实时多播流的弹性公平性和基于门限的拥塞控制策略

该文提出了关于实时IP多播流基于速率弹性的公平性定义，并结合有关多播组规模估计的机制和多播拥塞控制算法。实现了一种基于门限的实时多播业务的等级拥塞控制策略，使实时多播流在满足瞬态弹性公平性的同时，也基本满足稳态的比例公平性，并对此进行了仿真验证。

Elastic Fairness Definition and Threshold-Based Congestion Control Policy for Real-Time IP Multicast Flows

This paper proposes a new elastic fairness for real-time IP multicast traffic. The definition can be applied in performing flow and congestion control for real-time IP multicast traffic, together with multicast group measurement and multicast congestion control mechanisms. Different services have different service requirements, and data service is more sensitive for bandwidth, although real-time service focuses more on delay and jitter. Elastic fairness specifies different levels for different sessions from the jitter point of view. So fairness is guaranteed not only in bandwidth, but also in rate variation. Authors further devise two algorithms to apply the generalized fairness concept in congestion control. Time-based algorithm adjusts sending rate gradually in a step length of round trip times. Loss rate-based algorithm adjusts sending rate by different equation, which depends on the session weights. Therefore, real-time multicast flow meets both elastic fairness in small time scale and proportional fairness in large time scale.