An Approximate Inner Bound to the QoS Aware Throughput Region of a Tree Network under IEEE 802.15.4 CSMA/CA and Application to Wireless Sensor Network Design

TL;DR

This paper derives approximate conditions for tree networks using IEEE 802.15.4 CSMA/CA to meet QoS targets like delivery probability and delay, providing insights for wireless sensor network design.

Contribution

It introduces a set of sufficient conditions based on fixed point equations to evaluate QoS compliance in tree networks and identifies shortest path trees as optimal for maximum throughput under certain conditions.

Findings

Sufficient conditions relate network performance to arrival rates and topology.

Shortest path trees maximize throughput among bounded trees with equal arrival rates.

Conditions provide practical guidelines for QoS-aware wireless sensor network design.

Abstract

We consider a tree network spanning a set of source nodes that generate measurement packets, a set of additional relay nodes that only forward packets from the sources, and a data sink. We assume that the paths from the sources to the sink have bounded hop count. We assume that the nodes use the IEEE 802.15.4 CSMA/CA for medium access control, and that there are no hidden terminals. In this setting, starting with a set of simple fixed point equations, we derive sufficient conditions for the tree network to approximately satisfy certain given QoS targets such as end-to-end delivery probability and delay under a given rate of generation of measurement packets at the sources (arrival rates vector). The structures of our sufficient conditions provide insight on the dependence of the network performance on the arrival rate vector, and the topological properties of the network. Furthermore, for the special case of equal arrival rates, default backoff parameters, and for a range of values of target QoS, we show that among all path-length-bounded trees (spanning a given set of sources and BS) that meet the sufficient conditions, a shortest path tree achieves the maximum throughput.