Opportunistic Message Broadcasting in Campus Environments Salih Safa Bacanlı, Gürkan Solmaz, Damla Turgut Networking and Mobile Computing Lab Department of Computer Science University of Central Florida, Orlando, FL December 9, 2015 Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 1 / 1
Introduction Opportunistic Communication University Campuses as application scenario People are on foot Event - fire alarms, weather alerts (e.g., hurricane alert), closed roads State based Campus Routing Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 2 / 1
Opportunistic Network We collected mobility traces in UCF P 5 =61 min P 2 =54 min P 4 =12 min P 1 =225 min P 3 =248 min People spend time in buildings or they walk Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 3 / 1
Application Scenario Messages are created based on events Smartphones as nodes Message creators may be anyone Messages are broadcasted to everyone Messages are carried by store and forward manner Communication via Bluetooth Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 4 / 1
State based Campus Routing Protocol No packet exchange besides actual messages Session similar to Epidemic (Vahdat and Becker, 2005) Figure: Session example in epidemic routing A. Vahdat and D. Becker, "Epidemic routing for Partially-Connected ad hoc networks," CS-200006, Duke University, Tech. Rep., Apr. 2000. [Online]. Available: http://issg.cs.duke.edu/epidemic/epidemic.pdf Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 5 / 1
state=active state=idle S 1 S 3 state=active S 1 S 1 S 4 S 1 S 2 S 4 Hotspot P 1 12:30 pm 1:30 pm Figure: A mobile node s encounters Active state -encounter frequency increased Idle state -encounter frequency decreased Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 6 / 1
State Decision Function Probability to forward if node is idle Probability to forward if node is active Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 7 / 1
Simulation study SCR is compared with Epidemic PROPHET Pinit = 0.75 γ = 0.98 β = 0.25 Epidemic with Times-To-Send Random routing Message TTL value is 48 hours Parameter Value α 0.25 P wanted 0.99 λ 0.99 Table: SCR Parameters Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 8 / 1
Datasets University of Milano campus human walktrace 1 10 meter communication distance 44 people ( faculty members, doctoral students, and technical staff) 19 days University of Cambridge campus human walktrace 2 Bluetooth 12 people (doctoral students) 6 days 1 P. Meroni, S. Gaito, E. Pagani, and G. P. Rossi, "CRAWDAD data set unimi/pmtr (v. 2008-12-01)," Downloaded from http://crawdad.org/unimi/pmtr/, Dec 2008. 2 J. Scott, R. Gass, J. Crowcroft, P. Hui, C. Diot, and A. Chaintreau, "CRAWDAD data set cambridge/haggle (v. 2006-01-31)," Downloaded from http://crawdad.org/cambridge/haggle/, Jan 2006. Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 9 / 1
Success Rate 1 0.9 Cumulative Distribution Function 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 SCR Epidemic PROPHET Epidemic w/ TTS Random 20 40 60 80 100 Message delivery success (%) Figure: Success Rate Results for University of Milano Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 10 / 1
Cumulative Distribution Function 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 SCR Epidemic PROPHET Epidemic w/ TTS Random 20 40 60 80 100 Message delivery success (%) Figure: Success Rate Results for University of Cambridge Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 11 / 1
Message Delay 1 0.9 Cumulative Distribution Function 0.8 0.7 0.6 0.5 0.4 0.3 SCR 0.2 Epidemic PROPHET 0.1 Epidemic w/ TTS Random 0 2 4 6 8 10 12 14 16 Message delay (s) x 10 4 Figure: Message Delay Results for University of Milano Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 12 / 1
1 0.9 Cumulative Distribution Function 0.8 0.7 0.6 0.5 0.4 SCR 0.3 Epidemic PROPHET 0.2 Epidemic w/ TTS Random 0.1 2 4 6 8 10 12 14 16 Message delay (s) x 10 4 Figure: Message Delay Results for University of Cambridge Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 13 / 1
Number of Packets Sent 3 x 104 2.5 Number of Packets Sent 2 1.5 1 0.5 0 SCR Epidemic PROPHET Epidemic w/ TTS Random Routing Methods Figure: Number of Packets Sent in University of Milano Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 14 / 1
5000 4500 4000 Number of Packets Sent 3500 3000 2500 2000 1500 1000 500 0 SCR Epidemic PROPHET Epidemic w/ TTS Random Routing Methods Figure: Number of Packets Sent in University of Cambridge Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 15 / 1
Conclusions We proposed an application scenario of opportunistic communication in university campuses We designed a routing algorithm (SCR). We compared its performance with epidemic, PROPHET, epidemic with TTS, and random. SCR gives nearly the same performance results with Epidemic and PROPHET whereas decreasing the number of packets sent by 20 to 30%. Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 16 / 1
Thank you Email: bacanli@knights.ucf.edu Github: github.com/cosai/eons Bacanlı, Solmaz and Turgut (UCF) IEEE GLOBECOM 15 December 9, 2015 17 / 1