Study of Throughput and Delay in Finite-Buffer Line Networks

TL;DR

This paper investigates how finite buffer sizes impact throughput and delay in line networks with erasure links, proposing an approximate method to analyze performance and optimize memory allocation.

Contribution

It introduces a simplified iterative scheme to approximate the steady-state distribution, enabling analysis of buffer effects on network performance.

Findings

Accurately predicts throughput and delay distribution variations

Provides a scalable approximation method for performance analysis

Classifies nodes for congestion-aware memory allocation

Abstract

In this work, we study the effects of finite buffers on the throughput and delay of line networks with erasure links. We identify the calculation of performance parameters such as throughput and delay to be equivalent to determining the stationary distribution of an irreducible Markov chain. We note that the number of states in the Markov chain grows exponentially in the size of the buffers with the exponent scaling linearly with the number of hops in a line network. We then propose a simplified iterative scheme to approximately identify the steady-state distribution of the chain by decoupling the chain to smaller chains. The approximate solution is then used to understand the effect of buffer sizes on throughput and distribution of packet delay. Further, we classify nodes based on congestion that yields an intelligent scheme for memory allocation using the proposed framework. Finally, by simulations we confirm that our framework yields an accurate prediction of the variation of the throughput and delay distribution.