Application-Aware Network Traffic Management in MEC-Integrated Industrial Environments

• Published in: Future internet 2023-01, Vol.15 (2), p.42
• Main Authors: Bellavista, Paolo, Fogli, Mattia, Giannelli, Carlo, Stefanelli, Cesare
• Format: Article
• Language: English
• Subjects: application-aware network traffic management; Cellular communication; Communications traffic; Confidentiality; Cryptography; Data encryption; Decision making; Edge computing; in-network processing; Industrial applications; Industrial equipment; industrial Internet of things; Infrastructure; Inspection; Internet; Internet of Things; Middleware; Mobile computing; multiaccess edge computing; Payloads; Software; Software-defined networking; Traffic flow; Traffic management
• ISSN: 1999-5903; 1999-5903
• DOI: 10.3390/fi15020042

The industrial Internet of things (IIoT) has radically modified industrial environments, not only enabling novel industrial applications but also significantly increasing the amount of generated network traffic. Nowadays, a major concern is to support network-intensive industrial applications while ensuring the prompt and reliable delivery of mission-critical traffic flows concurrently traversing the industrial network. To this end, we propose application-aware network traffic management. The goal is to satisfy the requirements of industrial applications through a form of traffic management, the decision making of which is also based on what is carried within packet payloads (application data) in an efficient and flexible way. Our proposed solution targets multi-access edge computing (MEC)-integrated industrial environments, where on-premises and off-premises edge computing resources are used in a coordinated way, as it is expected to be in future Internet scenarios. The technical pillars of our solution are edge-powered in-network processing (eINP) and software-defined networking (SDN). The concept of eINP differs from INP because the latter is directly performed on network devices (NDs), whereas the former is performed on edge nodes connected via high-speed links to NDs. The rationale of eINP is to provide the network with additional capabilities for packet payload inspection and processing through edge computing, either on-premises or in the MEC-enabled cellular network. The reported in-the-field experimental results show the proposal feasibility and its primary tradeoffs in terms of performance and confidentiality.