📚 课程目录第4章边缘计算框架选型与部署

4. 边缘计算框架选型与部署

任务目标

为工业IoT边缘智能网关选择并部署合适的边缘计算框架，实现数据接入、规则处理与云端转发的最小闭环。本案例聚焦于框架选型对比与轻量级部署验证，不涉及具体业务算法。

框架选型结论

框架	适用场景	资源占用	选型建议
Node-RED	快速原型、流式编排	~80MB RAM	适合中小型网关，开发效率高
EdgeX Foundry	复杂设备接入、协议转换	~500MB RAM	适合大型网关，功能全但偏重
EMQX + eKuiper	高并发MQTT + SQL规则处理	~120MB RAM	适合工业实时数据处理场景

本案例选用Node-RED作为部署对象，原因：资源占用低、内置MQTT与HTTP节点、可视化流编辑降低维护成本，且可通过Docker实现分钟级部署。

环境依赖

操作系统：Linux x86_64 / ARM64（树莓派4B及以上）
Docker Engine ≥ 20.10
Docker Compose ≥ 2.0
内存 ≥ 512MB（建议1GB）

完整实现

创建docker-compose.yml，一次性启动Node-RED与Eclipse Mosquitto（MQTT Broker），形成数据采集→规则处理→转发的最小链路：

version: "3.8"

services:
  mosquitto:
    image: eclipse-mosquitto:2.0
    container_name: iot-mqtt-broker
    ports:
      - "1883:1883"
    volumes:
      - ./mosquitto/config:/mosquitto/config
      - mosquitto-data:/mosquitto/data
      - mosquitto-log:/mosquitto/log
    restart: unless-stopped

  node-red:
    image: nodered/node-red:3.1.9
    container_name: iot-node-red
    depends_on:
      - mosquitto
    ports:
      - "1880:1880"
    environment:
      - TZ=Asia/Shanghai
    volumes:
      - ./node-red-data:/data
    restart: unless-stopped

volumes:
  mosquitto-data:
  mosquitto-log:

创建Mosquitto配置文件mosquitto/config/mosquitto.conf：

listener 1883
allow_anonymous true
persistence true
persistence_location /mosquitto/data/
log_dest file /mosquitto/log/mosquitto.log

启动服务并导入Node-RED流（flows.json），实现：订阅sensor/data主题 → 提取温度值 → 超过50℃则发送告警到alarm/notice，同时将原始数据转发至cloud/upload：

[
  {
    "id": "mqtt-in-1",
    "type": "mqtt in",
    "topic": "sensor/data",
    "broker": "mqtt-broker-1",
    "name": "Sensor Data",
    "qos": "1",
    "datatype": "auto"
  },
  {
    "id": "mqtt-broker-1",
    "type": "mqtt-broker",
    "broker": "mosquitto",
    "port": "1883",
    "clientid": "node-red-gateway",
    "usetls": false,
    "protocol": "mqtt"
  },
  {
    "id": "function-1",
    "type": "function",
    "name": "Threshold Check",
    "func": "const payload = JSON.parse(msg.payload);\nif (payload.temperature > 50) {\n    msg.topic = 'alarm/notice';\n    msg.payload = { deviceId: payload.deviceId, temp: payload.temperature, ts: Date.now() };\n    return msg;\n}\nreturn null;"
  },
  {
    "id": "mqtt-out-1",
    "type": "mqtt out",
    "broker": "mqtt-broker-1",
    "topic": "cloud/upload",
    "qos": "1",
    "name": "Cloud Upload"
  },
  {
    "id": "mqtt-out-2",
    "type": "mqtt out",
    "broker": "mqtt-broker-1",
    "topic": "alarm/notice",
    "qos": "1",
    "name": "Alarm Output"
  },
  {
    "id": "link-1",
    "type": "link out",
    "name": "To Alarm"
  },
  {
    "id": "link-2",
    "type": "link in",
    "name": "From Function"
  }
]

将上述JSON保存为node-red-data/flows.json后重启Node-RED容器：

docker compose up -d
docker restart iot-node-red

运行与验证

访问http://<网关IP>:1880，确认Node-RED界面中四个节点已自动部署且无红色错误标记。
在终端发布模拟数据：

mosquitto_pub -h localhost -t sensor/data -m '{"deviceId":"dev-001","temperature":48.5}'
mosquitto_pub -h localhost -t sensor/data -m '{"deviceId":"dev-001","temperature":55.2}'

订阅输出主题验证：

mosquitto_sub -h localhost -t cloud/upload -v
mosquitto_sub -h localhost -t alarm/notice -v

预期输出（由代码逻辑直接推出）：

cloud/upload收到两条原始数据（48.5和55.2均转发）
alarm/notice仅收到第二条（55.2 > 50触发告警）

通过条件：上述两条预期均满足，且Node-RED日志无异常报错，即判定框架部署成功且规则链路工作正常。

常见问题

端口冲突：若1880或1883被占用，修改docker-compose.yml中映射端口后重新docker compose up -d。
流未自动加载：确认flows.json位于node-red-data/目录且文件名正确，重启容器后生效。
MQTT连接失败：检查Mosquitto容器是否健康运行，执行docker logs iot-mqtt-broker查看日志。