IoT Middleware, Devices and Edge Processing

Middleware Definition & Purpose

• Serves as an intermediary layer facilitating communication and data management between devices and applications.
• A software layer connecting different applications or services, enabling effective data communication and management.
• Simplifies the design of distributed mobile applications.
• Reduces the complexity of developing applications in dynamically changing environments.

The Things

• IoT devices are foundational elements that collect and transmit data, including sensors, actuators, smart meters, and wearable devices.
• Sensors detect and measure physical phenomena, converting them into signals.
• Actuators perform actions or control devices based on received data or instructions.
• Producers refers to the source of the data, which is the IoT device itself.

Network and Edge Processing

• IoT Gateways act as intermediaries between IoT devices and the cloud.
• IoT Gateways provide connectivity, local processing, data aggregation, and security.
• Device Management is crucial for managing and controlling IoT devices, including provisioning, configuration, updates, monitoring, and fault management.
• Service Agents, software components on gateways or edge servers, handle tasks like data translation, protocol conversion, and local data processing.
• Applications, UI, and DB components within the gateway run localized applications on the edge, providing user interfaces and local databases for data storage or caching.

Data Processing and Security

• Data Processing encompasses data-related functions:
  • Data Transformation: Converts data into a consistent and usable format.
  • Analytics: Analyzes data to gain insights, identify patterns, and make predictions.
  • Context Processing: Adds contextual information to make data more meaningful.
  • Storage: Persists data for later use and analysis.
• Security is paramount in IoT systems:
  • Authentication: Verifies the identity of devices and users.
  • Authorization: Controls access to resources and data based on roles and permissions.
  • Accounting: Tracks data access and usage for billing or auditing.
  • Logging: Maintains records of events for security analysis and troubleshooting.

API (Application Programming Interface)

• APIs are crucial for enabling communication and data exchange between different components and systems, i.e. between the gateway and the cloud platform, like between microservices within the platform.

Middleware Purpose/Need

• Communication Management: Ensures seamless communication between devices and layers in fog and edge architectures.
• Network Management: Manages network resources and optimizes data flow.
• Task Scheduling: Allocates tasks efficiently across available resources.
• Mobility Management: Handles the dynamic nature of mobile devices and ensures continuous service.
• Security Management: Provides mechanisms for secure data transfer and access control.

Communication Management

• Allows dynamically joining or leaving the network (e.g., MQTT for lightweight messaging).
• Establishes channels between requesting devices and ad-hoc discovered devices for data acquisition and processing.
• Maintains continuous communication, even in unreliable or low-bandwidth networks.
• Manages secure, scalable, and reliable message exchange between devices.

Network Management

• Distributes applications using SDN or virtual network topologies across fog, edge, and cloud layers.
• Monitors device connections and triggers procedures to resume lost connections.
• Manages connections between devices using technologies like TCP sockets, Wi-Fi, Wi-Fi Direct, or Bluetooth.
• Minimizes bandwidth usage while ensuring efficient data transfer across the network.

Task Scheduling & Management

• Surrogate Selection: Chooses appropriate devices to execute tasks based on policies like energy-aware selection, delay tolerance, or context-aware selection.
• Resource Monitoring: Continuously monitors the availability of resources (e.g., new user devices, VMs in fog or cloud layers).
• Dynamic Task Assignment: Matches incoming tasks with devices based on real-time requirements.
• Optimized Task Distribution: Distributes FEA tasks among multiple edge devices to minimize latency and resource usage.

Mobility Management

• Follow Me Cloud (FMC): Ensures middleware services and data follow mobile devices, like using Locator/ID separation protocols like LISP.
• Service Continuity: Guarantees uninterrupted service in highly dynamic scenarios.
• VM Migration: Manages the migration of VMs across edge and fog nodes to accommodate device mobility.
• Dynamic Device Tracking: Updates the system as devices change location using GPS or accelerometer data.

Security Management

• Authentication: Authenticates users and devices to prevent unauthorized access using lightweight mutual authentication schemes.
• Privacy Protection: Ensures data privacy through anonymization or encryption techniques, implementing policy-based access control mechanisms.
• Encryption: Encrypts data during communication to ensure confidentiality using crypto processors or lightweight encryption methods to reduce computation overhead.
• Secure Offloading: Safely offloads application tasks to other edge devices for outsourcing.

Design Challenges

• Ensuring sensing devices have the necessary contextual information to perform tasks effectively.
• Minimizing the cost associated with transferring and processing data across different tiers of the Fog and Edge Architecture (FEA).
• Handling the dynamic nature of edge devices, along with dynamic changes in their context and mobility.
• Meeting strict real-time response requirements while ensuring processing is done close to the source.

Design Goals

• Ad-Hoc Device Discovery: Dynamically discover and establish communication with edge devices.
• Run-Time Execution Environment: Provide a platform for remote code execution on edge devices.
• Minimal Task Disruption: Minimize interruptions during task execution due to mobility.
• Overhead of Operational Parameters: Minimize bandwidth and energy usage during middleware operations.
• Context-Aware Adaptive Design: Adapt to dynamic changes in device context and environment.

Quality of Service (QoS) Design Goals

• Ensures middleware meets specific QoS requirements of applications.
• Real-Time Response: Applications like autonomous vehicles or emergency response systems require ultra-low latency.
• Data Relevance: Acquired data must be relevant and accurate for the application.
• Uninterrupted Data Acquisition: Middleware must ensure continuous data collection even in dynamic environments.

State-of-the-Art Middleware Infrastructures

• Supports real-time data streaming applications like traffic monitoring, real-time replay in stadiums, video analytics, emergency rescue, missing persons search, and smart vehicle-to-vehicle systems.
• Recent Middleware Solutions:
  • FemtoCloud supports data analytics and optimized device selection.
  • Nakamura et al. proposes in-device processing for IoT.
  • CloudAware adapts dynamically to changing network conditions.
  • MEC-based and Fog-based middleware include Carrega et al. and Grewe et al.
• Middleware Approaches proposes microservices-based and service-oriented middleware. It also uses context-aware middleware (MobiPADs, MobiCon) and cloudlet-based solutions for real-time response and offloading.

Microservice Approach

• Designed with independent, lightweight microservices that interact via APIs.
• Dynamically scale and deploy authentication, scheduling, and data analytics services in fog nodes.
• Advantages: Scalability, Modularity, Flexibility.

Service-Oriented Middleware (SOM)

• Applications interacts through well-defined service interfaces.
• Global Sensor Networks (GSN) are widely used, processing data using a service-oriented approach.
• Key Features:
  • Interoperability: Applications can use standard service calls (SOAP, REST).
  • Dynamic Composition: Services can be composed dynamically based on application needs.
  • Scalability: Can be deployed across multiple edge/fog nodes.

Context-Aware Middleware

• Adapts dynamically based on real-time contextual changes (e.g., user location, device movement, network conditions).
• Examples: MobiPADs and MobiCon based on sensor inputs.
• Advantages: Personalized User Experience, Efficiency, Resilience.

Cloudlet-Based Middleware

• Uses cloudlets (small-scale, low-latency cloud nodes close to edge devices) to handle computing and storage.
• The CloudAware middleware allows edge devices to seamlessly offload tasks, using a "data center in a box" model with real-time processing.
• Advantages: Low Latency, Energy Efficiency, Support for Mobile Users.

Middleware – Proposed Architecture

• Middleware Services in Fog Edge Computing: API Code, Security with Authentication, Privacy and Encryption; and Device Discovery
• Middleware Components in Fog Edge Computing: Context Monitoring and Prediction; Selection of Participating Devices with Energy-Aware Selection, Delay-Tolerance-Based Selection, Context-Aware Selection; Data Analytics; Scheduling and Resource Management; Network Management; Execution Management; Mobility Management

Case Study 1: Smart City Traffic Management

• Scenario: A smart city uses traffic cameras, sensors, and fog servers to monitor congestion.
• Problem: Sending all raw data to the cloud causes delays, network congestion, and high costs.
• Solution: Use Fog Computing for the below advantages:
  • Discover nearby cameras and traffic sensors.
  • Locally processes Images and only sends critical alerts to authorities.
  • Assigns tasks to another server if one fog server is too busy.
  • Ensures continuous data tracking across edge devices.
  • Ensures only authorized devices can access sensitive traffic data.

Case Study 2: Smart Thermostat Functionality

• Middleware enhances smart thermostat through: -Detecting the current room conditions and user behavior to optimize thermostat operation as well as using weather forecast data to pre-adjust heating/cooling before drastic temperature changes
  • Filtering unnecessary data to reduce cloud communication and to send only relevant information
  • Using device discovery to determine if nearby devices detect if someone is in the room.
  • Assigns tasks to the right processing unit and balances processing between multiple smart home devices to avoid delays.
  • Ensures secure data from cyber threats and prevents unauthorized controls.
  • Enables seamless automation by controlling other smart home devices and by switching to eco mode if no one is home.