Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An advanced scheme for queue management inTCP/IP networks

Abderrahmane Boudi, Malik Loudini

TL;DR

This paper addresses fairness among RTT-heterogeneous TCP flows by introducing FuzzyRTT, a stateless, RTT-aware AQM that leverages Explicit RTT Notification (ERN) to share per-flow RTTs with routers. It combines a queue-management fuzzy-PI controller with a RTT-based scheduling scheme that categorizes flows into five RTT bins (≤40, 80, 160, 320, 640 ms) and uses per-category controllers plus a RTT-interpolation mechanism to compute drop probabilities. Through OMNeT++ simulations, FuzzyRTT demonstrates superior fairness and near-optimal utilization across bandwidths and RTT distributions, at the cost of slightly higher transient queuing delays due to its lower update frequency. The work highlights deployment-friendly RTT-sharing approaches and points to future work on stability analysis and online parameter tuning. Overall, FuzzyRTT advances RTT-aware congestion control by integrating flow-specific RTT information into a pure AQM framework without requiring end-host protocol changes.

Abstract

Active Queue Management (AQM) is a key congestion control scheme that aims to find a balance between keeping high link utilization, minimizing queuing delays, and ensuring a fair share of the bandwidth between the competing flows. Traditional AQM mechanisms use only information that is present at the intermediate nodes (routers). They do not take into account the particularities of the flows composing the traffic. In this paper, we make use of a mechanism, called Explicit RTT Notification (ERN), that shares with routers information about the Round Trip Times (RTTs) of the flows. We propose a new fuzzy logic based AQM controller that relies on the RTTs of the flows to improve fairness between them. The performances of the new proposed method, FuzzyRTT, is examined and compared to existing schemes via simulation experiments.

An advanced scheme for queue management inTCP/IP networks

TL;DR

This paper addresses fairness among RTT-heterogeneous TCP flows by introducing FuzzyRTT, a stateless, RTT-aware AQM that leverages Explicit RTT Notification (ERN) to share per-flow RTTs with routers. It combines a queue-management fuzzy-PI controller with a RTT-based scheduling scheme that categorizes flows into five RTT bins (≤40, 80, 160, 320, 640 ms) and uses per-category controllers plus a RTT-interpolation mechanism to compute drop probabilities. Through OMNeT++ simulations, FuzzyRTT demonstrates superior fairness and near-optimal utilization across bandwidths and RTT distributions, at the cost of slightly higher transient queuing delays due to its lower update frequency. The work highlights deployment-friendly RTT-sharing approaches and points to future work on stability analysis and online parameter tuning. Overall, FuzzyRTT advances RTT-aware congestion control by integrating flow-specific RTT information into a pure AQM framework without requiring end-host protocol changes.

Abstract

Active Queue Management (AQM) is a key congestion control scheme that aims to find a balance between keeping high link utilization, minimizing queuing delays, and ensuring a fair share of the bandwidth between the competing flows. Traditional AQM mechanisms use only information that is present at the intermediate nodes (routers). They do not take into account the particularities of the flows composing the traffic. In this paper, we make use of a mechanism, called Explicit RTT Notification (ERN), that shares with routers information about the Round Trip Times (RTTs) of the flows. We propose a new fuzzy logic based AQM controller that relies on the RTTs of the flows to improve fairness between them. The performances of the new proposed method, FuzzyRTT, is examined and compared to existing schemes via simulation experiments.
Paper Structure (15 sections, 10 equations, 10 figures, 1 table)

This paper contains 15 sections, 10 equations, 10 figures, 1 table.

Table of Contents

  1. Introduction
  2. Background and related work
  3. FuzzyRTT Controller design
  4. Queue management
  5. Fuzzification and defuzzification
  6. The rule base
  7. The scaling factors
  8. Scheduling
  9. Simulation setup
  10. Results and discussion
  11. Varying the bandwidth
  12. Log normal distribution of RTT
  13. Transient state
  14. Deployment considerations
  15. Conclusion

Figures (10)

  • Figure 1: Membership functions of the FLC input variables
  • Figure 2: Membership functions of the FLC output variable $\Delta p_{i}(kT)$
  • Figure 3: Membership functions of the input variable $RTT$
  • Figure 4: Membership functions of the output variable $p$
  • Figure 5: Dumb bell topology
  • ...and 5 more figures