Non-Opportunistic Data Transfer for IoT and Cyber-Physical Systems with Mostly Sleeping Nodes

Abstract

Sensor networks are frequently used to monitor our environment. From monitoring the habitat of seabirds [1], to the structural integrity of bridges [2]. They can also be used to monitor the arctic tundra to help us monitor climate change.

The arctic tundra does however place additional requirements on a monitoring system. Low access to energy sources, human intervention, and networks to transfer the results back, combined with a high likelihood of being destroyed by the environment makes it difficult to successfully retrieve any measurements. The nodes should therefore replicate any measurements among themselves while minimizing the energy consumption.

In this thesis, we describe four approaches to schedule connections to share data between a neighborhood of nodes. We also present the implementation of a simulation to evaluate the approaches based on energy usage, broadcast-latency and broadcast-throughput.

We conclude that scheduling connections in a ring-like or cluster structure has in general the lowest energy usage at the cost of latency and throughput. However, more work should be done to get a more accurate estimation of the energy usage of the systems.