Understanding Congestion Control in Multihop Wireless Mesh Networks

1
Understanding Congestion Control in Multi-hop
Wireless Mesh Networks
Sumit Rangwala, Apoorva Jindal, Ki-Young Jang,
Konstantinos Psounis, and Ramesh
Govindan University of Southern California
Problem Description Congestion Control in
802.11x Wireless Networks
What is wrong with TCP?
Congestion is a Neighborhood Phenomenon
Link sender-receiver pair
Stack Topology
TCP starves the middle flow
Proposed Solution Neighborhood-Centric Transport
WCP AIMD Based Transport Protocol
WCPCap Explicit Rate Control Protocol

• Neighborhood congestion detection
• Any link in the link neighborhood is congested
• Neighborhood RTT
• Maximum RTT of all flows traversing a link
  neighborhood
• End-to-end behavior
• Reduce a flows rate if at least one of the
  traversed neighborhoods is congested
• Use the maximum RTT among all traversed
  neighborhood RTTs to clock the rate changes

• Neighborhood per flow rate
• Calculates sustainable fair rate for each flow in
  a link neighborhood
• Requires spare capacity calculation
• End-to-end behavior
• Flows send at a rate that is minimum of the rate
  assigned at each traversed neighborhood

Evaluation WCP fairer than TCP, WCPCap achieves
max-min fairness
Experimentation Setup

• Simulation
• Qualnet 3.9.5
• 802.11b with default parameters
• 11Mbps , no rate adaptation, 512 byte data packet
• Zero channel losses
• Buffer size 64 packets
• Real World Experiments
• Click modular router on Linux
• Same code as in simulation

Simulations Diamond Topology
Chain-cross Topology
Simulations Chain-cross Topology
Simulations Stack Topology
Real-world Experiments Arbitrary Topology
Real-world Experiments Stack Topology