Aggregate Processes as Distributed Adaptive Services for the Industrial Internet of Things

TL;DR

This paper proposes using Aggregate Programming to simplify the development of distributed, intelligent services at the edge of IIoT systems, demonstrating its feasibility through simulations and physical experiments.

Contribution

It introduces the aggregate process construct within Aggregate Programming as a novel approach for building distributed IIoT services at the edge.

Findings

Feasibility demonstrated through simulations with FCPP

Physical experiments show practical applicability

Approach reduces complexity of distributed service development

Abstract

The Industrial Internet of Things (IIoT) promises to bring many benefits, including increased productivity, reduced costs, and increased safety to new generation manufacturing plants. The main ingredients of IIoT are the connected, communicating devices directly located in the workshop floor (far edge devices), as well as edge gateways that connect such devices to the Internet and, in particular, to cloud servers. The field of Edge Computing advocates that keeping computations as close as possible to the sources of data can be an effective means of reducing latency, preserving privacy, and improve the overall efficiency of the system, although building systems where (far) edge and cloud nodes cooperate is quite challenging. In the present work we propose the adoption of the Aggregate Programming (AP) paradigm (and, in particular, the "aggregate process" construct) as a way to simplify building distributed, intelligent services at the far edge of an IIoT architecture. We demonstrate the feasibility and efficacy of the approach with simulated experiments on FCPP (a C++ library for AP), and with some basic experiments on physical IIoT boards running an ad-hoc porting of FCPP.