Functional Compression for Networked Applications in Practice

This work presents a practical implementation of Functional Compression (FC) in a real-life control application. We use a Rotary Inverted Pendulum (RIP) as the cyber-physical system and implement FC over the network. We compared the performance of FC with the controller implemented locally and the lossy compression of the sensor data. For the comparison, we used the average quadratic cost of the controller when using the different approaches. The results show that for a fixed compression level, FC can achieve a lower control cost than the lossy compression of the sensor data. We achieved a 69.4% in compression ratio and around 68% and 32% reduction in control costs compared to the traditional over-network control and networked lossy compression approaches, respectively.