Network Functional Compression for Control Applications

TL;DR

This paper demonstrates the implementation of network Functional Compression to significantly reduce data traffic in control systems by transmitting compressed sensor data to in-network controllers, achieving over 50% compression.

Contribution

It introduces a practical implementation of network Functional Compression for control applications, showing substantial data reduction and efficiency improvements.

Findings

Achieved over 50% data compression in control scenarios

Distributed cascade FC further reduces data and computational costs

Enables goal-oriented communication for cyber physical systems

Abstract

The trend of future communication systems is to aim for the steering and control of cyber physical systems. These systems can quickly become congested in environments like those presented in Industry 4.0. In these scenarios, a plethora of sensor data is transmitted wirelessly to multiple in network controllers that compute the control functions of the cyber physical systems. In this paper, we show an implementation of network Functional Compression (FC) as a proof of concept to drastically reduce the data traffic in these scenarios. FC is a form of goal-oriented communication scheme in which the objective of the sender receiver pair is to transmit the minimum amount of information to compute a function at the receiver end. In our scenario, the senders transmit an encoded and compressed version of the sensor data to a destination, an in-network controller interested in computing as its target function, a PID controller. We show that it is possible to achieve compression rates of over 50% in some cases by employing FC. We also show that using FC in a distributed cascade fashion can achieve more significant compression rates while reducing computational costs.