1. A SDN-Controlled Underwater MAC and Routing Testbed
Ruolin Fan∗, Li Wei† Pengyuan Du*, Ciarán Mc Goldrick♠
and Mario Gerla∗
* University of California Los Angeles, Los Angeles, CA, USA
† Michigan Technological University, Houghton, MI, USA
♠ Trinity College Dublin, Dublin, Ireland

2. Outline Introduction Background Design Testbed Implementation
Testbed Usage
Conclusion

3. Introduction Scientific and military operations Ocean floor mapping
Ancient shipwrecks exploration

4. Introduction
Search and rescue missions (downed airplanes)
MH-370

5. Supervised Search Man guides search from vessel in real time
Tether support fiber optic communications
Supplies power to USV (Underwater Support Vehicle) and in turns UUVs
Covert operations possible by using optics only for the swarm search

6. Autonomous search Non real time search Wave actioned generator
No man in the loop
Wave actioned generator
Guarantees resupply
Unlimited searches
Several km fronts can be explored with multiple USVs
Video inputs processed on USVs
Results reported to base via satellite

7. Outline Introduction Background Design Testbed Implementation
Testbed Usage
Conclusion

8. Choices for PHY Under Water
Conventional radio waves are absorbed too quickly: not feasible
Acoustic PHY
Long propagation distances (kilometers)
Very high latency (speed of sound)
Small transmission rates (kbps)
Complexity: high latency require different packet collision models
Optic PHY
Short propagation distances (tens of meters)
Lower latency (speed of light in water)
Fast transmission rates (mbps)
Complexity: requires line of sight for transmission

9. Acoustic vs. Optical Telemetry Method Range Data Rate Efficiency
Propagation Speed
Acoustic
Several km
1 kbps
100 bit/Joule
1500 m/s
Optical
100 meter
1 Mbps
30,000 bit/Joule
2.55 * 108 m/s
* Farr, N.; Bowen, A.; Ware, J.; Pontbriand, C.; Tivey, M.; , "An integrated, underwater optical /acoustic communications system," OCEANS 2010 IEEE - Sydney , vol., no., pp.1-6, May 2010

10. Optical or acoustic? It depends on many things:
Water quality
Turbulence
Covertness
Mobility
Optical needs alignment
Energy availability
May need to support multiple modes on the same UUV, switching from one mode to the next dynamically

11. Proposal: Underwater SDN
Software defined networking (SDN): separation of control plane with data plane
Allows for flexibility and simplicity
Centralized network controller defines network behavior of other nodes
Control plane: acoustics
Data plane: (mostly) optics, acoustics optional

12. Outline Introduction Design Testbed Implementation Testbed Usage
Conclusion

13. General SDN Framework
Introduction to SDN/OpenFlow

14. Architecture and Components
SDN components for Software-Defined Mobile Network
SDN controller: The central intelligence of an SDN-based Mobile Cloud
Communicates with UUVs using long-range acoustics
Directs UUV movements in addition to networking
SDN wireless node: The UUVs that explore the ocean
Sends data to the controller mainly using optics
Networks as directed by the controller

15. The Underwater SDN Architecture
Active UUV
Recharging UUV
Centralized Network
Controller
Docking Station
Sleeping UUVs

16. Architecture (cont) SDN wireless node internals
Local Agent contains recovery mechanisms so that system can still function when communications with SDN controller are lost or disrupted
Acoustic
Optic
Acoustic

17. The U/W SDN Control Channel
Requirements:
Both positioning and commands
Covert, encrypted, secure..
One to many – efficient broadcast
Virtual Nets ( different missions)
Network function virtualization
JANUS

18. Control Channel Standard - JANUS
The primary advantages:
Simplicity of design
Among the least complicated forms of acoustic communications yet devised.
Robust to noise
This signal should be detected when the signal to noise ratio (SNR) in a given band is at better than -2 dB.
Robust without tracking for “reasonable” amounts of relative speed (range rate).

19. Control Channel Standard - JANUS
Optimal approach for asynchronous, multi-access (multi-user) applications
Optimal for robustness in the presence of all types of interference, including intentional jamming. 
Potentially difficult for third parties
Undetectable by energy detectors 
A “constant envelope” waveform
Transmitters not concerned with amplitude crest factors
Allows for maximum power allocation to the transmission.
Depending on SNR

20. Outline Introduction Design Testbed Implementation Testbed Usage
Conclusion

21. Testbed Implementation
Parts of the acoustic version of the system implemented in our WaterCom testbed
Small water tank
Lined with foam to attenuate acoustic waves
Compartmentalized with foam to limit connectivity
6 OFDM acoustic modems
3 large models with long-range signals
3 educational models with short-range signals

22. Testbed Implementation
All modems connected to the WaterCom server
Doubles as the SDN controller
Accessible remotely via <apus.cs.ucla.edu>
Uses the underwater protocol stack SeaLinx
Allows for flexible loading of protocols at different layers

23. WaterCom Implementation

24. Testbed Network Topology

25. Outline Introduction Design Testbed Implementation Testbed Usage
Conclusion

26. Comparing UW MAC Protocols
Using our testbed, we compared 2 existing UW MAC Protocols
Slotted FAMA (S-FAMA)
UW-Aloha
Under-water multi-hop scenario using acoustic radios
5-minute test cases
Varying packet size
Varying packet sending rates

27. Experiment Network Topology

28. Results: Throughput

29. Results: Packet Delivery Ratio

30. Outline Introduction Design Testbed Implementation Testbed Usage
Conclusion

31. Conclusion Design of an under-water SDN architecture
Acoustic control plane
(Mostly) optical data plane
Implementation in the WaterCom testbed
Comparison of S-FAMA and UW Aloha
UW Aloha has higher throughput and packet delivery ratio