Ensemble-based Network Edge Processing Ioan Petri 1 , Ali Reza Zamani 2 , Daniel Balouek-Thomert 2 , Omer Rana 3 , Yacine Rezgui 1 , and Manish Parashar 2 1 School of Engineering, Cardiff University, UK 2 Rutgers Discovery Informatics Institute, Rutgers University, USA 3 School of Computer Science & Informatics, Cardiff University, UK contact author: petrii@cardiff.ac.uk Abstract—Estimating energy costs for an industrial pro- cess can be computationally intensive and time consuming, especially as it can involve data collection from different (distributed) monitoring sensors. Industrial processes have an implicit complexity involving the use of multiple appliances (devices/ sub-systems) attached to operation schedules, elec- trical capacity and optimisation setpoints which need to be determined for achieving operational cost objectives. Addressing the complexity associated with an industrial workflow (i.e. range and type of tasks) leads to increased requirements on the computing infrastructure. Such require- ments can include achieving execution performance targets per processing unit within a particular size of infrastructure i.e. processing & data storage nodes to complete a computational analysis task within a specific deadline. The use of ensemble- based edge processing is identified to meet these Quality of Service targets, whereby edge nodes can be used to distribute the computational load across a distributed infrastructure. Rather than relying on a single edge node, we propose the combined use of an ensemble of such nodes to overcome processing, data privacy/ security and reliability constraints. We propose an ensemble-based network processing model to facilitate distributed execution of energy simulations tasks within an industrial process. A scenario based on energy profiling within a fisheries plant is used to illustrate the use of an edge ensemble. The suggested approach is however general in scope and can be used in other similar application domains. Keywords-Edge computing, Energy Efficiency, Internet of Things, Industrial Processes I. I NTRODUCTION The integration of industrial workflows/ processes with edge devices provides numerous opportunities in automa- tion, optimisation, intelligent manufacturing and smart in- dustry, moving towards an on-demand service model. This leads to potentially new revenue models and facilitates industrial transformation. Integration of manufacturing in- dustries with Cloud-based analysis has been extensively investigated, considering the abundance of potential data generated from sensors integrated into industrial processes. Edge computing can represent a new solution to enhance and complement cloud-based data centers with support for real-time analysis of such data, enabling distributed exe- cution of tasks and enabling support for security/ privacy policies. Significant work has emerged in recent years on how edge devices can be used to extend the capability of cloud-based analysis for other latency sensitive applications, such as health care, security, smart cities, traffic control, transportation, production automation and many others all looking into how to implement the edge/fog models [1]. There are a number of industries and businesses which act as energy prosumers (both users and producers) often located close to large urban centres. To survive, they must be innova- tive in their business practices, controlling their cost base by the use of intelligent techniques for managing their energy consumption (a major factor in operational expenditure). For example, the fish processing industry is going through a paradigm shift from a unidirectional, demand driven industry with large centralised power generation to a market driven operational environment making use of smart grids, where supply and demand will be balanced with variable and intermittent renewable energies in a more localised manner. This will require intelligent systems to enable end users to satisfy demand within the peaks and troughs of the energy market. Therefore, supporting energy efficiency in industrial plants represents a prime objective for energy policy at regional, national and international levels. Studies have also indicated that although people and organisations are often aware of the benefits of using energy more efficiently, a variety of social, cultural, and economic factors often prevent them from doing so [2], [3]. Energy optimisation demonstrates real time use of sensor data, where a number of parameters need to be optimised based on a particular model representation. Based on such real-time readings from sensors it has become possible for site managers to take decisions in order to reduce energy consumption. As sensors can provide readings within an interval of 15-30 minutes, it is necessary for any simulation/ optimisation to also be carried out over a similar interval. The efficiency of the optimisation process depends on the capacity of the computing infrastructure used to execute demand analysis tasks. There has also been recent focus on the integration of sensor networks with decentralised dis- tributed systems based on the emergence of various network and IP-based technologies, where monitoring devices do not simply act as sensors, but feature computational, storage, and networking resources. Decentralized processing of data on Internet of Things (IoT) devices supported by cloud tech- nologies and virtualization has proved to be an efficacious method for reducing communication overheads and data