High Performance Fairness Algorithm for Resilient Packet Ring

Resilient Packet Ring (RPR) is MAN technology with two counter-rotating rings that multiple stations share the bandwidth. The stations on ring must negotiate the allowed rate that they can transmit fairness eligible traffic based on the total amount of uncommitted bandwidth. RPR draft employs distributed bandwidth control algorithm in order to assure global fairness. In this paper, we suggest a new fairness control algorithm termed Congestion Distributed Fairness Algorithm (CDFA) that exhibits better pe...

Congestion Control Algorithm for Resilient Packet Ring

A congestion control algorithm is proposed for resilient packet ring (RPR) in this paper. In this algorithm, nonlinear explicit rate feedback control is used to ensure fast convergence and smooth equilibrium behavior. The algorithm combines explicit rate control with a deficit round robin (DRR) scheduler, which not only ensures fairness, but also avoids the implementation difficulties of explicit rate control algorithms. The algorithm has good features of fairness, fast convergence, smooth equilibrium, low queue depth, and easy implementation. It is insensitive to the loss of congestion control packets and can adapt to a wide range of link rates and network scales. It has solved the unbalanced traffic problem of spatial reuse protocol (SRP). The algorithm can be implemented on the multi-access control layer of RPR nodes to ensure fair and efficient access of the best-effort traffic.

Flood Avoidance Mechanisms for Bridged Resilient Packet Rings

Resilient Packet Ring （RPR）, or the Standard IEEE 802.17, is a new IP-based network technology proposed to replace SONET/SDH in metropolitan area networks. RPR is well-adapted to handle multimedia traffic and is efficient. However, when RPR networks are bridged, inter-ring packets, or packets with the destination on a remote RPR network other than on the source network, are flooded on the source and the destination networks, and also on the path of the intermediate networks between the source and the destination networks. This decreases the available bandwidth for other traffic in those networks and is inefficient. As a result, we propose two solutions based on topology discovery, global topology discovery （GTD） and enhanced topology discovery （ETD）, that prevent the flooding of inter-ring packets. GTD enables the bridges to determine the next-hop bridge for each destination. ETD enables the source node to determine a default ringlet, so that packets reach the next-hop bridge without flooding the source network. The proposed solutions were analyzed and the overhead bandwidth and stabilization time were shown to be bounded. Simulations performed showed that the proposed solutions successfully avoid flooding and achieve optimal efficiency in the intermediate and destination networks, and in the source networks with one bridge.

Packet Queueing Delay in Resilient Packet Ring Network Nodes

The packet queueing delay is one of the most important performance measures of a data net-work and is also a significant factor to be considered in the scheduling buffer design for a network node. This paper presents a traffic queueing model for resilient packet ring (RPR) networks and a method for quantitatively analyzing queueing delays in RPR nodes. The method was used to calculate the average queueing delays of different priority traffic for different transit queue modes. The simulations show that, in the transmit direction, lower priority traffic is delayed more than higher priority traffic, and that Class-A traffic is delayed more in a single-queue ring than in a dual-queue ring. In the transit direction, the secondary tran-sit buffer in the dual-queue ring contributes more to the traffic delay than the primary transit buffer in the sin-gle-queue ring, which in turn causes more delay than the primary transit buffer in the dual-queue ring.