Adaptive Network Traffic Reduction on the Fly with Programmable Data Planes

• Published in: IEEE access Vol. 11; p. 1
• Main Authors: Gyorgyi, Csaba, Voros, Peter, Kecskemeti, Karoly, Szabo, Geza, Laki, Sandor
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.01.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: 5G mobile communication; Cloud computing; Communications traffic; Control systems; Controllers; Data exchange; Edge computing; Electronic devices; Industrial communication; Internet of Things; Packet transmission; Power consumption; Protocols; Radio links; Real time; Real-time systems; SCADA systems; Software-defined networking; Switches
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2023.3255985

Industrial networks rely on standard real-time communication protocols, such as ProfiNet RT. These real-time protocols use cyclic data exchange between IO devices and controllers. Each IO device reports its internal state to the controller at a predefined frequency, even if the state of the device is unchanged. These reports are essential to accurately monitor the health of the devices, but network resources are limited and it is not advisable to overload the network with unnecessary packets. The traffic generated by a single device is insignificant, but in an industrial site with hundreds of such devices, the number of packets to be transmitted adds up. As cloud-based industrial controllers (e.g., cloud-based soft-PLCs) become more prevalent, all generated IO device traffic must be forwarded over the access link to edge computing or private/public cloud infrastructure. Wireless (e.g. 5G radio) transmission of many small packets leads to spectrum efficiency issues and high power consumption. In this paper, we propose an in-network solution to significantly reduce industrial network traffic by cooperating with two P4 programmable network elements deployed on both sides of an access link. Excess traffic is filtered out and new data content is cached at both ends while detecting both link and device failures in real-time. The adaptive mechanism introduced allows the system to automatically optimize its efficiency and performance by dynamically enabling and disabling traffic filtering/caching. Our measurements show that the method can significantly reduce the wireless link load while being seamlessly deployable in existing industrial environments without modifying the protocol, IO devices, and controllers.