1. STUMP: Exploiting Position Diversity in the Staggered TDMA Underwater MAC Protocol Kurtis Kredo II, Petar Djukic, Prasant Mohapatra IEEE INFOCOM 2009

2. Outline  Introduction  Staggered TDMA Underwater MAC Protocol  Network Model  Conflict-Free Scheduling  Scheduling Algorithms  Numerical Results  Conclusions

3. Introduction  Acoustic communication forces protocol designers for underwater networks  Avoiding or reducing collisions becomes vitally important  Previous work has focused on overcoming the challenges of the acoustic channel  STUMP schedule overlapping transmissions without collisions

4. Staggered TDMA Protocol  To communications without collisions and increase channel utilization STUMP nodes develop schedule constraints share propagation delay estimates share time slot requirements among two-hop neighbors

5. Network Model  Scheduled protocols require nodes maintain time synchronization define σ as the maximum synchronization error at any node from a global time any two nodes differs by at most 2σ  Define π as the maximum error experienced in estimating the one way propagation delay

6. Four Possible Conflicts

7. Conflict-Free Scheduling  A set of time slots assigned to each node for transmission, S = {s i }, prevents all conflicts  Define constraints, C, ensure node transmission times are sufficiently separated  Binary ordering variables, O = {o ij } o ij = 1, node i transmits before receiving node j ’ s packet  Finding sets S and O that satisfy the schedule constraints C and node demands Δ

8. STUMP Schedule Constraints  Node i transmitting to neighbor j transmits for Δ ij time slots starting in slot s ij  Define p ij as the propagation delay from node i to node j  p ij may not equal p ji due to different propagation paths between the two nodes

9. RX-RX Conflicts (1)  Node k finishes receiving node i ’ s packet  This yields the inequality

10. RX-RX Conflicts (2)  Node j ’ s transmission does not cause a collision with node i in the next frame Node i : 2 slot packet Node j : 3 slot packet K m Frame size

11. RX-RX Conflicts (3)

12. TX-RX-TX Conflicts  Ensure an interference packet does not arrive at a node while it is receiving a valid packet nearly identical to the RX-RX conflict

13. TX-TX Conflicts  Neighbor 1 is farther than neighbor 0, neighbor 2 is farther than neighbor 1  Node k is the next farther neighbor than node j from node i

14. TX-RX Conflicts  TX-RX conflicts ensure nodes do not transmit while receiving a packet

15. TDMA Schedule Constraints  Define G i as the guard slots required after the transmission of node i using TDMA  Ensure nodes cause interference to each other are assigned non-overlapping time slots

16. Centralized Scheduling Algorithms  An appropriate objective function, minimum frame size or minimum uplink delay yields an integer linear programming problem enough computational resources require significant overhead to collect the network information

17. Distributed Scheduling Algorithms  Determine the ordering variables, o ij, by prioritizing nodes node i has a higher priority than node j, then oij = 1, otherwise o ij = 0  With fixed ordering, scheduling constraints become a set of difference equations Bellman-Ford algorithm

18. Determine Ordering Variables  A simple way to find node priorities is to select them at random does not guarantee any level of performance  Leaf nodes have the highest priority (an RX-RX conflict) may get the same priority may resulting in non-optimal uplink delay.

19. Numerical Results  Compare STUMP with the TDMA and Aloha protocols evaluating average throughput and delay performance over 100 random topologies  Nodes small movements caused by ocean currents the synchronization and propagation delay estimate error parameters σ and π

20. Normalized Throughput as σ Varies

21. Average Maximum Uplink Delay

22. Average Maximum Delay as σ Varies

23. Traffic Load Varies

24. Conclusion  The Staggered TDMA Underwater MAC Protocol increases the performance by using propagation delay estimates to schedule overlapping transmissions  Provide users to develop scheduling algorithms suited to their application