The Web of Things
Guest Editors' Introduction • Steven Gustafson and Amit Sheth • March 2014
Translated by Osvaldo Perez and Tiejun Huang
The Internet of Things (IoT) is an extension to the current Internet that enables connections and communication among physical objects and devices (see the September 2013 Computing Now theme for more on IoT and its role in ubiquitous sensing). Estimates suggest that there will be 50 billion devices and people connected and leveraging the vision and technology behind IoT by 2020. A related term that's currently somewhat in vogue is Internet of Everything (IOE), which recognizes the key role of people, or citizen sensing (such as through online social media), to complement the physical sensing implied by IoT.
The term Web of Things (WoT) goes beyond the focus on the Internet as the mode of exchanging data, instead bringing in all resources and interactions involving devices, data, and people on the Web. Correspondingly, it brings into focus a wide variety of challenges and opportunities while paving a way to a variety of exciting applications for individuals to industries.
WoT: Supercharging IoT and IoE
Sensors and devices collect data with a wide variety of types (such as temperature, light, sound, and video) and that are inherently diverse (data quality and validity can vary with different devices through time; data is often location- and time-dependent, and so on). WoT resources can be ubiquitous and are often constrained in terms of power, memory, processing, access, availability, attention, and communication capabilities. The resources' heterogeneity, ubiquity, and dynamic nature, coupled with the wide range of data, make discovering, accessing, processing, integrating, and interpreting the data on the Web a challenging task.
A rich cyber component of WoT includes Web-resident data, knowledge (in Wikipedia or Linked Open Data, for example), information exchanged over social media (such as sites in which patients share health-related information), and user-submitted physical world observations and measurements. Integrating physical, cyber, and social resources enables the development of physical-cyber-social (PCS) applications and services that can incorporate situation- and context-awareness into the decision-making mechanisms, and can create smart data (see http://wiki.knoesis.org/index.php/Smart_Data) out of big data by harnessing volume, variety, velocity and veracity to create actionable information. Examples of such applications range from personalized health, fitness, and well-being to energy and an increasingly wide variety of business and industrial activities.
Industrial Internet: Enabling Self-Directed and Autonomous Capabilities
WoT and IoT applications exploit Web-enabled and intermediated devices, data, human and social activities, and resources. Often under the umbrella term Industrial Internet, industries have begun to realize the value of applying IoT and WoT concepts to connect machines, industrial big data, and people. Recently, Moor Insights & Strategy proposed a WoT and IoT segmentation to capture the human-centric and interactive applications as the Human Internet of Things (HIoT), and the self-directed and autonomous applications as the Industrial Internet of Things (IIoT).
The Industrial Internet concept was proposed to translate the technologies behind the consumer Internet to the industrial space. The Industrial Internet can be thought of as an application of IoT connectivity and WoT services concepts, with additional focus on issues such as scalability, reliability, security, predictive capabilities, and virtualization of operations technology. Like the WoT, the Industrial Internet imagines that the "things" or resources connected are intelligent and can act with various degrees of autonomy.
By enabling vast amounts of sensors to be more easily connected, data to be more seamlessly collected, and analytics to be constantly looking to improve operations, the Industrial Internet aims to transform industries with applications such as smart manufacturing, intelligent hospital bed scheduling, and adaptive airline flight routes. These applications are estimated to bring tremendous value by reducing inefficiencies in various industrial segments. For example, in a 2012 whitepaper, GE estimates that a 1 percent efficiency improvement in how freight is moved across rail networks is valued at US$27 billion over a 15 year period. If we focus in on very related IoT application areas such as smart equipment servicing, there are about 120,000 diesel-electric locomotives worldwide. These assets require about 52 million labor hours to maintain, which is valued at $3 billion per year (as described in a 2013 report by GE). With 7 million people employed worldwide in this industry, there is a great opportunity to enable crowdsourcing, collaboration, and new operational technologies aimed at new productivity in field services and service and operational centers. This month's CN Industrial Perspective video explores such efforts.
WoT applications can potentially exploit all IoT data, citizen sensing observations, data in Web repositories, and more. That means they also face challenges and opportunities related to big data's volume, variety, velocity, veracity, and value properties. We begin this month's theme with "From Data to Actionable Knowledge: Big Data Challenges in the Web of Things" from IEEE Intelligent Systems' November/December 2013 special issue on WoT. It discusses the WoT Big Data issues.
Our next article, also from that special issue, is "Farming the Web of Things," in which Kerry Taylor and her colleagues discuss how sensor data is used to monitor a smart farm in New South Wales, Australia. In their Smart Farm application, a set of environmental monitoring sensors is deployed to provide (near) real-time information related to different situations on a farm. They also discuss the business challenges, barriers, and drivers in using WoT technologies and sensor data in a smart-farm environment.
Huansheng Ning and his colleagues' "Human-Attention Inspired Resource Allocation for Heterogeneous Sensors in the Web of Things" discusses adapting different human attention models—including sustained, selective, and divided attention—and describes a resource-allocation model that uses prior and posterior attention data to dynamically allocate resources for a WoT application.
In both human- and industrial-focused applications, the ability to leverage data from large networks of sensors is key. To enable robust and scalable systems, sensor-selection management must be adaptive and consider noisy and failure-prone sensors.
Charith Perera and his colleagues' "Context-Aware Sensor Search, Selection and Ranking Model for Internet of Things Middleware" proposes a context-driven way to select sensors using both semantic querying and quantitative reasoning. Given that WoT applications will require different degrees of sensor reliability and redundancy, this type of solution could be useful to build on existing infrastructure, as well as new, purpose-built sensor networks.
In "Semantic-Based Knowledge Dissemination and Extraction in Smart Environments," Michele Ruta and his colleagues apply a similar context-aware theme from the previous article to the entire WoT-composition process. By defining and leveraging a "ubiquitous knowledge base" in the middle of the WoT stack, or Semantic WoT (SWoT), more advanced reasoning and inference could address backward compatibility to information and resources, as well as improve the handling of sensor-data streams.
The IoT realization and deployment, whether human-focused or leveraging enriched web technology (as in the WoT paradigm) or self-directed and industry-focused (as in the Industrial Internet), creates a new demand and opportunity for platform technology. Fei Li and his colleagues' "Web-Scale Service Delivery for Smart Cities" describes a platform-as-a-service for delivery of smart city services. Such workflows and service platforms will need to be deployed to connect the smart data and devices with underlying infrastructure for computation and storage, while addressing issues such as quality of service for both human and machine users.
Crowdsourcing, Collaboration and the IoT Revolution
This month's Industry Perspective video comes from Joseph Salvo, an R&D leader from GE Global Research in NY, who describes the Industrial Internet perspective and how technologies will enable the modularization and composition of advanced, complex industrial systems. At the movement's core are the implicit technology adoptions of WoT and IoT technologies coupled with the societal change to collaborative and crowdsourced solutions.
S. Gustafson and A. Sheth, "Web of Things," Computing Now, vol. 7, no. 3, Mar. 2014, IEEE Computer Society [online]; http://www.computer.org/web/computingnow/archive/march2014.
Steven Gustafson is an R&D leader at GE Global Research where he leads a team of 15 scientists developing and applying advanced technology across industries. He has a PhD in computer science from the University of Nottingham. His technical interests include big data systems, semantics and knowledge capture, and machine learning and artificial intelligence. Gustafson has chaired several conferences and workshops in the field of evolutionary computing, is a technical editor-in-chief of the Memetic Computing Journal, serves on several journal editorial boards, and has been awarded several patents in the area of computational intelligence. Check his interview on Big Data and Industrial Internet or contact him at email@example.com.
Amit Sheth is the LexisNexis Ohio Eminent Scholar and executive director of the Ohio Center of Excellence in Knowledge-en9abled Computing (Kno.e.sis) at Wright State University. His technical interests include semantic/social/sensor webs, big and smart data sciences, physical cyber social computing, and computing for human experience. Sheth's R&D has resulted in many deployed applications and several commercial products, and he has (co-)founded three companies. Contact him at firstname.lastname@example.org or http://knoesis.org/amit.