Towards a Robust Protocol Stack for Diverse Wireless Networks

1. Towards a Robust Protocol Stack for Diverse Wireless Networks Arun Venkataramani (in collaboration with Ming Li, Devesh Agrawal, Deepak Ganesan, Aruna Balasubramanian, Brian Levine, Xiaozheng Tie at UMass Amherst)

2. Cellular DTN Mesh WLAN MANET High mobility Little mobility Some mobility Wireless network landscape today • Diverse wireless networks coexist • Adapt TCP/IP stack in different ways

3. Cellular Mesh WLAN MANET DTN High mobility Little mobility Some mobility Interconnecting diverse networks? • Vision: A simple robust protocol stack to interconnect diverse wireless networks.

4. Cellular Mesh WLAN MANET DTN High mobility Little mobility Some mobility Why does TCP/IP not suffice? • E2E route disruptions cause unavailability

5. Cellular Mesh WLAN MANET DTN High mobility Little mobility Some mobility Why does TCP/IP not suffice? • Gap between what you buy vs. get Advertised capacity: 11Mbps

6. Principle: Design for high uncertainty • Observe what works at an extreme point in the design space--always-partitioned DTNs--for insights into a robust design.

7. Outline • Motivation • Block transport • Replication routing • Research challenges

8. 1. E2E Transport • E2E rate control is error-prone • E2E retransmissions are wasteful E2E Feedback Rate Control Dest Source Congestion? Link loss?

9. 1. E2E Transport • E2E rate control is error-prone • E2E retransmissions are wasteful E2E Retransmissions P Source Dest Redundant Transmissions

10. RTS/CTS Listen Backoff 2. Packet as Unit of Control • Channel access • Link layer ARQ

11. 2. Packet as Unit of Control • Channel access • Link layer ARQ Backoff Timeout Timeout Backoff

12. 3. Complex Cross-Layer Interaction • Link-layer ARQ/backoff hurts TCP rate control Rate Control Highly Variable RTT Transport Link Link ARQ Link ARQ Link ARQ

13. Hop: Clean-Slate Transport Protocol • End-To-End • Packets • Complexity Hop-by-Hop Blocks Minimize interaction Block-switched networks: A new paradigm for wireless networks, Li et al. [NSDI 2009]

14. Hop Design Virtual Retransmission Backpressure Multi-hop Per-hop ACK Withholding Micro-block Prioritization Reliable Block Transfer

15. TXOP CSMA Reliable Per-Hop Block Transfer • Mechanisms • Burst mode (TXOP) • Block ACK based ARQ • Benefits • Amortizes control overhead

16. Hop Design Virtual Retransmission Backpressure Multi-hop Per-hop ACK Withholding Micro-block Prioritization Reliable Block Transfer

17. C E A D B B-SYN B-ACK DATA Virtual Retransmission (VTX) • Mechanism • Leverages in-network caching • Re-transmits blocks only when unavailable in cache • Benefits • Fewer transmissions • Low overhead • Simple

18. E A D B B-SYNVTX B-SYN VTX Timer B-ACK E2E ACK VTX Timer Virtual Retransmission (VTX) • Mechanism • Leverages in-network caching • Re-transmits blocks only when unavailable in cache • Benefits • Fewer transmissions • Low overhead • Simple

19. Hop Design Virtual Retransmission Backpressure Multi-hop Per-hop ACK Withholding Micro-block Prioritization Reliable Block Transfer

20. B-SYN B-SYN B-SYN Backpressure • Mechanism • Limits outstanding blocks per-flow at forwarder Hold B-ACK Slow A B C D E Dest Source Limit on outstanding blocks=2

21. Backpressure • Mechanism • Limits outstanding blocks per-flow at forwarder • Benefits • Improves network utilization 20Mbps D E Dest 20Mbps 20Mbps 10Mbps A B C 20Mbps Source 1Mbps F G Dest Aggregate goodputwithout backpressure: 6Mbps

22. Backpressure • Mechanism • Limits #outstanding blocks per-flow at forwarder • Benefits • Improves network utilization 20Mbps D E Dest 20Mbps 20Mbps 10Mbps A B C 20Mbps Source 1Mbps G F Dest Limit of Outstanding Blocks=1 Aggregate goodputwith backpressure: 10Mbps

23. Hop Design Virtual Retransmission Backpressure Multi-hop Per-hop ACK Withholding Micro-block Prioritization Reliable Block Transfer

24. B-SYN B-SYN A B C B-ACK Withhold B-ACK DATA B-ACK DATA Ack Withholding • Mechanism: • Receiver withholds all but one BACK • Benefit: • Low overhead • Less conservative • Simple

25. Hop Design Virtual Retransmission Backpressure Multi-hop Per-hop ACK Withholding Micro-block Prioritization Reliable Block Transfer

26. FTP SSH B-SYN B-SYN DATA B-ACK Micro-block Prioritization • Mechanisms • Sender piggybacks small blocks to B-SYN • Receiver prioritizes small block’s B-ACK • Benefits • Low delay for small blocks Sender Receiver Sender 64B 1MB

27. Testbed • 20 nodes on the 2nd floor of UMass CS building • Apple Mac Mini • 1.8GHz, 2GB RAM, Atheros 802.11 a/b/g card

28. Single-flow Single-hop Performance 1.2x TCP Hop 1.6x 28x Hop achieves significant gains over TCP

29. Single-flow Multi-hop Performance 1.9x TCP Hop 2.3x 2.7x Hop achieves significant gains over TCP

30. Graceful Degradation with Loss Emulated link layer losses at the receiver TCP Hop TCP breaks down at moderate loss rates

31. Mean goodput Scalability to High Load • 30 concurrent flows 2x 20x TCP Hop 150x Hop-by-hop TCP Hop is much fairer than TCP

32. Hop Performance Breakdown

33. AP Low Delay for Small Transfers • 1 small transfer competing with 4 large Small transfer size (KB) Hop lowers delay across all file sizes

34. Related Work • Fixing E2E rate-control • Separating loss/congestion [Snoop, WTCP, Westwood+, ATCP, TCP-ELFN] • Network-assisted rate control [ATP, NRED, IFRC, WCP] • Hop circumvents rate control • Backpressure • ATM, theoretical work [Tassiulas et al.] • Tree/chain sensor data aggregation [Fusion, Flush] • Reliable point-to-point transport [RAIN, CXCC, Horizon] • Hop reduces backpressure overhead using blocks • Batching • Common optimization at link [802.11e/802.11n, WiLDNet, Kim08, CMAP], transport [Delayed-ACK, DTN2.5], and network [ExOR]layers • Hop leverages batching across layers

35. Outline • Motivation • Block transport • Replication routing • Research challenges

36. Replication routing P P P P • Replication reduces delay under topology uncertainty X Z Y W

37. Routing as resource allocation problem • Problem: Which packets to replicate given limited bandwidth to optimize a specified metric? RAPID Protocol (X,Y): 1. Control channel: Exchange metadata 2. Direct Delivery: Deliver packets destined to each other 3. Replication: Replicatein decreasing order of marginal utility 4. Termination: Until all packets replicated or nodes out of range Y X Change in utility Packet size DTN routing as a resource allocation problem, Balasubramanian et al. [SIGCOMM 2007]

38. Objective: Minimize average delay Define U(i) = -(T + D) T = time since created, D = expected remaining time to deliver Simplistic assumptions uniform exponential meeting with mean ¸ global view Utility computation example D = ¸ D = ¸/2 D =¸/3 X Y W Z i i i

39. Utility computation example (cont’d) j j i Deadline of i < T Deadline of j = T1 > T Metric: Min average delay Metric: Maximize #packets delivered within deadline X Y W Z Replicate j Replicate i

40. Deployment on DieselNet

41. Results based on DieselNet traces

42. Outline • Motivation • Block transport • Replication routing • Research challenges

43. DTN Mesh WLAN MANET High uncertainty Replication useful Low uncertainty Forwarding suffices C1: When to replicate? relays • Replication useful under • Topology uncertainty • Delay optimization • Light load scenarios X1 X2 src dst Claim: Delay benefit of replication is unbounded. Xk Xi = r.v. for delay of path i Forwarding delay = mini(E[Xi]) Replication delay = E[mini(Xi)]