1. What is the optimal future architecture for spectrum monitoring?
Focus Group 1A
What is the optimal future architecture for spectrum monitoring?

2. Wanted a System That Scaled
Could handle 500K devices/sq km in urban areas
Could also work in rural areas

3. System of Systems/Hierarchy of Trust
High Quality, Trusted Measurement Platforms
Expensive, so sparsely deployed (1 per sq/km or less)
Includes special purpose equipment that may even more limited deployment
Medium Quality, Less Trusted Measurement Platforms
Carrier equipment on cell towers, “volunteer” spectrum sensing from universities or others
Lowest Trust, Measurement Platforms
Cell phones, etc.
Integrating information across these levels can yield pervasive and verified sensing capabilities
Unanimity that something like this is the way to go forward

4. Looking More Deeply into the System Architecture
System will likely need to support at least two modes of use
Long running sensing activities
Tracking noise temperatures over time
Focused activities
Locate the pizza delivery car using GPS blocker
Not just restricted to passive sensing – active sensing via pinging also possible
So system needs mechanisms for collection and reactive tasking
DARPA RadioMap has shown one way to the do the reactive tasking for passive sensing in a platform independent way (can also do collection)

5. How Is Tasking Done? A range of choices
Local devices work together to get job done
Proximate controller in local cloudlet directs sensors in its area
More centralized controller directs sensors over wide area
Example of airport suffering GPS jamming
Too many local devices for local coordination (passengers and their mobile devices)
Coordination from a national center seems to remote
Instinct suggests local cloudlet is just right
Where data is stored and how it is shared (is an open question)
Process most data locally to reduce backhaul capacity requirements and costs
Rely on a set of criteria to trigger a push of specific data to a centralized place (e.g. passing the change of noise over time)
Standardize the data interfaces of various sensors

6. Testing Driving with Applications
Measuring noise temperature
To get really good measurements need specialized equipment with limited sensing range due to front end requirements
Specialized, trusted devices in selected areas around the country, whose data provides baseline/ground truth that can also leverage mobile self reported data
Finding the GPS blocker in the pizza delivery vehicle
Trusted device reports a GPS
Can crowd source other devices to confirm that problem is not a failure of trusted device – and to find areas suffering the GPS failure
Note: loss of accurate clock an issue – may need a GPS applet that one sends to devices when trying to find jammer which is robust to loss of accurate clock…
A suggestion that the system of systems architecture should work
Also highlights some of the challenges (e.g. define what problems to solve)

7. Final Comments This is not a solution Cost…
It is a sturdy architecture in which many aspects remain research questions
But gives confidence that this kind of infrastructure is feasible
Cost…
Varies according to density of deployment of trusted nodes and willingness of others to permit use of their devices
$0.5Billion estimate for 1 trusted device per urban sq km (US telecom revenues in 2014 were $167B – so < 0.1% of annual revenue for such a system – easily repaid by better spectrum use)