1. Wireless Mesh Networks 1

2. Architecture 2

3. Wireless Mesh Network A wireless mesh network (WMN) is a multi-hop wireless network that consists of mesh clients and mesh routers. Mesh routers form the backbone of WMNs. Some of mesh routers are called gateway nodes that are connected to the wired Internet. provide Internet access 3

4. Benefits Reduction of installation costs Only a few mesh router have cabled connections to the wired network. Large-scale deployment WLAN: One hop communication has limited coverage. WMN: Multihop communication offers longer distance communication through intermediate nodes. Reliability Redundant paths between a pair of nodes in a WMN increases communication reliability. Self-Management A WMN is a special ad hoc network. 4

5. Applications broadband home networking community and neighborhood networking enterprise networking metropolitan area networks transportation systems building automation health and medical systems security surveillance systems … 5

6. Multi-channel Multi-Radio System There are multiple non-overlapping channels IEEE 802.11b/a standards offer 3 and 12 non- overlapping channels, respectively. Each node is equipped with multiple radios interference reduction communicate with more than one neighbor at the same time full duplex operation throughput improvement 6

7. 1-Radio vs. 2-Radios 7

8. 8

9. Serving both Mobile Users and Downlink 9

10. Topology Control in WMNs A topology consists of a set of nodes and links, and it describes the connectivity information of the network. Links in topology are the result of some controlled parameters, such as transmission power and channel assigned. A good topology is critical to network performance. too dense  energy consumption & interference  throughput  too sparse  long path, disconnected network Reducing energy consumption and interference may be conflicting goals. We focus on topology control for reducing interference. 10

11. Topology Control in WMNs Topology control in WMNs includes two steps: Power adjustment Channel assignment Power adjustment Define the physical topology of network A link between two nodes if they are reachable via transmission power Channel assignment Define the logical topology on the top of the physical topology A link between two nodes if they are reachable and use a common channel 11

12. Network Model V : A set of nodes, representing the wireless devices in the Euclidean plane. p max (v): the maximum transmission power of node v p(u, v): the least required energy to transmit a message from u to v G(V, E): network graph, any link e = (u, v)  E if p max (u) ≥ p(u,v)// reachable by max power G P (V, E P ): physical topology, E P  E G P is a subgraph of G, which is obtained by removing some long-distance links through power control 12

13. Network Model C: number of channels Q(v): number of radios on node v, and typically Q(v) < C A(v): the set of channels assigned to v, |A(v)|=Q(v) G L (V, E L ): logical topology, any logical link e = (u, v; k)  E L iff (u, v)  E P and k  A(u)  A(v) There may be multiple logical links between a pair of nodes in G L, and it is a multi-graph. 13

14. Example of physical & logical topologies network graphphysical topology logical topology 14

15. Interference Model Interference model specifies conditions where a signal can be successfully received. Physical Model transmission from u to v (SNR: signal-to-noise ratio, SS: signal strength) 15

16. Interference Model Protocol Model (transmission from u to v) p(u)  p(u, v), and no other interfering transmitter w, such that: d(w, v)  (1 +  )∙ d(u, v) (  > 0) // d(w,v): distance Physical Interference Model Based on SNR level at receiver side IEEE 802.11 MAC protocol RTS-CTS Symmetrical communication: Both the sender and the receiver should be free from interference for a successful transmission. 16

17. Objectives of Interference Reduction Interference reduction based on network topology only network planning MIN interference while keeping certain network properties, such as k-connectivity and spanner Interference reduction based on network topology and traffic flows between nodes network planning and routing MAX network throughput 17

18. Power Adjustment Reduce interference by reducing transmitting power Link Interference Reduction define the interference of a link Node Interference Reduction define the interference of a node 18

19. Link Interference Reduction Minimize the number of nodes that are within the interference range of a link set of nodes interfered by link e = (u,v): I(e) ={w  V| d(u, w)  d(u, v)}  { w  V| d(v, w)  d(v, u)} The network interference can be defined as the maximum (or total) number of nodes interfered by any (or all) links in the physical topology. MST is the optimal solution when minimizing the max link interference in a connected physical topology. 19

20. Node Interference Reduction Minimize the number of nodes that are within the interference range of a node the transmission power of u: the number of nodes interfered by node u: I(u) = | {v| p(u,v) ≤ p(u)} | Minimize the max node interference while keeping the network k- connected or spanner Mnimize the total node interference in a connected topology I(v) = 4 I(u) = 1 20

21. Channel Assignment Efficient channel assignment can greatly reduce the network interference Two links interfere with each other iff one node of a link is within the interference range of the other link and both link use the same channel Channel assignment Methods dynamic assignment: assign channels transceivers at flow setup time (or even at packet level) Require channel switch at very fast scale difficult to let two end-nodes of an edge agree on the same channel (particularly at packet switching level) static assignment: assign channels to all links statically waste resource (not all link are used simultaneously) 21

22. Channel Assignment: conflict graph 22

23. Static Channel Assignment Algorithm Goal of channel assignment: assign channels to network links such as the total number of interfering links (after channel assignment) is minimized Two links within the interference range will not interfere with each other if they are assigned with different channels Given conflict graph G c (V c,E c ) and K channels, Max-K cut problem: partition V c into K partitions, aiming to maximize the number of edges whose end-points lie in different partitions. Channel assignment problem for minimizing total interference is equivalent to Max-K cut problem. Many heuristic algorithms were proposed to solve it. 23

24. A Distributed Channel Assignment Method Starting from any node in a network graph, partition its neighbors into K groups and assign one group (i.e., one channel) to one of its radios Choose one of the previous node’s neighbor to continue the above process, while maintaining the channels already assigned by the previous node Repeat the above process until all nodes have channels assigned 24 Example of a grid network, K = 4, r = 2

25. Joint Channel Assignment & Routing Cyclic Dependency Problem: Given traffic demand from each node, a cyclic dependency between routing and channel assignment Routing depends on link capacity and link capacity depends on channel assignment (due to interfering links) But, channel assignment depends on traffic load on links, which depends on routing How to break this cycle? 25

26. Iterative Method of Channel Assignment & Routing 26 Iterative Improvement Method: 1.Start with initial link load estimation without regard of channels 2.Assign channel to each radio (of nodes) according to link traffic load 3.Do the routing again based on the actual link capacity calculated from the assigned channels. 4.Repeat the above step until all link traffic load is less than the capacity

27. Future Work Which interference criterion is more proper? What is the appropriate optimizing objective? Many optimization problems of topology control are NP-hard so that efficient algorithms are required. especially for channel assignment Distributed algorithms for practical networks. Consider power adjustment and channel assignment together. Interference-aware routing QoS call admission QoS multicast call admission 27