Complex Behavior at Scale: An Experimental Study of Low-Power Wireless Sensor Networks

TLDR

A new class of networked systems is emerging that involves very large numbers of small, low‑power, wireless devices. Our experiments instrumented the network to separate protocol‑stack effects, measuring link‑layer packet reception, range, and asymmetry, MAC‑layer contention, collisions, and latency, and application‑layer tree structures from flooding. The large‑scale study of over 150 nodes at varying transmission powers shows that even simple flooding protocols can exhibit surprising complexity at scale, providing a foundation for broader algorithmic research.

Abstract

A new class of networked systems is emerging that involve very large numbers of small, low-power, wireless devices. We present findings from a large scale empirical study involving over 150 such nodes operated at various transmission power settings. The instrumentation in our experiments permits us to separate effects at the various layers of the protocol stack. At the link layer, we present statistics on packet reception, effective communication range and link asymmetry; at the MAC layer, we measure contention, collision and latency; and at the application layer, we analyze the structure of trees constructed using flooding. The study reveals that even a simple protocol, flooding, can exhibit surprising complexity at scale. The data and analysis lay a foundation for a much wider set of algorithmic studies in this space.