PestiSafe 🌿

An Android IoT application for real-time pesticide residue detection on agricultural produce. PestiSafe connects to an electrochemical sensor over a local WiFi network, converts raw voltage readings into pesticide concentration values (ppm) via a calibration curve, and compares results against FAO Codex Maximum Residue Limits (MRLs) to give a clear safe / unsafe verdict on sampled produce.

---

Table of Contents

• Overview
• Features
• Architecture
• Screen Flow
• Database Schema
• Setup & Requirements
• Dependencies
• How It Works
• Project Structure
• Known Limitations

---

Overview

PestiSafe is designed for field use by agricultural inspectors, food safety auditors, or researchers who need a fast, portable way to test produce for pesticide contamination. The workflow is:

1. Connect your Android device to the same WiFi network as the sensor.
2. Fetch the latest pesticide & MRL database from the FAO Codex API.
3. Select the pesticide you are testing for.
4. Scan the local network to find the sensor device.
5. Calibrate the sensor using known reference concentrations.
6. Point the sensor at a food sample — view live concentration readings and an instant safety verdict.
7. Export results as CSV, JSON, or Excel for reporting.

---

Features

Feature	Detail
Live sensor readings	Polls the IoT sensor over HTTP every 3 seconds and displays real-time voltage and concentration
Calibration engine	Collects voltage readings at 8 known concentrations, fits a least-squares linear regression, and validates with R² ≥ 0.9
FAO MRL database	Fetches and caches the complete FAO Codex pesticide + Maximum Residue Limit dataset on-device
Safety verdict	Compares live concentration against the FAO MRL for the selected pesticide-commodity pair; displays a clear SAFE ✅ or UNSAFE ⚠️ result
IP Scanner	Scans the local /24 subnet to discover sensor devices automatically
Data export	Export all captured readings as CSV, JSON, or Excel (.xlsx) and share via Android's standard file share sheet
User accounts	Local account system with BCrypt-hashed passwords and a DOB-verified password reset flow
Secure networking	Cleartext HTTP restricted to LAN IP ranges only (FAO API uses HTTPS)

---

Architecture

PestiSafe follows Modern Android Architecture with a clean separation of concerns:

UI Layer          →   Jetpack Compose + Material Design 3
State Management  →   ViewModel + MutableStateFlow / LiveData
Data Layer        →   Room Database + Repository pattern
Async             →   Kotlin Coroutines + SupervisorJob
Navigation        →   Jetpack Navigation Compose

Key design decisions

• Single ViewModel (MainViewModel) acts as the central state owner for sensor data, calibration coefficients, and navigation state.
• Repository pattern wraps all DAO access so the ViewModel never talks to Room directly.
• Reactive UI — all DAO queries return Flow or LiveData, so the chart and readings update automatically when new data is captured.
• Side-effect safety — all navigation calls and state mutations are performed inside LaunchedEffect blocks, never directly during Compose's composition phase.

---

Screen Flow

Login / Sign-Up
      │
      ▼
Main (Connect Screen)
  • Check WiFi connection
  • Fetch FAO pesticide + MRL dataset
      │
      ▼
Pesticide Selection
  • Choose target pesticide from FAO list
      │
      ▼
IP Scanner
  • Scan local /24 subnet (254 hosts concurrently)
  • Select sensor device IP
      │
      ▼
Calibration
  • Record voltage at 8 known concentrations
  • Fit linear regression (requires R² ≥ 0.9)
      │
      ▼
Home (Live Monitoring)
  • Real-time voltage + concentration display
  • Line chart of all captured readings
  • Select pesticide + commodity → view FAO MRL
  • SAFE / UNSAFE verdict
  • "Capture Reading" button → saves to database
  • Export data (CSV / JSON / Excel)

A dropdown menu in the top app bar allows jumping between any screen at any time.

---

Database Schema

Five Room entities:

User

Column	Type	Notes
id	Int (PK, autoGen)
name	String	Display name
username	String	Unique — enforced by index
email	String
password	String	BCrypt hashed
dob	String	Used for password reset verification

Pesticide

Column	Type	Notes
id	Int (PK)	FAO Codex pesticide ID
name	String	English name from FAO API

Commodity

Column	Type	Notes
id	Int (PK, autoGen)
name	String	Crop / food product name

MRL

Column	Type	Notes
mrlID	Int (PK, autoGen)
pesticideID	Int (FK → Pesticide)	CASCADE delete
commodityID	Int (FK → Commodity)	CASCADE delete
mrl	Double	Maximum Residue Limit in ppm

DataValue

Column	Type	Notes
id	Int (PK, autoGen)
voltage	Double	Raw sensor reading (V)
concentration	Double	Calculated concentration (ppm)

---

Setup & Requirements

Prerequisites

• Android Studio Hedgehog or newer
• Android device or emulator running Android 7.0+ (API 24+)
• A compatible electrochemical sensor device on the same WiFi network

Build

1. Clone the repository:

   git clone https://github.com/Shash976/wifi-data-app.git
   cd wifi-data-app

2. Open in Android Studio.
3. Let Gradle sync and download dependencies.
4. Run on a device or emulator (Run > Run 'app').

Permissions

The app requests the following at runtime:

Permission	Reason
INTERNET	FAO API calls + sensor HTTP polling
ACCESS_WIFI_STATE	Check WiFi connectivity before connecting
ACCESS_NETWORK_STATE	Network availability checks
READ/WRITE_EXTERNAL_STORAGE	File export on Android 9 and below (not needed on API 29+)

Cleartext HTTP is allowed only for local LAN IP ranges (192.168.x.x, 10.x.x.x, 172.16.x.x) via a Network Security Config. All external traffic (including FAO API) uses HTTPS.

---

Dependencies

Library	Version	Purpose
Jetpack Compose BOM	2025.06.00	Declarative UI framework
Material Design 3	—	UI components and theming
Room	2.7.1	Local SQLite ORM
Navigation Compose	2.9.0	Screen routing
Lifecycle ViewModel	2.9.1	State management
Kotlin Coroutines	1.8.0	Async operations
Vico Charts	2.0.0-alpha.19	Line chart rendering
Gson	2.10.1	FAO API JSON deserialization
BCrypt (jbcrypt)	0.9.0	Password hashing
Apache POI OOXML	5.2.3	Excel (.xlsx) export
WorkManager	2.10.1	Background task scheduling (future use)

SDK versions: minSdk = 24, targetSdk = 35

---

How It Works

Calibration

The sensor produces a voltage proportional to pesticide concentration. Calibration establishes the mapping:

Concentration (ppm) = (Voltage - Intercept) / Slope

The app steps through 8 reference concentrations [1.0, 10.0, 5.0, 7.5, 6.0, 2.5, 4.0, 1.25] ppm. At each step, the user places the sensor in a reference solution and captures the voltage. Once ≥ 3 points are collected, least-squares linear regression is fitted. Calibration is accepted only when R² ≥ 0.9.

FAO MRL Lookup

On the Connect screen, the app fetches from two FAO Codex endpoints:

• Pesticide list: https://www.fao.org/jsoncodexpest/jsonrequest/pesticides/index.html
• MRL details per pesticide: https://www.fao.org/jsoncodexpest/jsonrequest/pesticides/details.html?id={id}

Results are cached in the local Room database and deduped on subsequent fetches.

Safety Verdict

Once a pesticide and commodity are selected on the Home screen:

1. The MRL for that pair is retrieved from the local database.
2. The current concentration (converted from live sensor voltage using the calibration coefficients) is compared to the MRL.
3. If concentration > MRL → UNSAFE (red warning).
4. If concentration ≤ MRL → SAFE (green confirmation).

Data Capture & Export

Live readings update the display continuously but are not auto-saved. Tap "Capture Reading" to intentionally commit a reading to the database. Captured readings appear in the chart and can be exported:

• CSV — comma-separated with headers
• JSON — Gson-serialized array
• Excel — .xlsx workbook with column headers and auto-sized columns (via Apache POI)

Files are named {username}_{pesticide}_{commodity}_data.{ext} and shared via Android's standard share sheet.

---

Project Structure

app/src/main/java/com/example/pestisafe/
├── MainActivity.kt              # Entry point, Navigation host, Scaffold
├── MainViewModel.kt             # Central state, sensor polling, FAO API, calibration math
├── Database.kt                  # Room entities, DAOs, Repository
├── Routes.kt                    # Navigation route enum
│
├── LoginSignUp.kt               # Auth screen (login, sign-up, password reset)
├── ConnectScreen.kt             # WiFi check + FAO data fetch
├── PesticideSelectionScreen.kt  # Pesticide picker dropdown
├── ScanScreen.kt                # LAN IP scanner
├── CalibrationScreen.kt         # Sensor calibration wizard
├── HomeScreen.kt                # Live monitoring, MRL verdict, export
│
├── MathFuncs.kt                 # Linear regression, R² calculation
├── Chart.kt                     # Vico chart configuration
├── Deserializers.kt             # Gson deserializers for FAO API JSON
├── DownloadFuncs.kt             # CSV / JSON / Excel export + file sharing
├── Formats.kt                   # Export format enum
│
└── ui/theme/
    ├── Theme.kt                 # Material3 theme setup
    ├── Color.kt                 # Color palette
    └── Type.kt                  # Typography

---

Known Limitations

• Offline mode — The app requires WiFi throughout: for the FAO API fetch on first use and for sensor communication. No offline fallback exists once data is cached.
• Single sensor — The app connects to one sensor device at a time.
• HTTP sensor protocol — Sensor devices communicate over plain HTTP (acceptable for LAN-only use but not suitable for untrusted networks).
• Legacy v1 code — The v1/ package contains an older Activity-based implementation. It is not used by the current Compose app and will be removed in a future cleanup.

---

License

This project is developed as part of an academic research project. All rights reserved.