1. Harnessing Mobile Multiple Access Efficiency with Location Input Wan Du * and Mo Li School of Computer Engineering Nanyang Technological University, Singapore

2. Main access to WLAN 2 “Smart phones overtake client pcs in 2011,” 2012. www.canalys.com/newsroom/smart-phones-overtake-client-pcs-2011

3. Pervasive Location Information Outdoors – GPS (meters) Indoor Localization –Sound (centimeter) –WiFi (meter) –Camera (meter) 3

4. Location Based Applications Navigation Augmented reality Fine-grained location in supermarkets 4

5. Key Observation Improving the communication efficiency using location input –Hidden terminal and exposed terminal problems in mobile WLAN In two campus WLAN of CENTAUR, 40% links of exposed terminals and 10% links with 70% throughput reduction due to hidden terminals. 5 Location error of localization Indoor: <1m Outdoor: <13.7m << Communication range of WiFi Indoor: >50m Outdoor: >200m

6. outline Problem review and State-of-the-Art Design of CO-MAP Implementation and Evaluation Conclusions 6

7. Hidden Terminal Detect this relation Prevent concurrent transmissions 7 Collision!

8. State-of-the-Art Extra coordination channel –DC-MAC (TPDS 2012) New hardware or USRP implementation Conflict map based scheduling –RXIP (INFOCOM’ 12) Overhead of map learning Centralized control for downlinks 8

9. Exposed Terminal Detect this relation Enable concurrent transmissions Multiple exposed terminal problem 9 Collision!

10. State-of-the-Art Extra coordination channel –Attached-RTS (TPDS 2012) New hardware or USRP implementation Conflict map based scheduling –CMAP (NSDI’ 08) and CENTAUR (MobiCom’ 09) Overhead of map learning Multiple exposed terminal problems Centralized control for downlinks 10

11. Co-Occurrence MAP - Overview 11 Co-Occurrence MAP log normal shadowing propagation model Exposed TerminalsHidden Terminals Minimize collisionMaximize spatial reuse Dynamic packet size Enchanced CSMA Fast Uniform Distributed

12. Exposed Terminal 12

13. Exposed Terminal 13

14. Concurrent Transmissions 14

15. Concurrent Transmissions 15

16. Concurrent Transmissions 16 Multiple Exposed Terminals  Enhanced CSMA

17. Concurrent Transmissions 17

18. Concurrent Transmissions 18 ACK Lost Problem  Windowed ACK

19. Hidden Terminal 19

20. Hidden Terminal 20

21. Hidden Terminal 21 Important Parameters: Number of HTs Packet Size

22. Dynamic Packet Length for Hidden Terminals 22 Probability of node i transmiting in slot s Packet size Number of hidden terminal Number of contending nodes

23. Implementation Testbed of six laptops –Intel Wireless 4965AGN network adapter –MAC80211 and iwlegacy wireless drivers. Three Components –CO-MAP –Header and concurrent ET transmission –Packet length adaptation Data rate adaptation – Minstrel (Default) 23

24. Implementation Header in data packets –Thirteen bytes (address and CRC) in PHY header 24

25. Evaluation – Exposed Terminal 25 78%

26. Evaluation – Hidden Terminal 26 39%

27. Large Scale Network on NS-2 27 Network layout –Three APs separated about 60m –Nine clients. –Thirty topological configurations 48% exposed links and 19% hidden terminals

28. Large Scale Network on NS-2 28 39%

29. Large Scale Network on NS-2 29 39% 19%

30. Tolerance to Position Inaccuracy 30 Location Error Range Misclassification percentage Wrong ETMissing ETWrong HTMissing HT 1m0.2%0.3%0.2% 5m1.2%1.4%1.1%0.8% 10m2.1%2.3%2.4%1.4%

31. Conclusion A practical work leveraging pervasive location information to improve spatial reuse and reduce hidden collisions in mobile WLAN Distributed design with rapid construction of conflict map Successful practice using sensor hints in protocol design 31

32. Thanks. Questions? Wan DU, duwan@ntu.edu.sg Research Fellow @ NTU, Singapore