1. Extending Cognitive Radios with New Perspectives Delay Tolerant Networks and Social Network Analysis Suzan Bayhan* and Jussi Kangasharju Department of Computer Science, University of Helsinki, Finland *bayhan@hiit.fi ICUFN 2014, July 8- July 11, Shanghai, China

2. 2/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China CR Vision in 1999 and in 2014  In 1999, PhD thesis by Joseph Mitola III, KTH  A cognitive radio: understands self, understands the user’s goals, understands networks, and understands radio. In 2014, Infrastructure-based and centralized architectures limiting us Mobility largely overlooked Context is missing (actors and their interactions)

3. 3/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Outline  An Overview of New Perspectives  Delay Tolerant Networks  An example scenario  Social Network Analysis  An example scenario  Conclusions

4. 4/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China An Overview of New Perspectives Social Network Analysis (SNA) Social Awareness Robustness to challenging environments Improved spectrum capacity Grasp the information in the network Social-aware operation Delay Tolerant Networks (DTNs)

5. 5/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Delay Tolerant Networks (DTNs) store-carry-forward No dependency on infrastructure Like ad hoc networks, but no tight requirement on end- to-end paths/delay Opportunistic, aka proximity- based communications Traffic offloading Censorship/monitoring-free! No one is in communication range of the other, how to communicate? Exploit MOBILITY of nodes

6. 6/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Opportunities of Opportunistic Communications for CRNs DTN opportunism: if nodes happen to be at the transmission range of each other at a time (time and space) DTN intermittently connected due to low node density, node mobility, etc. Mobility as enabler Robust to connection breakdowns CRN opportunism: if frequency is idle (time, space, frequency dimensions) CRN intermittently connected due to disruptions from PUs Mobility as a curse Robust to lack of assigned spectrum Can CR learn from DTNs how to tackle the intermittent connections? Retain key strengths of each paradigm DTN also known as opportunistic communications

7. 7/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China An Example Scenario: Contact Capacity w/wo CR-opportunism Contact capacity: Amount of data that can be transmitted at a contact Contact duration Wireless channel capacity (bandwidth)

8. 8/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Three Scenarios Scenario I: DTN: DTN contact capacity (only ISM band) Spectrum Scenario III: RAND: Random selection contact capacity (ISM + 1 PU band) Coordinate via ISM and tune to one PU channel Sense it Transmit via idle PU channel + ISM Scenario II: REM: REM-enabled contact capacity (ISM + m PU bands) Retrieve the list of best channels from REM Sense m channels Aggregate idle PU channels + ISM Radio Environment Map (REM) Channel availability at each location

9. 9/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China A Closer Look Contact starts Contact ends Contact starts Contact ends REM information, spectrum sensing REM information, spectrum sensing Coordinate to switch to the same channel o Contact duration T, ISM Channel capacity: B bps o N PU channels (f i, p i, mean availability p), capacity: βB o Spectrum sensing overhead: αT Contact starts t=0 Contact ends t=T Transmission time Channel capacity DTN REM (m sensing) RAND transmission Sensing and transmission

10. 10/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Contact Capacity α: sensing overhead δ: coordination overhead β : PU channel capacity/ISM capacity p: Mean PU channel availability p j : availability of channel f j m: # of channels sensed

11. 11/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Contact Capacity Gain Effect of Sensing Overhead If sensing is not cumbersome, contact capacity increases with CR-type opportunism. Capacity: REM > RAND > DTN RAND, p=0.5 RAND, p=0.8 m=1, N=4, δ=0.01 Contact capacity gain = C (*) /C dtn

12. 12/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Effect of Sequential Sensing (m) Maximum m is bounded by min(T/α, N) Highest gain m* based on p and α 30% 170% 200% 300%

13. 13/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Summary J. Huang et al. Mobility-Assisted Routing in Intermittently Connected Mobile CRNs, IEEE TPDS, 2014. Jing Zhao, G. Cao, Spectrum-Aware Data Replication in Intermittently Connected CRNs, INFOCOM 2014.

14. Social Network Analysis and Social-awareness “Understand the self, Understand the network” 14/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Social network analysis (SNA) or network science tools: extract information hidden in the system either about individual nodes or the network as a whole.

15. 15/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Two Layered View of the Network (WCL) (SCL)

16. 16/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China An Example Scenario: Social-aware Cooperative Sensing Assume a mesh CRN in WCL An Erdös-Renyi CRN in SCL A CR i assesses its neighbors according to: Probability of sensing: is altruism realistic? (SCL) Social distance: length of shortest path bw. two nodes (SCL) SCL WCL Probability of sensing accurately (WCL)

17. 17/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China An Example Scenario: Social-aware Cooperative Sensing (2) F-sense: only friendship is considered (α f = 1) S-sense: only sensing accuracy (α s = 1) FSW-sense: All considered equally(α f = α s = α w = 1/3) A CR asked for cooperation senses probabilistically depending on the social distance (w j,i, n i,j -1 ) Who are the best cooperators? Should I sense for this CR? Social distance Probability of sensing accurately Probability of Sensing (willingness)

18. 18/21 ICUFN 2014, Session 8A Cognitive Radio. July 10, 2014, Shanghai, China Performance Evaluation Rejection ratio and social-distance between the cooperators can be decreased by using the social connectivity information. p j,I : U(0.7,1) Lowest reject ratio Lower waste of energy Lowest social distance (# hops bw. two nodes) More trusted cooperation Similar sensing performance

19. 19/21 ICUFN 2014, Session 9A Cognitive Radio. July 10, 2014, Shanghai, China Challenges Prediction of spectrum availability for the DTN relays and where contacts occur Privacy leakage vs. social- awareness Overhead of SNA (where to run) Overhead of exchange

20. 20/21 ICUFN 2014, Session 9A Cognitive Radio. July 10, 2014, Shanghai, China Conclusions Cognitive radios can be empowered using the well-established knowledge in DTNs and SNA. DTNs can improve by applying CR-type opportunism. Social-connectivity layer hosts variety of information, so exploit it! Occupy the spectrum and let the CRs be move and be social! Thank you. Suzan Bayhan bayhan@hiit.fi http://www.hiit.fi/u/bayhan

21. 21/21 ICUFN 2014, Session 9A Cognitive Radio. July 10, 2014, Shanghai, China References C Güven, S Bayhan, F Alagöz, Effect of Social Relations on Cooperative Sensing in Cognitive Radio Networks, BlackSeaCom, 2013. J. Huang et al. Mobility-Assisted Routing in Intermittently Connected Mobile Cognitive Radio Networks, IEEE Transactions on Parallel and Distributed Systems, 2013. Zhu, Ying, et al. "A survey of social-based routing in delay tolerant networks: positive and negative social effects." IEEE Communications Surveys & Tutorials, 15.1 (2013): 387-401. Kas, Miray, et al. "What if wireless routers were social? approaching wireless mesh networks from a social networks perspective." IEEE Wireless Communications, 19.6 (2012): 36-43.