be Fused
Cooperative and non-Cooperative Channel Sharing
These Topics Have No Equivalent
Wired IP Metaphor

4. Overarching Thrusts of the DARPA Adaptive Networking Programs
Discussion Points
Overarching Thrusts of the DARPA Adaptive Networking Programs
Current Technical Progress in Adaptive Spectrum
Future Issues with Growth of non-Linear Effects (Front End Linearity)
Leveraging Adaptive Spectrum to:
Lower Device Cost
Address Dense RF Environments
Create Adaptive Network and Services

5. Infrastructureless Networking: How Do I Network Without?
XYZ.COM
Cell
Towers
Unlimited
Energy
Central
Servers
Frequency &
Network
Planning
Known
Topology
Cache
Servers
Index
Services
Enter Search
Fiber &
Wires

6. Technology Implications
Current
(Possible) Future
Extensive Pre-Planning
Self Monitoring and Adaptation
Fixed Network Structure
Self Forming
Asymmetric, Hub-Spoke (ex. Cellular, VOIP)
Peer to Peer
Separate Network and User Equipment
Security at the Perimeter
Services and Info at Core
User Equipment is the Network
Distributed Security
Info at Edge

7. How Can We Operate in a Wide Range of Environments, Missions, Densities, …
Device Adaptation is Key to Needed Technology
Self Organize to Operate Without Infrastructure
But Leverage Whatever Infrastructure is Available
Locate Unused Radio Spectrum
While Ensuring no Interference to Other Users
Continuously Measure Environment
Avoid Need for Detailed Pre-planning, Deconfliction, …
Morph Physical and Media Access Layers
Leverage Multi-path with MIMO
Select Waveforms, bandwidths, Access Control Mechanisms
Apply Multiple Network Topologies
Optimize Packets, Streams, Multicasts, …
Adapt Topology to Mission and Opportunities
Opportunistic (instead of Fixed) Topology
Continuously Adjust Topology to Meet Instantaneous Network Needs
And Do All this at New Levels of Affordability
Essential to Achieving Required Density
Adaptation Can Mitigate Engineering Performance Compromises and Use Less Costly Devices

8. Some Wireless Networking Research Programs at DARPA
neXt Generation Communications (XG)
Develop Technology to Sense and Situationally Adapt Spectrum Usage for a 10 Times Increase in Spectrum Access
Disruption Tolerant Networking
Extending the Network Framework to Include Storage and Delivery for Episodic Networks
(Leverage Whatever Connectivity is Available)
MnM Mobile MIMO
Extending Multiple Input/Multiple Output Technology to Mobile Platforms for a 9 times increase in Throughput
Wireless Network After Next
Develop the Technology for Dense, Low Node Cost (Under $500) Networks with High QOS
Connectionless Networking
Dynamically Adapting Network Time Synchronization, Routing, Energy Strategy to Achieve High Performance at 100th the Energy

9. Some Wireless Networking Research Programs at DARPA (Continued)
Polarization Rotation Modulation
Encode Information in Polarization to Achieve a 9x More Space to Space and Space to Ground Bandwidth
Radio-Isotope Micro Power Sources
Power “Manufactured In” for Life of Device
Ultra-wideband Networking (NETX)
Close in LPD Communications in Harsh Environments
Very High Accuracy Positioning
Spatial Optical (Air/Air and Air/Ground)
“Wireless Does Not Always Mean RF!”
Adapting for the Reliability of RF, and Speed of Fiber
Air to Air and Air to Ground Providing for Wireless Communications at Fiber Speeds (ORCLE)
Passive Optical Tags that provide 100’s of Kilobits/Sec

10. DARPA XG Program
All Spectrum May Be Assigned, But…
XG is Developing the Technology and System Concepts for DoD to Dynamically Access All Available Spectrum
…Most Spectrum Is Unused!
Goal: Demonstrate Factor of 10 Increase in Spectrum Access

11. DARPA XG Program Investments
Sensor Technology
Signal Processing Algorithms
Spectrum Measurements
Capability
and Affordability
Subnoise
Detection
Distributed Sensing
Algorithms
Increased
Awareness
Dynamics
Spectrum Awareness
Optimizing
Strategies
Tactics
Interference
Effects
Assessment
Adaptive
Network Operation
XG Prototype
& Demonstration
Assessments
Non-Interfering
Operation
Performance
Experiments
Spectrum Adaptive
Networking
Behavior
Engineering
Basis
Enforcement
Implementation
Policy Language
Methodology
Policy
Reasoning
Framework & Semantics
Policy
Description
IEEE 1900

12. Yes/No or Additional Constraints
XG Operation
RF Info
Acquisition
Sensing Loop
Message Flow
Transceiver
System
Strategy
Reasoner
Policy
Reasoner
RF Resource
Request
Develop
Options
Process
Request
Radio
Platform
Accredited
Policy
Determine
Opportunities
RF Transmit Plan
Select
Opportunities
Yes/No or Additional Constraints
Policy Engine

13. What We Expect
Simulation of Early Designs Using Recorded Spectrum Measurements Shows Significant Potential

14. XG Prototype Development and Demonstration Schedule
Phase
Phase
Task Name
Task Name
Phase 3b
Phase 3b
Phase 3c
Phase 3c 3a 3a
Demonstrations
Demonstrations
#3 adds:
#3 adds:
Criteria
Criteria
Dev’t
Dev’t
6 mobile nodes
6 mobile nodes
1. Two Node
1. Two Node 1 1
Network Functions
Network Functions
#5 adds:
#5 adds:
2. Five Node
2. Six Node 2 2
Low Power Detector
Low Power Detector
25 mobile nodes
25 mobile nodes
3. Technology
3. Technology 3 3
Enhanced Detector
Enhanced Detector
4. Initial System
4. Initial System 4 4
5. Upgraded System
5. Upgraded System 5 5
#1 demonstrates:
#1 demonstrates:
6. Final Configuration
6. Final Configuration 6 6
2 nodes
2 nodes
7. Final Urban
7. Final Urban
#2 adds:
#2 adds:
#4 adds:
#4 adds: 7 7
Demo Spectrum Policy
Demo Spectrum Policy
8. Final Test Range
8. Final Test Range
6 nodes
6 nodes
Policy Enforcer
Policy Enforcer 8 8
Performance Detector
Performance Detector
Group Behaviors
Group Behaviors
Ultimate Modem
Ultimate Modem
Rendezvous
Rendezvous
# 6, 7, & 8 Demonstrate
# 6, 7, & 8 Demonstrates
Frequency Selection
Frequency Selection
All Technologies
All Technologies
Conduct Series of Field Demos
Incrementally Add Performance
Midterm Demonstration (Demo 3)
6 Nodes with Core Capabilities
Provide Confidence to Stakeholders
Final Demonstration
25 XG Nodes with Advanced Capabilities
Show Capability for Transition
XG 008 v6
XG 008 v6

15. Demonstration Objectives to Show:
No Harm
Detects Potential Victims Before Causing Harmful Interference
Disseminates Sensor Data For Spectrum Awareness
Incorporates Automated Spectrum Rule Enforcement Algorithms
XG Works
Demonstrates Automated Rendezvous and Frequency Selection
Enables Node Transitions Between Networks
Conducts Operations With Multiple Cooperative and Non-Cooperative networks
Adds Value
Increases Spectrum Access and Communications Capacity
Implements No Setup Networking
Implements Automated Spectrum Rule Updating Algorithms

16. Equipment Installed in Vans
NC DoD Radio and Test Equipment
XG Radio and Test Equipment
XG Radios (mobile)
NC DoD Radio (fixed)
Legacy Radio Link
XG Link

17. XG Behaves!! With XG – Sense and Jump Without XG – Signals Collide
With XG -- Moved Before Interfering

18. Speeded Up Transit of XG Past a Set of Non-Cooperative Radios
Colors are Different Frequencies
Dots are XG Radios
Triangles are non-Cooperative Networks
XG’s Always try to be on Lowest Frequency Possible (Blue)

19. Simple Scenario, With and Without XG Military PSC-5 FM Radio Victim
Victim Radio Bit Error Rate Without XG
Bit Error Rate With XG
Bit Error Rate
Typical impulsive Environmental Noise
Scenario Time
Node Separation (mtrs)
XG Caused Noise
Bit Error Rate
Only Background Impulsive Noise
XG Node Interferes within 600 meters
Node Separation (mtrs)
Scenario Time

20. Not Essential to Establish New Regulatory Framework, either
XG Strategy
XG is not the universal solution to all problems
Some cases are likely to be too hard to deal with
One hard case does not does not invalidate XG usefulness
Testing to be focused on identifying and validating
low hanging fruit
candidate spectra that are unsuitable for sharing
Anticipate incremental adoption on a “Not to Interfere” basis
Military on military (10x greater packing of radios)
Military on shared (Technical framework for sharing)
Opportunistic (Widespread NIB operation)
Not Essential to Establish New
Regulatory Framework, either
Nationally or Internationally

21. Policy Compliant Behavior
Rejected Notion that General Solution to Adaptive Radio Could be “Just Code”
Almost 200 Sovereignties
Thousands of Bands
Emerging Technologies
Investing in Computer Science Structure for Policy Reasoning
Reasoning Technologies to Control Real-Time Process
Declarative Language Expression
Provable Policy Expression and Implementation
Partitioned Policies into Two Categories
Policy Enforcement (Permitted Operation)
Optimizing (Network and Above)

22. Future Cognitive Components Will Not Come From Just One Source
Spectrum
Policy
Country #1
Network
Service #1
Persistence
Policy
Spectrum
Policy
Country #2
Network
Service #2
Topology
Optimizing
Spectrum
Leasing
Policy
Multiple
Dynamically
Interacting
Cognitive
Processes
Enterprise 1
Policies
Network
Optimizing
Security
Mode
Enterprise 2
Policies
Security
Mode
Interoperable
Mode # 4
Interoperable
Mode # 2
Device Mfr.
Constraints
Interoperable
Mode # 3
Industry
Standard A
Industry
Standard B
Interoperable
Mode # 1
Industry
Standard C

23. XG Program – Transition
Anticipate Incremental Adoption on a Not to Interfere Basis (NIB)
Military on Military (10x Greater Packing of Radios)
Coordinated Sharing (Military with Coordinated Users)
Opportunistic (Widespread NIB Operation)
Incremental Rollout Enables Near-Term Deployment as Appliqué Into Existing Systems
Add Protocols and Adaptation Software to Digital Networking Radios
Add Spectrum Sensing Algorithms
Military on Military
Coordinated Sharing
Opportunistic Spectrum
Not Necessary to Establish New Regulatory
Framework, Either Nationally or Internationally

24. DARPA Tuner Study Conclusion
Looking at High Density RF Environments
Even With Ultra-High Quality Front Ends, Inter-Modulation Will Cause 20 dB (100 Times) Increase in Noise for Wideband Front Ends in Contemplated Density
Narrow Analog Filters Essential to Survival in Typical Military Dense RF Environments
Can Not Throw Linearity at the Problem
Energy Costs of High Linearity Unacceptable in Battery Devices
Responsive Strategy
Accelerate High Performance Tuneable Filter Technology
100’s of Individual Filter Settings/Octave
DARPA ASP Program
Do Not try to Survive on Any Given Frequency
Use Adaptive Spectrum to Resolve Remaining In-Filter Conflicts
INPUT SIGNAL
Spectrum Appears to Offer Lots of Opportunities
Post-RF Preprocessing
Non-Linear
Distortions
Results Shown for 10w, 10dB Gain, IIP3 = 50dBm (Ultra High Quality) LNA w/ Prime Pwr > 12 Watts
Noise Floor Increases 20 dB!
Even Dynamic Spectrum Can Not Help the Radio with 20 db of Intermod!!

25. Adaptation Is Key
Without Tunable Ultra-High Q Front End Filters (Unavailable), Reliable Wideband Radio Operation in Dense Spectrum Can Not be Assured
Legacy Narrow and Fixed Band Radios Could Rely on Fixed Filters
Can De-conflict Intended Frequencies, but not Non-Linear Products
Co-Site Like Situations More Common
Large Energy Costs for Linearization Unacceptable for Many Applications
Spectrum Adaptation (as in XG) Essential to Move Around Strong Signals Automatically
IIP3 > 20dBm Required to Avoid 50% of Spectrum Having 5 dB Intermod Noise
But,… Even 20 dBm of IIP3 requires at least 5 watts of Prime Power (Ideally)

26. DARPA Director’s Vision of Communications Technology Evolution
Key to Infrastructureless Comm
MEMS/NEMS
Self-Forming
Mobile
Small Unit Operations-Situational Awareness System Program– SUO SAS
Mobile Networked MIMO Program – MNM
Future Combat System – Comms Program - FCS-C
Packet Radios
Future Networks
Dynamic Spectrum
Global Mobile Program - GLOMO
Next Generation Comms Program - XG
Wireless Network after Next Program - WNAN
Topology Optimizer
Control-Based Mobile Ad-Hoc Networking Program - CBMANET
Cognitive Technology
Information Protection
Dynamic Quarantine of Computer Based Worm Attacks Program – DQW
Defense Against Cyber Attack in MANETs Program – DCMANET

27. What We Need to Do Next Develop the Technology to: Achieve:
Adapt Network in order to Operate Radios with 20 db Lower SFDR and Linearity at the Same Performance
Existing Programs Provide toolkit for the Physical Layer, but Have no Network to Exploit the Opportunities
Scale Understanding to Ultra-Large Mobile Networks
Extend Concept of Packet Networks to Directly Implement Tactical Broadcast /Streaming Service
Operate Multiple Network Technologies Simultaneously to Meet Each Mission QOS Need
Achieve:
100 Time Reduction in Network Radio Cost ($500 per 4 Channel Radio)
10 Time Reduction in Network Area Coverage Cost
1,000 Times Increase in Demonstrated and Objective Network Scale (100’s to 100,000’s)
Higher Goodput/Throughput Ratio to support Broadcast/Netted Voice/Video

28. Adaptive Radio Uses All Network Layers to Resolve Issues
Each Technology Can Throw “Tough” Situations to other More Suitable Technologies without Impact on User QOS
Topology
Planning
Re-plan
Topology
Spectrum
Planning
Re-plan
Across Network
Network-Wide
MIMO
Spectrum
Too Tight
Dynamic Spectrum Key to Adaptive Networking
No Good MIMO Paths
Dynamic
Spectrum
Need
More
Range
Relocate
Around Spur
Radio Device
Unavoidable
Strong
Signal
Move to New Preselector Band
Device
Spurs, …
Beam
Forming
Nulling
Strong
Neighbor
Signal
Link

29. The DARPA WNAN Network & Radio
Optimizing Layer
“Looks Through” Lower Layers to Make Globally Optimizing Decisions
Topology Layer
Makes the Network Topology Achievable by the Radios. Plans Network Around Spectrum, Power, Channel, …
Network Layer
Multiple, Unique Networks Optimized for Stream (Voice and Video), Broadcast (GBS-Like) and Packet Services
MAC Layer
Adaptive Spectrum, MIMO, and Beamforming Modes
PHY Layer
Commercial Component-Based
Mitigated H/W Weaknesses
Standard RF Slice Widely Replicated
Global and Local
Optimizing
CBMANET
(As
Applicable)
New WNAN
Technology
CBMANET
(As
Applicable)
New WNAN
Technology
Dynamic
Spectrum
(XG)
MIMO
(MnM)
COTS-Based
Mil Radios
MEMS Filters and
Sensors (MTO)
Existing Program
Technology
New Program
Technology

30. Hardware Platform
Single RF Processing Slice Replicated to form 1, 2 and 4 channel MIMO/XG/ Beamforming Capable Radios
Reverse of Standard ATO Approach
Build Early H/W and Incrementally Add Network Capability
Have Early Demonstrator of DARPA Philosophy and Technology
Approach:
Develop early Prototypes By Leveraging Available Commercial Chips (TV-Tuners and Others)
Use Cost Pressure to Force Innovation for Lower Cost/Higher Performance
Contribution from MTO New Analog Signal Processing (MEMs Filter Program) Essential
Frequency
900 MHz to 6 GHz
Power
36 dBw
SFDR
60 dB
IP3
What it is!
Peak
10 Mbps
GPS Access Interleaved by Connectionless Networking Digital Post Processing
Control -
Based
MANET
New
Technology
New
Technology
Dynamic
$ 500 per 4 Channel Node, Spectrally Adaptive, MIMO, Beamforming, Member of Four Simultaneous Subnetworks, Ultra Low Latency
MIMO
Spectrum
(MnM)
(XG)
COTS
Chip Set

31. Conclusion Dynamic Spectrum Technology is:
Maturing
Being Tested and Demonstrated Now!
Being Worked into the Regulatory Process
Fundamental Physics-based Constraints Limits Effectiveness of Frequency Assignment (Co-channel) Management in Dense RF Environments
May Offer Opportunity for Nearer Term Affordability and Performance Benefits
Need to Start Research into how to Best Exploit the Dynamics that Adaptive Spectrum Makes Possible
Not Clear that this is Evolutionary from Current Technology
Dynamic Spectrum Access is the Key Technology Needed to Enable a New Generation of Low Cost, Adaptive Wireless Network Devices

32. Thank You! Preston Marshall Photonic Avionics
Adapt,
Morph,
Proliferate
Wireless Network After Next
Preston Marshall
Defense Advanced Research Projects Agency
Advanced Technology Office
Radio-Isotope
Micro-power
Sources
One of DARPA’s top 5 programs!
Defense News, June 3, 2002
Wolfpack
Photonic
Avionics
Polarization
Modulation
Rotation
Connectionless
Networks
Rescue
Radio
Opinions expressed are those of the authors, and do not represent the position of DARPA, the Department of Defense, or the United States Government