1. IEEE Globecom 2010 Tan Le Yong Liu Department of Electrical and Computer Engineering Polytechnic Institute of NYU Opportunistic Overlay Multicast in Wireless Network

2. Outline  Introduction  Motivation  Goal  Network Model and Assumptions  Definition  Minimum Steiner Tree Opportunistic Routing  Simulation  Conclusion

3. Introduction  Deliver multicast services efficiently over multi-hop wireless networks:  Increasing popularity of wireless devices  New wireless applications  It is well-known that the general minimum-cost multicast routing problem is NP-hard.  Interferences  Collisions  Volatile link

4. Introduction  Opportunistic Routing (OR) has been proposed to improve the efficiency of unicast in multi-hop wireless networks.  The advantage of OR:  Reduce the number of transmissions necessary to deliver a packet.  Combine multiple weak links into one strong link. S A BD C Traditional Routing: S→ B→D 4 + 4 = 8 Opportunistic Routing: S →ABC 1/(1-0.75^3)=1.73 A|B|C→D 4 S→ABC→D 1.73+4=5.73 25%

5. Introduction  Overlay networks have been widely employed to deliver multicast services on the Internet.  A virtual network that is built on top of an existing network  Implement a service that is not available in the existing network.

6. Introduction  For wireless multicast, we build an overlay Steiner tree to connect the source with all receivers. Src Recv 1 Recv 2

7. Introduction Goal  Build the Minimum Steiner Tree with Opportunistic Routing (MSTOR).  Minimize the total transmission cost to reach all receivers

8. Network Model and Assumptions  We consider a network of N static wireless nodes, including  One source node S  A set of K receivers R={R 1, R 2, …, R k }, where K<N  N-K-1 relay nodes  Wireless links between neighbor nodes are not reliable.  The success probability of packet transmission on a link is depends on  Distance  Density  Traffic  MAC scheduling scheme

9. Network Model and Assumptions  The length of a OR path from node i to node j is the expected number of packet transmissions to send a packet from i to j along the OR path. ijk 25% 10% 4 10 14

10. Minimum Steiner Tree Algorithm  S AB R CD 11 11 3 2 6 S R Total length = 6 S AB R CD 1 3 6 S R Total length = 5 AB C 11 1 2

11. Minimum Steiner Tree Algorithm  Approximate algorithm: Shortest Path algorithm  Start with a subtree T consisting of one terminal node  While T does not span all terminals Select a terminal x not in T that is closest to a vertex in T. Add to T the shortest path that connects x with T.  Delete non-terminals of degree 1 from this spanning tree  Until there are no such non-terminals S AB R CD 3 6 S R AB C 11 1 2 D 1

12. Simulation  Compare the performance of MSTOR with  The shortest path unicast routing (UR) The source sends packets to each receiver one by one  Unicast opportunistic routing (OR) The source sends packets to each receiver one by one by following the optimal unicast OR path.  Tree-based unicast routing (tree-based UR). packets are sent to receivers along the minimum Steiner tree with the unicast distance as link weight  The sending rate at physical layer is set to be 250Kb/s.  We configure the transmission power with the effective radio coverage of 200m.

13. Simulation  Topology  The source is node 1, with two receivers node 7 and 10.  The grid size is 70m.  The average packet loss rate on each wireless link is around 10%

14. Simulation  Number of packets delivered  MSTOR outperforms other routing schemes

15. Simulation  Number of packets delivered  This demonstrates that MSTOR can improve the multicast efficiency in wireless networks with high link loss rate.

16. Simulation  Topology  30 static nodes randomly located in an area of 1000m×1000m.  We randomly choose one source and five receivers

17. Simulation  Number of packets delivered  The big improvement of MSTOR is originated from the gains OR distance higher packet loss rates higher wireless node density

18. Simulation  Packet latency  The time lag from the source sends out the first copy of a packet until all receivers receive the packet successfully.  This is because of the advantages of dynamic routes of OR over the static routes of UR in a lossy wireless network.

19. Conclusion  In this paper, we studied opportunistic overlay multicast for wireless networks.  MSTOR  simple protocol  can be easily deployed in multi-hop wireless networks  achieve high multicast efficiency