INTELLIGENT INDUSTRIAL FOG COMPUTING: AN OVERVIEW OF THE WI-NEXT ARCHITECTURE FOR INDUSTRIAL INTERNET OF THINGS.

Transcription

1 INTELLIGENT INDUSTRIAL FOG COMPUTING: AN OVERVIEW OF THE WI-NEXT ARCHITECTURE FOR INDUSTRIAL INTERNET OF THINGS

2 The industrial environment is an evolving enterprise comprising different types of sensors, machines, and instruments of varying vintage, each with its own demands for monitoring, control, and maintenance. Some of these machines are Internet-enabled (either through hardwired connectors or via Wi-Fi), but a broad range of older machines are not likely to be inherently connectable to a network. Yet the traditional approaches of manually overseeing and managing the nodes within an industrial environment can be radically improved by configuring the devices into an intelligent connected ecosystem. This ecosystem embodies the concept of an Industrial Internet of Things, or IIoT, which is a type of environment architecture that connects a broad community of equipment, devices, sensors, and machines within a mesh-connected network. However, enabling computational control embedded within the nodes of this mesh elevate the holistic system into a more In this paper we discuss the topology of an edge-centric Internet of Things (IoT) and an objective description and overview of the components and logical architecture for configuring and deploying an intelligent wireless mesh network within an industrial environment. intelligent environment. As opposed to a cloud-based framework in which all data is streamed to a single central server, the intelligence can be distributed across the mesh, thereby creating what we can call an Intelligent Industrial Fog. In an Intelligent Industrial Fog, software can be deployed at various points in the network to not only automate monitoring and control, but also to apply embedded intelligent agents that can adjust device behaviors in relation to ongoing performance variables, reduce running costs by reducing power consumption during off-cycles, or even detect imminent failures and notify technicians to perform preventative maintenance. In this paper we discuss the topology of an edge-centric Internet of Things (IoT) and an objective description and overview of the components and logical architecture for configuring and deploying an intelligent wireless mesh network within an industrial environment. We then consider the characteristics that transform this mesh network into an Intelligent Industrial Fog, a paradigm that implies distribution of processing, managed communications, and integration of embedded analytics at different levels of the mesh hierarchy to facilitate smart monitoring, control, and predictive/prescriptive oversight of the overall health of the industrial ecosystem. We describe the components of the WI-NEXT architecture and how they are leveraged to enable an Intelligent Industrial Fog. Finally, we consider the benefits to the business and suggest next steps in proving the value proposition within your organization. 2

3 THE INDUSTRIAL INTERNET OF THINGS (IIOT) IOT There are different conceptual architectures for an Internet of Things, with each configured in a way that balances the business s business applications data creation, communication, and computation needs. An edge-centric IoT is characterized by independent self-sufficient devices, distributed computation, and centralized communication generally limited to the purposes of coordination and network-wide analysis. The objective of an edge-centric IoT is three-fold: leverage the ability to push decision-making as close to the device as possible, enable continued monitoring and control when there is intermittent enterprise-wide connectivity, and limit and control exposure of data beyond the organization s firewall. An edge-centric IoT architecture is nicely suited to industrial environments in which devices and assets are remotely connected, where the devices and machines have computing capability, and the devices can be configured for computation and communication. A cohort of smart devices can be logically clustered around (and communicate with) an Edge node hub that is used for monitoring, coordinating, and controlling the end-nodes within its sphere of influence. Edge nodes are also connected to each other, providing a dynamically-configurable mesh network topology. In addition, the edge-nodes may communicate with a designated central server (used to accumulate and analyze streamed data from all the nodes in the mesh) that is either located within the corporate firewall or a cloud-based server accessible via the external Internet. 3

4 THE INTELLIGENT INDUSTRIAL FOG There are two key features of the edge-centric IoT: There must be an efficient publish/subscribe communications protocol designed for rapid data sharing, and There must be an environment for developing intelligent applicationware that that can be deployed at any point in the mesh. The ability to push the calculation of localized analytics all the way to the machines and the devices enables the creation of an intelligent industrial fog. As opposed to a cloud-based analysis framework, in which all the data is accumulated for centralized analytics, relevant decisioning is performed at the specific location of impact in the environment. An industrial environment poses some particular added challenges. Manufacturing facilities are prone to various forms of interference dust particles that can affect electronic devices, a variety of radio communications creating signal interference, as well as hard barriers through which signals cannot penetrate. 4

5 For a mesh network topology to be adapted to configure an intelligent industrial fog, it must demonstrate these kinds of characteristics: Ruggedness: The computational components must be reinforced industrial-grade devices designed for enhanced sustainability in an industrial environment. Integrate-able: The devices must be able to be integrated into a dynamically-configurable topology in which edge nodes are connected in a mesh while many end nodes are configured around a hub edge node. This topology allows information to be propagated in an efficient manner. Transparent: Depending on the business application, there may be requirements for some visibility into alignment of the entire mesh replicated at each device. This implies a capability for shareddistributed data in which each device may manage a partial or complete view of the connected ecosystem. Fault-Tolerant: To mitigate the risks of operating in the hostile industrial environment, the mesh must provide tolerance to failures and allow for dynamic mesh reconfiguration. Scalable: Machines or devices can be added to the mesh with performance scaling linearly in the number of units. Collaborative: Distributed processing, where the necessary computation takes place at the most relevant location (for example, device monitoring happens at the end node, as opposed to communicating the data stream to a centralized server). Actionable: Actions to be taken at a specific device as a result of continuous monitoring can be triggered by the machine or device controller at that location in the mesh. Integrated Analytics: The applicationware deployed at the various points in the mesh can embed intelligent predictive and prescriptive models coupled with event stream processing so empower preemptive automated decision-making to improve overall operations. 5

6 THE WI-NEXT ARCHITECTURE Aside from the obvious environmental challenges, most industrial environments are subject to an additional obstacle: the lack of node intelligence and connectivity. Many heritage machines are bound to be missing an intelligent controller and have no means for connectivity. That means that in order to institute an intelligent industrial fog, one must be able to: Retrofit the existing machines with computing and communications capabilities. Integrate edge nodes that can coordinate the devices and machines within their sphere of influence. Employ a robust and scalable means of messaging that provides reliability and security. A connectivity solution for the industrial IoT space consisting of Edge-Nodes (hardware) with embedded software to build a wireless network, and End-Nodes (hardware) with embedded software to interface to the machines and surrounding sensors. Those are the drivers for the essential components of the Wi-NEXT architecture 6

7 COMMUNICATIONS AND MESSAGING APPLICATION LUA AGENTS MESSAGE MQTT TRANSPORT AGILE MESH AD-HOC AP/STA ETHERNET The dynamic yet hostile nature of the industrial environment means that an Intelligent Industrial Fog that leverages an Industrial Internet of Things must provide a more robust and trustworthy communications scheme to counter vibration, dust, obstacles and radio interference. In addition, there must be integrated mechanisms for ensuring reliability of the mesh connectivity by tolerating the need for reconfiguring the network when connectivity obstacles impact the environment. Wi-NEXT has attempted to address this need using their proprietary WiseMESH Network Operating System (NOS). First, WiseMESH can optimize connectivity despite the issues associated with an industrial environment by providing dynamic route reconfiguration, an ability to reevaluate the performance of the mesh and potentially pick nodes based on performance and changing environmental parameters. Second, WiseMESH uses open industry standards for communications (such as Wi-Fi, Ethernet, IP, Zigbee, and Bluetooth). Third, WiseMESH layers messaging using the MQTT (Message Queue Telemetry Transport) messaging protocol, which is a light-weight publish/subscribe messaging protocol employing TCP/IP to support remote connectivity for devices with limited memory/storage footprints and network bandwidth. The benefit of using MQTT for publish/subscribe is twofold it provides a lightweight means of broadcasting data to a wide set of listeners, yet supports security features for data transport. 7

8 END NODE To address the need to retrofit existing machinery, the Wi-NEXT architecture includes an end node that comes in two versions. Both versions are low-power Wi-Fi devices, have an embedded microcontroller, storage, general purpose I/O (GPIOs), and can be interfaced with external sensors. Both versions use the MQTT protocol for communications. The Power version, which is suited to retrofitting existing machinery, is connected via the power supply. This version monitors power consumption and communicates that data across the mesh network. In addition, the End Node Power can connect and collect data from network sensors as well as connect to and trigger connected actuators. An embedded (End Node Embedded) version has a small form factor and is suitable for direct incorporation within newly-designed devices and developing smart device applications. The embedded version is also available for partners developing OEM applications. 8

9 EDGE NODE The family of Wi-NEXT s Edge Nodes are more powerful devices that can play multiple roles in the intelligent mesh network. From a connectivity standpoint, the Edge Nodes act as network gateways or hubs for connecting end nodes, access points for Wi-Fi devices, or as a network repeater. The Edge Nodes are used for monitoring and controlling operations of the devices that are within its sphere of influence, sharing information with other edge nodes in the mesh as well as with any designated centralized point of coordination. Edge Nodes are designed with internal CPU, memory, and storage so that applications for analysis of connected devices and machines can be deployed. Some examples of features include: CPU/Computing: Atheros AR 7242 CPU board and are configured with DDR memory, flash storage, and a slot for an SDHC memory card. Wired connectivity: 2 RJ 45 Ethernet 10/100/1000 with an integrated Gigabit switch. Wireless connectivity: internal or external antennas, using IEEE abgn for high throughput. Roaming: Dynamic characteristics of roaming between access points. Network Profiles: Standard access point, router, repeater, or hotspot functionality). Environmental Robustness: Is configured to operate in varied temperature and humidity conditions, as well as reconfigurable for rerouting in the presence of interference, dust, or other obstacles. Security: Supports a variety of network and https security protocols and encryption. Communications: As with the End Nodes, the Edge Nodes communicate using the MQTT protocol. 9

10 COMPUTING The key to transforming an IIoT into an Intelligent Industrial Fog is the ability to embed analytic processing at all locations in the mesh network. In essence, the mesh network composes a loosely-coupled distributed computing environment with a virtual means of sharing memory while distributing storage. In the Wi-NEXT architecture, code for monitoring integrated and external sensors as well as controlling devices can be deployed at the End Nodes. The richer computing environment of the Edge Node allows for operational monitoring of a cohort of connected End Nodes, as well as more complex computation for analysis of streamed data for event triggering, predictive modeling of critical events, and notifications. In essence, it is at the Edge WiseMESH uses an advanced Fog Computing architecture to resolve the typical connectivity and processing hurdles by locating computing power, storage, networking and analytics at the edge. Node where opportunities for preemptive maintenance can be analyzed. Continuous monitoring of a community of devices will enable analysis to predict machine events or failure before they actually occur, thereby allowing preemptive maintenance to be performed to maintain continuity of operations. In the Wi-NEXT environment, the nodes are programmed using a light-weight scripting language called Lua. According to the Lua website ( Lua combines simple procedural syntax with powerful data description constructs based on associative arrays and extensible semantics. Lua is dynamically typed, runs by interpreting bytecode for a register-based virtual machine, and has automatic memory management with incremental garbage collection, making it ideal for configuration, scripting, and rapid prototyping. Lua is a reasonable choice as a development language for embedded applications. Its execution engine has a small footprint, uses a small amount of memory, yet is fast and easy to work with, as it will work on any platform with a standard C compiler. 10

11 COORDINATION WITH DATA-CENTRIC/CLOUD-BASED APPLICATIONS All coordination across the mesh network can be facilitated through collaborative computing distributed within the local mesh. In other words, the applications for monitoring, control, and predictive maintenance for all of the devices and machines in the environment can be deployed using algorithms whose computation is distributed among the Edge Nodes. However, the mesh topology can be adapted to other paradigms for coordination. One alternative to distributed monitoring and control is to use a local central server associated with a cohort of edge-nodes within a single environment. In this approach, the data from all of End Nodes and Edge Nodes can be accumulated at the local central server, which can provide the platform for control and analytical applications. Our Edgeware Technology enable our partner to put easily their decision-making and analysis platform next to the controlled devices or processes at the edge. In turn, these local central servers may also be mesh-connected. As an example, consider a company with multiple factory facilities. Each facility may manage its internal Industrial IoT, and the central servers for each facility communicate and share data for cross-facility analysis. Alternatively, selectively streamed data from across the mesh network can be accumulated and forwarded to partner cloud-based applications, in which the data can be analyzed and inform decisions impacting more than one network of industrial ecosystems. An example would be aggregating data from multiple facilities from multiple companies that use the same types of machines and devices. By sharing event and device failure data, a comprehensive analysis can help adjust model parameters that can ultimately pushed back down to the Edge nodes and End nodes to be incorporated into the predictive models. 11

12 SUMMARY In summary, the key components of the Wi-NEXT architecture enable the creation of an Intelligent Industrial Fog using an IIoT. Vintage machines can be retrofitted with the Power version of the End Node, while newer devices and equipment with digital interfaces can be upgraded with the embedded version. Edge nodes can be deployed for monitoring, communication, and oversight over a collection of devices, as well as provide the platform for embedded analytical models for control and predictive maintenance. Implementers can add additional intelligence at any layer of the mesh network, as well as connect to central/on-premise or cloud-based hosts for global oversight and analysis. Understanding that there are clear benefits of the Intelligent Industrial Fog should motivate some follow-up steps to determine the best opportunities for design and implementation: 1) Organizational Preparedness: Assess your existing industrial environment to determine its IIoT-readiness. Using an inventory of the devices and machines within your industrial environment, differentiate between those devices that are inherently connectable to embeddable End nodes and those vintage machines that could be retrofitted using the Power End nodes. 2) Assess the Control Environment: Determine how automation and control is already implemented in the environment, and identify specific types of devices and machines that can benefit from increased monitoring and control, what types of automation already exists in the environment, and where new opportunities for instrumentation can increase overall efficiency. 3) Devise the Value Proposition: Devise a cost/benefit model to calculate the potential economic upside of implementing the IIoT within the environment regarding predictability in asset monitoring and maintenance, reducing energy consumption, and overall improvements in efficiency and reliability. 4) Proof of Concept: Develop a project plan for a proof-of-concept to test out the technology, review how analytical models can positively impact a segregated collection of devices and machines, specify the success criteria, assess the degree to which those criteria were met, and refine the value proposition for broader deployment. 12

13 ABOUT THE AUTHOR David Loshin, president of Knowledge Integrity, Inc, ( is a recognized thought leader and expert consultant in the areas of analytics, big data, data governance, data quality, master data management, and business intelligence. Along with consulting on numerous data management projects over the past 15 years, David is also a prolific author regarding business intelligence best practices, as the author of numerous books and papers on data management, including the recently published Big Data Analytics: From Strategic Planning to Enterprise Integration with Tools, Techniques, NoSQL, and Graph, the second edition of Business Intelligence The Savvy Manager s Guide, as well as other books and articles on data quality, master data management, big data, and data governance. David is a frequent invited speaker at conferences, web seminars, and sponsored web sites and channels including and share additional content at his notes and articles at David can be reached at loshin@knowledge-integrity.com, or at (301)

14 CONTACTS: US OFFICE 530 Lytton Avenue, Palo Alto, California USA Ph. +1 (650) ITALY OFFICE Piazzale Biancamano 8, Milano - Italy Ph NOTE 2015 Wi-NEXT. All rights reserved. This document may be reproduced in whole but not in part. The information contained in this document is subject to change without notice and is made available in good faith without liability on the part of Wi-NEXT. All trademarks acknowledged. Company with Quality Management System Certificate Certificate N