Multipath forwarding strategies in Information Centric Networks with AIMD congestion control

TL;DR

This paper presents an analytical model to evaluate multipath forwarding strategies in ICNs with AIMD congestion control, analyzing existing strategies and proposing a new one, validated through PlanetLab experiments.

Contribution

It introduces a general analytical model for ICN multipath forwarding strategies with AIMD congestion control and evaluates both existing and novel strategies.

Findings

The model accurately predicts performance of multipath strategies.

Existing strategies have varying efficiency depending on network conditions.

The proposed strategy outperforms existing ones in realistic tests.

Abstract

A content can be replicated in more than one node, in Information Centric Networks (ICNs). Thus, more than one path can be followed to reach the same content, and it is necessary to decide the interface(s) to be selected in every network node to forward content requests towards such multiple content containers. A multipath forwarding strategy defines how to perform this choice. In this paper we propose a general analytical model to evaluate the effect of multipath forwarding strategies on the performance of an ICN content delivery, whose congestion control follows a receiver driven, loss-based AIMD scheme. We use the model to understand the behavior of ICN multipath forwarding strategies proposed in the literature so far, and to devise and evaluate a novel strategy. The considered multipath forwarding strategies are also evaluated in a realistic network setting, by using the PlanetLab testbed.