Optimizing Messaging and Designing Broker-Based SDKs in IoT Embedded Environments

gpt-4-turbo has translated this article into English.

1. Messaging and Asynchronous Structures in IoT Environments

• Data linkage issues between IoT and the cloud
• Cloud environments transitioning to Microservice Architecture (MSA)
  • Switch from traditional synchronous API-based to asynchronous messaging-centric structures
• Limitations in Embedded Environments: Overhead and Limited Computing Resources
  • Serial (UART, SPI, I2C) based data processing methods
  • Inefficiencies due to Polling-centric synchronous processing structures

2. Broker Strategies to Reduce Messaging Overhead

• Challenges of Implementing Asynchronous Messaging in Embedded Environments
  • Traditional synchronous processing methods lead to increased computational demand and real-time data loss
  • Network instability (LPWAN, LoRa, Wi-Fi, BLE) causing message integrity issues
• Need for Messaging Optimization Using Brokers
  • Suitable for cloud MSA through transition to asynchronous messaging (Sync → Async conversion)
  • Reduced computational load from message serialization (Serialize) and deserialization (Deserialize)
• Role of Brokers in Embedded Systems
  • Asynchronous processing using message queues in low-power devices
  • Conversion of serial data into message format for linkage with the cloud

3. Linking Serial Data and Messaging Systems

• Serial communications (UART, SPI, I2C) are key interfaces in embedded systems
• Efficient conversion of serial data into messaging format is necessary
• Application of lightweight serialization methods to reduce overhead
  • Utilizing CBOR, MessagePack, Protobuf, FlatBuffers instead of JSON
• Need for broker and SDK structure for Serial → Message conversion

Example Data Flow

flowchart LR
    A[IoT Sensor] --> B((Serial Interface))
    B --> C[Broker]
    C --> D((CBOR Conversion))
    D --> E[Cloud Messaging]

4. Lightweight Messaging SDK Design Based on Brokers

• Role of the SDK: Converting serial data into a messaging structure
• Application of RTOS-based Event-Driven Architecture
  • Elimination of Polling methods → Asynchronous event-based messaging processing
  • Use of Task Queue, Event Queue for low-power optimization
• Integration with Lightweight Messaging Protocols
  • Utilizing MQTT-SN, CoAP, Zenoh, Micro DDS

SDK Layer Structure

graph TD
    A[Sensor]
    B[Serial Parser]
    C[Data Serializer]
    D[Messaging Adapter]
    E[Cloud]

    A --> B
    B --> C
    C --> D
    D --> E

5. Integration with Cloud and Microservice Environments

• Standardization of interfaces between embedded environments and the cloud
• Scalability through messaging brokers
• Linkage strategies with Serverless and Event-Based Architectures

Example IoT Messaging Flow

flowchart LR
    A[Embedded Device] --> B[Edge Broker]
    B --> C[Cloud Messaging Broker]
    C --> D[Microservices]

6. Conclusion and Future Directions

• Instead of directly implementing asynchronous messaging in embedded systems, utilize brokers
• Essential SDK structure for linking serial data to cloud messaging
• Optimization needed by combining RTOS-based event-driven structure with lightweight serialization methods