1. 1 A Coordinate-Based Approach for Exploiting Temporal-Spatial Diversity in Wireless Mesh Networks Hyuk Lim Chaegwon Lim Jennifer C. Hou MobiCom 2006 Modified and Presented by Jihyuk Choi

2. 2 Contents Introduction Interference in multiple-hop wireless networks Proposed approach to mitigate interference –Topology discovery –Transmission scheduling –Packet transmission Experiment Results Conclusion

3. Wireless Mesh Network 3 A wireless network that allows wireless nodes to supply backhaul services to other nodes. Soekris board Mesh node Wireless ad-hoc network, multihop, GN(Gateway Node), static

4. 4 Interference in Mesh Network Interferences –Inter flow interference: interference between difference flows. –Intra flow (self) interference: interference between consecutive wireless links in the same flow Intra flow-interferenceInter flow-interference

5. 5 Mitigating Interference To mitigate interference and maximize the network capacity, there are several control knobs: –Transmit power: topology control. –Carrier sense threshold: trade-off between spatial reuse and interference level. –Channel diversity: use of non-overlapping channels. –Scheduling concurrent transmissions for least-interference connection. In this paper, the authors consider the problem of mitigating interference and improving network capacity from the angle of temporal-spatial diversity

6. 6 Interference Range Intra flow Interference Example of a single flow Pk 1 ABCDE Pk 2 F Observation: –The intra flow interference is considered as a self capacity limiting mechanism. –It cannot be avoided in a single flow.

7. 7 Temporal-Spatial Diversity What if we schedule packet transmissions as follows: case 1 case 2 WYZABCD Packet Transmission at node A time t Packets in node A’s queue sender   1 2 3 4 5 6 1 2 3 4 5 6

8. 8 Assumption The authors focus on transporting downstream traffic at gateway nodes. –most of the Internet accesses are intended for downloading large video/audio/text files –by virtue of the way how wireless mesh networks operate, all the downloaded traffic is handled by gateway nodes The authors restrict the measurement area to be within two hops from GN (Gateway Node)

9. 9 Issues to Be Considered Topology discovery –How to establish network topology to predict interference between nodes. Transmission scheduling –How to find sets of nodes that result in the least inter flow interference. Packet transmission –How to interleave packet transmissions for least- interference connections.

10. 10 Topology Discovery Goal: to facilitate the prediction of received signal strength (RSS) or interference strength between nodes. RSS prediction –Direct measurement: possible between neighbor nodes. –Indirect estimation: Signal from a non-neighbor node cannot be decoded. Use geographic locations and path loss model. Use a coordinate-based network topology constructed with pairwise RSS measurements.

11. 11 Topology Discovery (cont’d) Procedures: Pairwise RSS Measurements Cartesian Coordinate System Singular Value Decomposition (SVD) RSS Prediction Distance Metric Example of 3D representation distance  level of interference

12. 12 Topology Discovery (cont’d) - Notations M(GN) : the set of neighbor nodes that can directly communicate with GN and GN itself. The RSS measurements are represented by the p*p square matrix S. ( p = |M(GN)| ) The ith column vector of S, which denoted by s i, is the (-RSS)s measurement made in dBm by the ith node from all nodes in M(GN). –As the sign of the RSS measurement is negated, a smaller value of s i,j implies stronger signal strength. Pairwise RSS Measurements

13. 13 Project the p-dimensional space into a new q- dimensional space. Example of PCA (Principal Component Analysis) SVD (singular value decomposition) Topology Discovery (cont’d) Pairwise RSS Measurements Cartesian Coordinate System

14. 14 Topology Discovery (cont’d) Example (cont’d) –SVD of matrix D –Calculate coordinates of hosts in two-dimensional coordinate system Pairwise RSS Measurements Cartesian Coordinate System

15. Topology Discovery (cont’d) Find the optimal scaling factor α * that minimizes the following function is 0.6 The new coordinate of a node is written by 15 Pairwise RSS Measurements Cartesian Coordinate System

16. 16 Topology Discovery (cont’d) Determining coordinates for nodes that are two hops away GN Transmission range i j i’ ik k Pairwise RSS Measurements Cartesian Coordinate System

17. 17 Issues to Be Considered Topology discovery –How to establish network topology to predict interference between nodes. Transmission scheduling –How to find sets of nodes that result in the least inter flow interference. Packet transmission –How to interleave packet transmissions for least- interference connections.

18. 18 Transmission Scheduling Computing SNR between two-hop neighbor nodes to get least-interference nodes. If SNR  , the j th node is not an interfering node to the i th node. SNR = Pairwise RSS Measurements Cartesian Coordinate System RSS Prediction

19. 19 Transmission Scheduling (cont’d) Determining the transmission order of least- interference nodes. Procedure: –Pick the first packet in the queue. –Search up to N packets to obtain the set of non-interfering nodes. Pk 1 Pk 2 Pk 3 Pk 4 Pk 5 Pk 6 Queue of a node Select the first pkt......... Select more pkts depending on SNR.

20. 20 Issues to Be Considered Topology discovery –How to establish network topology to predict interference between nodes. Transmission scheduling –How to find sets of nodes that result in the least inter flow interference. Packet transmission –How to interleave packet transmissions for least- interference connections.

21. 21 Packet Transmission Basic idea: If a node is congested, it has to have a higher priority over neighbor nodes. –Without backoff, send packets in a bulk, and take a longer pause (backoff) time.  : # of packets sent in the previous transmission. busy Frame ACK SIFS BACKOFFDIFS Congested node Two nodes belonging to the same set of least interference nodes.

22. 22 Experiment Results We focus on transporting downstream traffic at gateway nodes –Gateway nodes are responsible for transporting a large amount of downstream traffic Champaign-Urbana community wireless network (CUWiN)

23. 23 Experiment Results Augmented NS-2 simulation –Real topology of CUWiN + Random topology Visualization of 2D coordinate system Throughput performance 20 % throughput improvement obtained !

24. 24 Experiment Results NS-2 Simulation –Star topology with multiple wireless paths –Transmission range: 100m, Interference range 220m

25. 25 Experiment Results Throughput performance 27 ~ 30 % throughput improvement obtained !

26. 26 Conclusion A coordinate-based approach is proposed for representing network topology and mitigating interference in wireless mesh networks. Future work –Topology construction with various performance metrics such as packet loss rate and delay. –More experiments in a large scale mesh network.