Achieving Low Latency Communications in Smart Industrial Networks with Programmable Data Planes

Area of Science:

• Computer Science
• Networking
• Industrial Automation

Background:

• Industrial networks are integrating Internet of Things (IoT) for enhanced efficiency and sustainability.
• A key challenge is meeting stringent response time requirements for time-critical industrial applications.
• In-network computing and edge computing offer solutions for improved communication quality.

Purpose of the Study:

• To present a novel solution for reducing response times in industrial IoT (IIoT) communications.
• To leverage in-network computing and data plane programming for faster data processing.
• To demonstrate the effectiveness of P4 language in achieving these goals.

Main Methods:

• Implemented an IIoT communication solution using data plane programming with P4 language.
• Utilized in-network computing to process Message Queuing Telemetry Transport (MQTT) packets directly in programmable network devices.
• Tested the solution on an experimental facility with a Netronome SmartNIC.

Main Results:

• Achieved a 74% reduction in response times during data processing.
• The delay introduced by P4 network processing was found to be insignificant.
• Successfully processed sensor data and generated alarms in the data plane based on predefined thresholds.

Conclusions:

• Data plane programming with P4 enables effective in-network computing for IIoT.
• The proposed solution significantly reduces response times in industrial networks.
• This approach is viable for enhancing the performance of time-critical IIoT applications.