1. INTERFERENCE MEASUREMENTS AND SPATIAL REUSE MAP GENERATION FOR WIRELESS MESH NETWORKS Pradeep Gopaluni Advisor: Dr. Bhaskaran Raman IIT Kanpur IIT Bombay.

2. Outline  Introduction to WMNs + Motivation  Problem Statement  Existing Work  RSSI Based Interference measurements  Signal Strength vs. Interference  RSSI – Analysis and Implications  Measurements - Setup  Future Work

3. Introduction  Wireless Mesh Networks  Self organized network of Mesh routers & Mesh clients  Low cost internet access network.  Flexible => Often unplanned  Scalable => May vary form small indoor settings to large and long-distance community networks.  Interference is one of the key factors that influence(degrade) the performance of Wireless Mesh Networks Source: http://research.microsoft.com/mesh/

4. Motivation  Interference graph  Used for Channel Assignment, Routing and Scheduling links in a WMN.  An automated mechanism of getting interference graph help us establish scalable and flexible WMNs  Periodic re-generation of Interference graph  Allows dynamic changes in network  Takes care of inherent variability

5. Problem Statement  To device an automated mechanism to measure interference among links and to detect the possibility of spatial reuse in WMNs  Envisioned scenario:  Should work for all Wireless mesh networks.  To be tested on an out- door medium range LACN type of network in FRACTEL deployment. FRACTEL example with LACN Ref: “FRACTEL: A Fresh Perspective on (Rural) Mesh Networks”, Kameswari Chebrolu and Bhaskaran Raman, NSDR’07 Cherukumilli Juvvalapalem Point-to-Point 802.11 Links with Directional Antennas Landline: wired gateway to the Internet Point-to-Multi-Point 802.11 link-sets using Sector Antennas 19 Km 19.5 Km Local- Gateway: gateway to LDN LACN: Local- ACcess Network at one of the villages (desired, not deployed) LDN: Long- Distance Network (deployed, in the Ashwini project) Jalli Kakinada Ardhavaram Kasipadu Alampuram Tetali Pippara Kesavaram Korukollu Polamuru Jinnuru Lankala Koderu Bhimavaram Tadinada IBhimavaram

6. Problem Statement  To device an automated mechanism to measure interference among links and to detect the possibility of spatial reuse in WMNs  Envisioned scenario:  Should work for all Wireless mesh networks.  To be tested on an out- door medium range LACN type of network in FRACTEL deployment. FRACTEL example with LACN Ref: “FRACTEL: A Fresh Perspective on (Rural) Mesh Networks”, Kameswari Chebrolu and Bhaskaran Raman, NSDR’07 Local- Gateway: gateway to LDN

7. Existing Work

8. How to measure interference? (Existing Work)  Exhaustive Measurements  Use pair-wise measurements to establish Interference  Link Interference Ratio (LIR): Ratio of good-put of two links operating separately vs. the good put when they operate simultaneously  O(N 4 ) measurements required for N nodes  Broadcast Interference Ratio (BIR): Uses broadcast measurements  O(N 2 ) measurements of N nodes REF : “Estimation of Link Interference in Static Multi-hop Wireless Networks” Jitendra Padhye, Sharad Agarwal, Venkata N.Padmanabhan, Lili Qiu,Ananth Rao, Brian Zill

9.  Prediction Models.  Distance Based Predictions / Path loss modes  Packet loss models  Pessimistic Estimations. - Highly incapable of representing real world scenarios How to measure interference? (Existing Work)

10. RSSI Based Measurements

11. RSSI based Measurement  Premise : There exists strong correlation between the received signal strength at a particular receiver from different senders and the amount of interference that a each of them exerts upon the other. Only O(N) measurements required for N-nodes. Can use the measured values to infer interference from multiple sources. Interference measurements are reliable compared to prediction based methods

12. Signal Strength vs. Interference @ Sender  At sender.  If above the sensitivity of radio, causes it to back-off while carrier sensing  Non-destructive Interference  RSSI received by S2 from S1- in Fig1 will determine the amount of carrier sense caused by S1 at S2  This relation need not be symmetric.

13.  At receiver.  Depends on the difference between interferer and sender signals strengths- SIR (signal to interference ratio)  Capture effect : If above SIR doesn’t exceed by some capture threshold packets are lost due to collision during reception  Destructive Interference.  The difference in Figure will determine how much the pair of links L1 and L2 interfere at R1 Signal Strength vs. Interference @ Receiver

14. Interference Cases  Interference occurred can be divided in cases where  Both senders carrier sense each other (channel shared)  One way hidden cases and one-way interference cases  Mutually hidden but still interfere with each other.  No interference cases (spatial reuse possible).

15. Other ISSUES  Effect of multiple interferers  Literature says multiple interference exists  How to gauge it?  Inter-channel interference  Can RSSI predict Inter-channel interference?  Transmission rates  How does interference vary with sender and interferer rates?

16. Measurements performed during the following work “On the Feasibility of the Link Abstraction in (Rural) Mesh Networks'', Dattatraya Gokhale, Sayandeep Sen, Kameswari Chebrolu, and Bhaskaran Raman” - The 25th Annual Conference on Computer Communications (IEEE INFOCOM), Phoenix, Arizona (USA), April 2008. RSSI – Analysis & Implications

17. Packet Error rate vs. RSSI  Existence of Steep region  High loss-rate even with less interfering packets  Use high power broadcasts Source : Thesis D.Gokhale IIT Kanpur

18. Packet Error rate vs. RSSI  Interference measurement should not be done at the steep region.  Use higher power broadcasts.  Helps if Interference range is higher than communication range.  Must validate linear transmission of SIR from High-power to low-power transmissions.

19. Variability of RSSI  Link Abstraction  Presence of variability  Forms a distribution with a peak  SIR can be determined by taking discrete convolutions  SIR between two nodes with probabilities P(X) and P(Y) is given as

20. Duration of Measurements ?  Implications of variability over Longer duration  Should determine the duration of measurements and interval between successive measurements  Average duration for >2db interval is ~25mins  More experimentation required for concluding. Graph plots CDF of every 1000 packet duration over an interval of 48hrs

21. Measurement Methodology

22.  Create a couple of link pairs  Interference at sender  Interference at receiver  Measure RSSI vs. Packet Error Rate  Simultaneous transmission from Sender and Interferer for 30s – Measure PER  Individual transmissions from sender and interferer for 30s - Measure RSSI  Use different transmission powers for interferer.  Determine the relation between RSSI and Interference

23. Final System Design  Each node broadcasts for a small duration (higher power), and also receives other broadcasts.  Interference is estimated according to the above determined relation  The measurement is repeated at regular intervals

24. Further Directions  Verify it in a real world deployment of size ~10 nodes  Evaluate the multiple interferer, inter-channel interference scenarios  Implement some TDMA using the interference graph generated.  Evaluate the interference measurement overhead

25. Thank you

26. References  “On the Feasibility of the Link Abstraction in (Rural) Mesh Networks‘”, Dattatraya Gokhale, Sayandeep Sen, Kameswari Chebrolu, and Bhaskaran Raman - The 25th Annual Conference on Computer Communications (IEEE INFOCOM), Phoenix, Arizona (USA), April 2008.  “Estimation of Link Interference in Static Multi-hop Wireless Networks” Jitendra Padhye, Sharad Agarwal, Venkata N.Padmanabhan, Lili Qiu,Ananth Rao, Brian Zill.  “Interference Map for 802.11” Dragos¸ Niculescu. IMC 2007.  “Quantifying the Interference GrayZone in Wireless Networks : A Measurement Study” Wonho Kim, Jeongkeun Lee, Taekyoung Kwon, Sung-Ju Lee, and Yanghee Choi. IMC 2007  “FRACTEL: A Fresh Perspective on (Rural) Mesh Networks”, Kameswari Chebrolu and Bhaskaran Raman, NSDR’07