Efficient Monitoring of CPS and IoT Systems: A Deployment Guide for Empirical Evaluations
Paper in proceeding, 2024
As the digitalization of our world continues, the volume of data produced escalates rapidly. Communication often constitutes the most energy-intensive aspect of applications, making it a prime target for optimization. This paper studies the efficiency and precision of monitoring Cyber-Physical Systems (CPS) and Internet of Things (IoT) devices in real experimental deployment scenarios. We evaluate in this work the performance of six different monitoring algorithms concerning CPU utilization, memory usage, network traffic, energy consumption, and monitoring accuracy. Our experimental setup is a network consisting of 38 Raspberry Pis connected to a single host computer via Ethernet. Monitoring was conducted both on the devices themselves and on the host computer, which served as the metrics-receiving stack. We offer here a comprehensive deployment guide serving as a valuable resource for reproducing similar monitoring experiments on IoT devices. Additionally, our own empirical evaluation indicate notable differences among the tested algorithms, highlighting the importance of application constraints on the selection of a monitoring algorithm. Seeking those real-world experiments aim to add substantial value to existing surveys by offering deeper insights, practical findings and empirical evidence.
adaptive filtering
LMS
PLA
static filter
adaptive sampling
monitoring
SIP
CPS
IoT