2006한국멀티미디어학회 국제학술대회Requires access

PROBABILITY ADAPTIVE RED IN TCP/IP NETWORKS

Taehoon Kim, Aeri Lee, ChangHee Kim, KeeHyun Lee

Open publisher page 0 citations

Abstract

The proposed AQM algorithm by the IETF is Random Early Detection (RED). The RED algorithm allows network operators simultaneously to achieve high throughput and low average delay. However, the resulting average queue length is quite sensitive to the level of congestion and the RED parameter settings, and is therefore not predicable in advance. To achieve a predictability of performance would require constant tuning of the parameters to adjust to current traffic conditions. We evaluate the performance of the RED and Adaptive RED algorithms. We propose a Probability Adaptive RED (PARED) that can maintain high throughput and low packet drop under dynamic traffic loads.

About this research paper

What this paper is about

The proposed AQM algorithm by the IETF is Random Early Detection (RED). The RED algorithm allows network operators simultaneously to achieve high throughput and low average delay. However, the resulting average queue length is quite sensitive to the level of congestion and the RED parameter settings, and is therefore not predicable in advance. To achieve a predictability of performance would require constant tuning of the parameters to adjust to current traffic conditions. We evaluate the performance of the RED and Adaptive RED algorithms. We propose a Probability Adaptive RED (PARED) that can maintain high throughput and low packet drop under dynamic traffic loads.

Why it matters

A significance statement is not available in the OpenAlex record.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

The proposed AQM algorithm by the IETF is Random Early Detection (RED). The RED algorithm allows network operators simultaneously to achieve high throughput and low average delay. However, the resulting average queue length is quite sensitive to the level of congestion and the RED parameter settings, and is therefore not predicable in advance. To achieve a predictability of performance would require constant tuning of the parameters to adjust to current traffic conditions. We evaluate the performance of the RED and Adaptive RED algorithms. We propose a Probability Adaptive RED (PARED) that can maintain high throughput and low packet drop under dynamic traffic loads.

Key concepts: Random early detection, Computer science, Throughput, Active queue management, Computer network, Network packet, Queue, Network congestion

Related papers

Back to paper searchBrowse research topicsOriginal source
PROBABILITY ADAPTIVE RED IN TCP/IP NETWORKS — Research Paper | ScholarLens