1. The Radio Environment Map Sami Lunnamo

2. Presentation outline REM Definitions Requirements Challenges Design Location and mobility Database WS availability Layered REM

3. REM - Definitions Geolocation service And more Centralized database Radio environment data Predictions of spectrum opportunities Cognition cycle

4. REM – Definitions - Analogy REM - Location (x, y, z) -Geographical location -Radio spectrum profile Tourist map of Beijing, source: http://images.chinatravel.com/city/beijing/beijing-city-map-large.jpg

5. REM – Requirements Normal CR requirements Safety Efficiency Centralized service Availability Reliability Throughput Latency Security

6. REM – Challenges – Signal overhead Communication between REM and CR Amount of relevant data Quality of connection Balance between overhead and benefits of REM REM Clipart from bestclipartblog.com

7. REM – Challenges - Validity REM is collection of data How long will any data be valid? Information exchange between layers of REM

8. REM – Challenges – Spectrum opportunities Ability to detect and predict spectrum opportunity Model-driven scheme to calculate opportunities Loss-ratio can be close to 0% Hey, I have data I’d like to send REM Nope, no free bands here Bands

9. REM – Challenges - Locations Determining a location is error-prone and expensive Could CRs use base station location as substitute for knowing own location accurately? Mobile CRs One-off events and wireless microphones

10. REM – Challenges - Bootstrapping New client Connecting with REM through base station Chicken-and-egg

11. REM – Design – Spectrum opportunities Model-driven schemes vs. Data-driven schemes Longley-Rice (L-R) with terrain data Climactic effects, soil conductivity, permettivity, Earth’s curvature and surface refractivity 8% spectrum opportunity loss and even less false positives In practice, no one model is sufficient alone

12. REM – Design - Location Different channels open in different areas If using base station location, REM would have to use channels that are open and available in every part of the coverage area 80% spectrum opportunity loss Location granularity 4km = 80% loss 800m = ~0% additional loss Bootstrapping beacon Data of available channels

13. REM – Design - Mobility Mobile CRs location is constantly changing CR might travel to an area where it uses a channel in use by PU Old data about spectrum availability Protection range for channels Polling frequency for spectrum updates 96 km/h, 60sec, 1.6km = 20% loss 96 km/h, 30sec, 800m = 0% loss

14. REM – Design – Database What kind of data should the database contain? TV transmitter data Tower locations, antenna heights, transmit powers… Client data Locations, IDs, channels, transmit powers… Cache

15. REM – Design – Layered REM Distributing and decentralizing Layers REM SA and REM Managers Subsidiarity Data stored and analysed in least centralized node possible National – regional – local Proportionality Data with shorter life span needs to be readily available REM SA and manager within quick reach

16. REM - Summary Geolocation service Analysis and historical data of spectrum opportunities Can detect and use ~92% of all spectrum opportunities in area No false positives Accommodates mobile CRs Mobility doesn’t cause breach of safety requirement Layered architecture gives robustness and eliminates overhead Uses broadcast beacons to bootstrap new clients