Distributed Hop-by-Hop Rate Adjustment for Congestion Control in Sensor Networks

1
Distributed Hop-by-Hop Rate Adjustment for
Congestion Control in Sensor Networks

• Presented by
• Vipul Bhasin
• Sapna Dixit
• Vishal Kumar Singh
• Manmohan Voniyadka

2
Goal

• Implement a distributed rate adjustment strategy
  to minimize congestion
• Adjust the source rate using congestion factor
  for better response

3
Design

• Calculate congestion factor
• Used queue length, channel load, and
  retransmission counter
• Adjust source rate when consecutive congestion
  factor found to increase
• Propagate cumulative congestion factor upstream
• Propagated only when threshold exceeded

4
Code Changes

• Suppress Message Behavior
• sendSupressMsg
• suppressMsgRcvd
• calcCongestionFactor
• Source Rate Adjustment (via AIMD)
• suppressMsgRcvd
• NAK Implementation (packet dropped from queue as
  criteria to post NAK)
• sendNakMsg
• NakMsgRcvd

5
Experiments

• Simple Model of VC vs. EC
• Varied source rates from 5 to 20
• EC outperformed VC
• Lossy Model of VC vs. EC
• Two models one-hop and generated (from lossy
  builder)
• Comparable results in generated and EC
  outperformed VC in one-hop
• Lossy Model Topology of VC vs. EC
• EC outperformed VC in new lossy topology

6
Results
7
Pros/Cons

• Pros
• Higher Responsiveness
• Local decisions can be made based on global
  congestion scenario
• NAK counter better estimates congestion
• Cons
• More suppress messages due to greater propagation
  of congestion factor