IoT Data Analytics & Concurrent Client Handling Sever

Java program to simulate an IoT analytics server which receives sensor data and provides various services to clients such as notification, aggregation, and predictive modelling. Through socket programming, the handling of concurrent clients was ensured while maintaining principles of Quality of Services (QoS). Event handling was implemented using a Earliest Deadline First (EDF) approach to optimize client satisfaction.

Implemented a serverless application running on AWS Lambda.template.yaml is the config file for the Lambda function that specifies the ReST API path : \lmbdafn, and the Handler : cpen221.mp3.lambda.Function::handleRequest. Then utilized AWS's sam to build and deploy mp3-Erling on AWS. Currently the lambda function is triggered at the endpoint https://2a9yquga3f.execute-api.us-east-2.amazonaws.com/default/IoT_Prediction_Overhead/lmbdafn with a HTTP GET request with three parameters. This endpoint is hit by a PredictorThread.

Key Components

Entities and Events

• Sensors are elements that generate data through SensorEvents, while Actuators are active entities with mutable states that can be controlled remotely via ActuatorEvents.
• Events are sent to the server through a socket connection, with a configurable frequency
• Entities requires an endpoint to send events.

Client-Server Interactions

Client Requests

• Supported request types include:
  • CONFIG: Configure entity settings.
  • CONTROL: Adjust actuator states.
  • ANALYSIS: Run analytics on sensor data.
  • PREDICT: Forecast sensor data patterns.

Server Service

• Offers control and analytics services.
• Manages actuator states and event ordering using an Earliest Deadline First scheduling algorithm.