List of Topics:
Location Research Breakthrough Possible @S-Logix pro@slogix.in

Office Address

Social List

Privacy-Aware Federated Learning Framework for Edge-Based IoT Devices using AWS Services

Edge-Based IoT

Edge-Based IoT Devices using AWS Services

  • Use Case: To enable federated learning on distributed IoT devices (e.g., wearables, health sensors, smart meters) where data privacy is crucial, by training models locally on devices and aggregating them securely without centralizing raw data.

Objective

  • To design and implement a privacy-preserving, scalable federated learning system using AWS SageMaker for orchestration and AWS IoT Greengrass for deploying local model training to edge IoT devices.

Project Description

  • This project develops a secure and efficient federated learning architecture where IoT devices collaboratively train machine learning models without sharing raw data to a centralized server. Each device runs a local model training job using private data, and only model weights or gradients are shared to a central aggregator (in the cloud). AWS IoT Greengrass v2 facilitates local ML training, while AWS SageMaker handles model aggregation and global coordination. AWS IoT Core ensures secure communication, and privacy-preserving techniques (e.g., differential privacy, secure aggregation) are applied during transmission.
    • Key Technologies & AWS Services :
    Category Technology / AWS Service Purpose
    Machine Learning Amazon SageMaker Global model orchestration, aggregation, and re-distribution.
    Machine Learning SageMaker Edge Manager Manage, monitor, and update models on edge devices.
    Edge Deployment AWS IoT Greengrass v2 Run local training, inference, and manage federated learning components.
    Edge Deployment ML Inference Runtimes (e.g., PyTorch Mobile, TensorFlow Lite) Local training and inference on IoT devices.
    IoT Communication AWS IoT Core Secure, bi-directional communication between devices and the cloud.
    IoT Communication MQTT / Local Broker Messaging and coordination for local devices.
    Security & Privacy AWS IAM Role-based access control for edge and cloud services.
    Security & Privacy AWS KMS / AWS Secrets Manager Manage secure keys for encryption and decryption of weights/parameters.
    Security & Privacy Differential Privacy Techniques Prevent data leakage from gradients or weights.
    Data Storage Amazon S3 Store global model artifacts, logs, and metrics.
    Monitoring Amazon CloudWatch Monitor Greengrass component performance and device logs.
    Monitoring AWS IoT Device Defender Monitor and audit IoT device security compliance.