Many-to-One Throughput Capacity of IEEE 802.11 Multi-hop Wireless Networks

TL;DR

This paper analyzes the many-to-one throughput capacity of IEEE 802.11 multi-hop networks, introducing canonical networks for easier analysis, and demonstrates how strategic routing and node placement can significantly improve throughput.

Contribution

It introduces canonical networks as a new analytical framework and shows how routing and node positioning can enhance throughput beyond typical random topology performance.

Findings

Upper bound of 3L/4 for canonical networks with multiple hops

Simulated throughput of approximately 0.690L to 0.740L in practical scenarios

Significant throughput improvements (>150%) by strategic route selection and node deactivation

Abstract

This paper investigates the many-to-one throughput capacity (and by symmetry, one-to-many throughput capacity) of IEEE 802.11 multi-hop networks. It has generally been assumed in prior studies that the many-to-one throughput capacity is upper-bounded by the link capacity L. Throughput capacity L is not achievable under 802.11. This paper introduces the notion of "canonical networks", which is a class of regularly-structured networks whose capacities can be analyzed more easily than unstructured networks. We show that the throughput capacity of canonical networks under 802.11 has an analytical upper bound of 3L/4 when the source nodes are two or more hops away from the sink; and simulated throughputs of 0.690L (0.740L) when the source nodes are many hops away. We conjecture that 3L/4 is also the upper bound for general networks. When all links have equal length, 2L/3 can be shown to be the upper bound for general networks. Our simulations show that 802.11 networks with random topologies operated with AODV routing can only achieve throughputs far below the upper bounds. Fortunately, by properly selecting routes near the gateway (or by properly positioning the relay nodes leading to the gateway) to fashion after the structure of canonical networks, the throughput can be improved significantly by more than 150%. Indeed, in a dense network, it is worthwhile to deactivate some of the relay nodes near the sink judiciously.