Implement mobile application for SCIoT client

[lbedogni]

Problem
Currently, the SCIoT system only supports Raspberry Pi and ESP32 clients for edge inference and offloading. Mobile devices (smartphones/tablets) represent a significant use case for edge AI applications but lack native client support.

Motivation

• Ubiquity: Mobile devices are widely available and have significant computational power
• Use Cases: Real-time image recognition, video analysis, sensor fusion
• Platform Coverage: iOS and Android represent the majority of edge computing devices
• Camera Integration: Built-in high-quality cameras and sensors

Proposed Implementation

Phase 1: Architecture & Design

• Design mobile-specific architecture considering battery constraints
• Evaluate cross-platform frameworks (Flutter/React Native vs native Swift/Kotlin)
• Define API compatibility layer for existing server endpoints
• Design power-efficient inference scheduling

Phase 2: Core Features

• Camera Integration: Real-time video capture with configurable resolution/FPS
• Model Loading: Support for TensorFlow Lite / Core ML / ONNX models
• Inference Pipeline: On-device inference with layer-wise profiling
• Offloading Client: HTTP/MQTT communication with edge server
• Network Monitoring: Adaptive offloading based on latency and battery level

Phase 3: Platform-Specific Implementation

iOS (Swift/SwiftUI)

• AVFoundation for camera capture
• Core ML for on-device inference
• Vision framework for preprocessing
• Network framework for adaptive connectivity

Android (Kotlin/Jetpack Compose)

• CameraX API for camera management
• TensorFlow Lite for inference
• ML Kit for preprocessing
• OkHttp/Retrofit for networking

Phase 4: User Interface

• Live camera preview with inference overlay
• Real-time metrics dashboard (latency, FPS, battery usage)
• Offloading strategy selector (always local / always remote / adaptive)
• Model selection and configuration

Phase 5: Testing & Optimization

• Battery consumption profiling
• Network efficiency testing (Wi-Fi, 4G/5G)
• Comparative analysis vs Raspberry Pi client
• User experience testing

Technical Considerations

• Power Management: Background inference throttling, adaptive frame rate
• Model Optimization: Quantization for mobile deployment
• Connectivity: Handle network transitions gracefully (Wi-Fi ↔ cellular)
• Privacy: On-device processing options, data retention policies
• Compatibility: Support iOS 14+ and Android 8+

Deliverables

• Mobile architecture design document
• iOS application (Swift/SwiftUI)
• Android application (Kotlin/Jetpack Compose)
• Mobile-specific configuration management
• Battery and performance benchmarks
• User documentation and setup guide

Dependencies

• Requires completion of Restructure the code #22 (code restructuring for modularity)
• May benefit from Consolidate config loading into a singleton Config class #8 (config consolidation)

Files/Components to Create

• src/mobile/ios/ - iOS application source
• src/mobile/android/ - Android application source
• src/mobile/shared/ - Shared business logic (if using cross-platform approach)
• docs/MOBILE_SETUP.md - Mobile development and deployment guide

Estimated Effort: Large (6-8 weeks for both platforms)

Priority: Medium (expands platform coverage, significant user value)

[lbedogni]

This issue has been split into smaller tasks for better management:

Sub-tasks Tracking

• [Sub-issue for #23] Mobile Core: Cross-platform Architecture Setup #27 Mobile Core: Cross-platform Architecture Setup
• [Sub-issue for #23] Mobile: Platform-specific Camera & ML Integration #28 Mobile: Platform-specific Camera & ML Integration
• [Sub-issue for #23] Mobile UI: Dashboard and Live Stream View #29 Mobile UI: Dashboard and Live Stream View