Source-linked AI summary
Context-aware Computing in the Internet of Things: A Survey on Internet of Things From Industrial Market Perspective
Charith Perera, Chi Harold Liu Member, Srimal Jayawardena, Min Chen
TL;DR
Industrial IoT solutions have expanded beyond the academic focus of existing surveys, creating a need to understand how context-aware technologies are used in marketplace products. This paper reviews and categorizes such solutions and evaluates their context-aware strategies through a theory-based framework. The survey identifies limited primary-context sensing, varied data processing, predominantly screen-based interaction, continued manual actuation, and interoperability and planning challenges, while outlining research and innovation opportunities.
Problem
Existing IoT context-aware surveys focus on academic research, leaving industrial IoT solutions and their marketplace use of context-aware technologies insufficiently surveyed.
Method
The paper reviews marketplace IoT products, categorizes them into five segments, and evaluates their context-aware features using a framework based on established context-aware computing theories.
Results
The survey finds limited primary-context sensing, substantial variation in data processing, mostly screen-based presentation, frequent manual actuation after notifications, and major interoperability concerns.
Takeaways & Limitations
The findings identify opportunities for sensing-as-a-service, hands-free interaction, domain-independent analytics, interoperable platforms, deployment-planning tools, and continued integration with legacy devices.
Abstract
from arXiv · showhide
The Internet of Things (IoT) is a dynamic global information network consisting of Internet-connected objects, such as RFIDs, sensors, and actuators, as well as other instruments and smart appliances that are becoming an integral component of the Internet. Over the last few years, we have seen a plethora of IoT solutions making their way into the industry marketplace. Context-aware communication and computing has played a critical role throughout the last few years of ubiquitous computing and is expected to play a significant role in the IoT paradigm as well. In this article, we examine a variety of popular and innovative IoT solutions in terms of context-aware technology perspectives. More importantly, we evaluate these IoT solutions using a framework that we built around well-known context-aware computing theories. This survey is intended to serve as a guideline and a conceptual framework for contextaware product development and research in the IoT paradigm. It also provides a systematic exploration of existing IoT products in the marketplace and highlights a number of potentially significant research directions and trends.
I. INTRODUCTION
The paper frames IoT as a rapidly expanding network of connected objects whose context-aware capabilities are increasingly important, while noting that existing surveys largely emphasize academic research. It surveys industrial marketplace solutions and evaluates them through a context-aware computing framework to identify trends and research opportunities.
- IoT and context-aware computing: IoT connects everyday objects, including sensors, actuators, RFID systems, appliances, and other devices, into a global information network.The paper describes IoT as extending Internet connectivity beyond computers and mobile devices to ordinary objects.
- IoT and context-aware computing: Context-aware computing uses information about entities and situations to provide users with information or services relevant to their tasks.The paper defines context as information characterizing an entity’s situation and context-awareness as task-dependent relevance.
- Research gap and contribution: Existing context-aware IoT surveys focus on academic research, leaving industrial IoT solutions without a dedicated survey.The paper therefore addresses products proposed, designed, developed, and marketed by organizations ranging from start-ups to large corporations.
- Research gap and contribution: The survey reviews industrial IoT products using a framework grounded in established context-aware computing theories.It gathers information from company websites, demonstration videos, technical specifications, and consumer reviews, then evaluates selected solutions from a context-aware perspective.
- IoT marketplace: IoT marketplace growth spans connected devices, sensors, RFID, smart cities, and other application domains.The paper reports projections of 50 to 100 billion Internet-connected devices by 2020 and a smart city market expected to exceed $1 trillion by 2016.
- Evolution of context-aware computing: Context-aware functionality evolved from application assistance and contextual menus toward location-, social-, mobile-, and object-based interactions.The Internet’s development is presented in five phases: computer networking, the World Wide Web, mobile Internet, social networking, and IoT.
IV. THEORETICAL FOUNDATION AND EVALUATION FRAMEWORK
The paper builds an IoT evaluation framework on established context-aware computing theories, focusing on tagging, selection and presentation, and execution. It applies this framework to common IoT data-flow stages and product classifications.
- Context-aware Computing Theories: The framework adopts Abowd et al.’s definition of context because it can identify context from data generally.
- Context-aware Computing Theories: Context-aware applications support presentation, execution, and tagging as three features applicable to IoT systems.
- Context-aware Computing Theories: Presentation selects information or services using context such as location and time, while execution automatically triggers context-dependent actions.
- Context-aware Computing Theories: Tagging collects, fuses, interprets, and associates context with sensor data to support understanding of situations.
- Evaluation Framework: The evaluation framework reviews products through context-aware tagging, context selection and presentation, and context-aware execution across adaptable IoT data-flow stages.
- Evaluation Framework: Products are classified into five categories, while the remaining evaluation columns cover tagging, context selection and presentation, and execution.
1) Context-aware Tagging Section:
The reviewed IoT products collect primary context, derive secondary context, and present or act on it through diverse channels, triggers, and processing locations. Examples show how sensing capabilities can be combined with software, cloud, and intermediary devices.
- Context-aware Tagging: SenseAware collects location, temperature, light, relative humidity, and biometric pressure to improve real-time shipment visibility.
- Context-aware Tagging: Secondary context is computed from primary context through sensor fusion or retrieval operations such as web-service calls.
- Context-aware Tagging: Mimo transfers primary sensor data from the lightweight turtle device to a rechargeable lilypad for communication and processing.
- Context Selection and Presentation: Products use notifications, visual presentation, and voice, gesture, or touch interactions, with some devices offloading presentation to smartphones and tablets.
- Context Selection and Presentation: IoT products present context through visual interfaces, including Fitbit’s mobile and web displays of activity and sleep information.
- Context execution: IoT solutions process data in sensors, local devices, or the cloud, using real-time or archival processing depending on application requirements.
V. REVIEW OF IOT SOLUTIONS
The paper evaluates a variety of marketplace IoT solutions using the previously introduced context-aware framework. Table I summarizes the framework, while Table II reports product-review results.
- Review of IoT Solutions: The review evaluates a variety of IoT marketplace solutions using the earlier context-aware evaluation framework.
- Review of IoT Solutions: Table I summarizes the evaluation framework used for the product review.
- Review of IoT Solutions: Table II presents the results of the IoT product review.
VI. LESSONS LEARNED, OPPORTUNITIES AND CHALLENGES
The paper’s lessons-learned section distills major findings from its review of IoT products. These lessons are presented as a basis for identifying opportunities and challenges.
- Lessons Learned, Opportunities and Challenges: This section presents major lessons learned from reviewing IoT products.
- Lessons Learned, Opportunities and Challenges: The lessons are derived from the paper’s IoT product review.
- Lessons Learned, Opportunities and Challenges: The section frames the review lessons within opportunities and challenges for IoT products.
A. Trends and Opportunities
The survey finds that IoT products collect limited primary context but process it in diverse ways, while interaction, automation, personalization, and economic payback remain important opportunities.
- IoT solutions collect mostly limited types of primary context, yet use roughly 30-40 sensor types across applications.Different processing methods derive different insights from the same data, supporting sensing as a service.
- Most products present summarized data through traditional screens, while voice, object-based, gesture, and touch interactions remain less common.Wearables commonly use touch, and smart watches or glasses may reduce smartphone-related distraction.
- Most IoT products stop after notifying consumers, leaving users to perform actuation manually.The survey associates this pattern with a lack of M2M communication standards and identifies a need for accessible analytical frameworks.
- Many products treat families as groups rather than individuals, limiting their ability to accommodate separate preferences such as desired temperature.The survey identifies individual recognition and preference handling as a critical future marketplace requirement.
- Nest’s Auto-Schedule feature is described as helping users save up to 20% on heating and cooling bills, supporting product-cost recovery.The example links adoption to recovering the purchase cost within a reasonable period.
B. Product Prototyping
IoT prototyping is supported by inexpensive modular platforms and small computers, while surveyed products span diverse sensing and application configurations.
- DIY prototyping platforms let users test IoT ideas quickly, cheaply, and modularly with limited budget, resources, and time.Examples include Arduino, .NET Gadgeteer, and LittleBits.
- Raspberry Pi is a credit card-sized single-board computer increasingly used for IoT product prototype development.The platform was originally developed to promote basic computer science education.
- Surveyed products cover sensors and outputs including environmental conditions, health measures, activity, location, energy, and air quality.The registry lists examples such as temperature, GPS, heart rate, accelerometry, pollution compounds, and energy usage.
- The surveyed marketplace examples include smart river management, flood detection, waste management, health monitoring, fitness tracking, smart homes, and weather stations.These examples pair domain-specific measurements with outputs such as prediction, reports, maps, navigation, and localization.
- Modular prototyping tools combine primary sensor context into secondary context, but secondary-context discovery often lacks standard module definitions and composability.The survey calls for standard context-discovery modules that can combine to derive more advanced context information.
C. Interoperability on Product and Services
Interoperability is presented as essential for customizable IoT ecosystems, but current approaches face scalability, market-domination, security, and privacy concerns.
- Interoperability lets consumers combine products and services or move between providers according to cost, functionality, appearance, and customer service.The survey also links interoperability with lowering entry barriers for smaller providers.
- IoT interoperability is mainly pursued through partnerships, open or closed standards, and adapter or mediator services.The paper describes these as the three principal marketplace approaches.
- Partnerships may not scale widely, while proprietary standards from large corporations may encourage market domination and hinder smaller companies’ innovation.The survey cites Apple and Google as examples of corporations developing standards and certifications.
- AllJoyn enables devices to advertise and share capabilities, such as a motion sensor informing a light bulb that a room is unoccupied.The paper notes that security and privacy require strengthening in this interoperability framework.
- IFTTT uses channels, triggers, actions, and personal recipes to connect otherwise different IoT products and services.A cited recipe sends a Twitter message to a family member when the user reaches home.
D. Resources and Energy Management
The section identifies gaps in planning, deployment, resource management, and context-sensitive configuration for IoT environments. It proposes tools that simulate deployments, optimize processing and sensing, and help users select compatible products.
- Planning and Deployment: IoT lacks marketplace solutions for planning or deploying smart environments, despite the importance of energy analysis in industrial and consumer settings.The gap includes smart home, office, and industrial deployment scenarios.
- Planning and Deployment: Current tools do not determine optimal sensor locations or the number of products needed to achieve coverage in a specified area.Examples include placing micro-climate sensors and estimating motion-sensor requirements.
- Planning and Deployment: Large-scale deployment tools could predict energy sources, battery sizes, and sensor-node quantities from contextual information.Such tools would support agricultural scientists and other nontechnical users during deployment planning.
- Context-Aware Configuration: Context-sensitive sensor configuration supports timely, location-aware data collection, while intelligent reconfiguration can save energy by eliminating ineffective sensing and communication.The purpose is to fuse and reason over sensor data to understand environments more effectively.
- Tool Requirements: A modular planning tool should simulate scenarios, recommend sensor parameters, visualize products, and combine compatible products using budget, preferences, and location.The proposed product library would let manufacturers add solutions and consumers select them visually.
- Resource Management: Resource-management tools should evaluate software components against computational network architectures to select processing locations using preferences, availability, context, and communication conditions.The evaluation concerns where reasoning should occur in the IoT network.
E. Privacy and Data Analytic
Privacy is a shared requirement across IoT device manufacturers, cloud and platform providers, and application developers. The section outlines responsibilities involving device controls, data portability, standards, malware prevention, and informed consent.
- Privacy Responsibilities: IoT marketplace participants must treat privacy as a serious requirement and challenge across devices, cloud services, platforms, and applications.The section groups responsibilities by the three major marketplace parties.
- Device Manufacturers: Device manufacturers should implement secure storage, data deletion, and access controls at firmware level while explaining collection and processing practices.Consumers should be informed about what data devices collect and when it leaves the device.
- Cloud and Platform Providers: Cloud and platform providers should use common standards so users can choose providers and seamlessly delete or move their data between services.Interoperability and portability depend on following common standards.
- Third-Party Developers: Application developers should certify apps against malware and provide clear information needed for explicit user consent.Required disclosures include app tasks, needed data, sensors, and aggregation or analysis practices.
F. Central Hubs
Central hubs consolidate communication, storage, and reasoning for IoT components that are too constrained to support substantial computation. However, proprietary hub designs reduce interoperability, motivating common platforms, hardware adaptors, and intermediary protocol bridges.
- Central Hubs: IoT sensors and actuators are kept small for deployment, limiting their ability to support significant computation and motivating the use of central hubs.A typical solution may also include processing and communication devices.
- Central Hubs: Central hubs are larger, support multiple wireless protocols, store data, and perform processing or reasoning, often with one hub covering a large area.Examples of hub functions include triggering IF-THEN rules.
- Interoperability: Designers’ tendency to build proprietary centralized hubs significantly reduces interoperability because hubs use custom firmware and hardware-specific software stacks.These stacks make it harder for other IoT solutions to use the hubs.
- Interoperability: A common software platform and standards could support adoption, with hardware adaptors enabling two-way interoperability between products and different hubs.The proposed model is compared with mobile application platforms, but IoT additionally requires hardware verification.
- Protocol Bridging: Intermediation nodes can bridge short-range protocols such as Bluetooth and ZigBee and connect sensors to central hubs through longer-range communication.These nodes may be distributed throughout a location.
G. Legacy Devices
IoT products can add smart capabilities to legacy devices without replacing them. Nest learns user behavior for thermostat control, while Leeo listens to existing alarms and forwards notifications when it detects alarm sounds.
- Legacy Devices: Nest enriches a traditional thermostat by learning users’ behavior and preferences and controlling temperature more efficiently and proactively.It can be installed by replacing an existing nonsmart thermostat.
- Legacy Devices: Leeo listens for smoke and carbon-monoxide alarm sounds, then sends notifications to users’ phones when something is wrong.The device also tracks home climate information.
- Integration Strategy: Leeo operates independently from legacy alarms, using their sounds as triggers rather than direct communication or system dependencies.This allows smart functionality to be added without replacing existing detection devices.
- Integration Strategy: Adding smart devices around legacy systems can avoid replacement costs that might discourage consumers from adopting IoT solutions.The paper presents this approach as a way to mitigate weaknesses in legacy devices that cannot interpret context intelligently.
VII. CONCLUDING REMARKS
The survey examines marketplace IoT solutions through context-aware computing, categorizes them into five segments, and derives lessons and research opportunities for future development.
- The survey reviews IoT marketplace solutions from a context-aware computing perspective.
- It categorizes market solutions into five segments: smart wearable, smart home, smart city, smart environment, and smart enterprise.
- The survey identifies seven lessons learned and opportunities for future context-aware computing research and development.
- Its intended foundation uses understanding of past marketplace developments to support more efficient and effective future planning.