Edge-Powered In-Network Processing for Content-Based Message Management in Software-Defined Industrial Networks

Traditional industrial networks were characterized by flat topologies, where industrial equipment exchanged a limited number of messages. In sharp contrast, modern manufacturing plants are evolving towards articulated environments generating an ever-increasing amount of network traffic. Such emerging environments prevent the adoption of traditional solutions based on end-to-end dispatching of few messages in reliable networks with bandwidth availability much greater than needed. This leads to the need to adopt proper message management strategies as close as possible to industrial equipment to avoid overwhelming the industrial network with non-mission-critical traffic at the expense of mission-critical one. This paper originally proposes edge-powered in-network processing to i) transparently manage messages sent by industrial equipment, ii) support a broad spectrum of message management strategies, ranging from efficient header-based solutions to expressive content-based ones, and iii) fulfill the application-dependent requirements demanded by industrial environments nowadays. Achieved performance results based on a proof-of-concept prototype demonstrate that the proposed solution efficiently provides content-based message management at the edge, even considering edge nodes with limited hardware capabilities.