network_speed_monitor 0.0.1

A pure Dart package to monitor real-time network upload/download speed by tracking actual app traffic. Includes a stream API, quality labels, and a ready-made widget.

network_speed_monitor

A Flutter package that monitors real-time network upload/download speed by tracking actual app traffic — no fake file downloads. Includes a stream API, quality labels, and a ready-made widget.

---

✨ Features

• 📡 Live speed stream — emits NetworkSpeedSnapshot every second (configurable)
• 🏷️ Quality labels — Excellent, Good, Poor, Offline with custom thresholds
• 🎨 3 widget styles — Card, Compact pill, and Dot indicator
• ⚡ Zero native code — pure Dart + Flutter, works everywhere
• 🌐 All platforms — Android, iOS, macOS, Windows, Linux, Web
• 🔧 Fully configurable — interval, probe URL, quality thresholds

---

📦 Installation

dependencies:
  network_speed_monitor: ^0.0.1

---

🚀 Quick Start

Stream API

import 'package:network_speed_monitor/network_speed_monitor.dart';

final monitor = NetworkSpeedMonitor();

monitor.stream.listen((snapshot) {
  print('↓ ${snapshot.downloadFormatted}');     // "12.4 Mbps"
  print('↑ ${snapshot.uploadFormatted}');       // "850 Kbps"
  print('Quality: ${snapshot.quality.label}'); // "Excellent"
});

await monitor.start();

// When done:
monitor.dispose();

Ready-made Widget

// Card (full details)
NetworkSpeedIndicator(
  style: SpeedIndicatorStyle.card,
)

// Compact pill (great for app bars)
NetworkSpeedIndicator(
  style: SpeedIndicatorStyle.compact,
)

// Minimal dot (status bar)
NetworkSpeedIndicator(
  style: SpeedIndicatorStyle.dot,
)

Sharing a monitor between widgets

final monitor = NetworkSpeedMonitor();
await monitor.start();

// Pass to multiple widgets — they all share the same stream
NetworkSpeedIndicator(monitor: monitor, style: SpeedIndicatorStyle.card)
NetworkSpeedIndicator(monitor: monitor, style: SpeedIndicatorStyle.compact)

---

⚙️ Configuration

final monitor = NetworkSpeedMonitor(
  config: NetworkSpeedConfig(
    // How often to emit snapshots
    interval: Duration(seconds: 1),

    // URL for HTTP probe (used on iOS, macOS, Web, Windows, Linux)
    probeUrl: 'https://speed.cloudflare.com/__down?bytes=10000',

    // Quality classification thresholds (in Kbps)
    thresholds: QualityThresholds(
      poorBelow: 512,    // < 512 Kbps  → Poor
      goodBelow: 10000,  // < 10 Mbps   → Good, else Excellent
    ),
  ),
);

---

📊 NetworkSpeedSnapshot

Property	Type	Description
downloadKbps	double	Download speed in Kbps
uploadKbps	double	Upload speed in Kbps
downloadMbps	double	Download speed in Mbps
uploadMbps	double	Upload speed in Mbps
downloadFormatted	String	Human-readable string e.g. "12.4 Mbps"
uploadFormatted	String	Human-readable string e.g. "850 Kbps"
quality	NetworkQuality	.offline / .poor / .good / .excellent
timestamp	DateTime	When this snapshot was captured

---

🔬 How It Works — Technical Deep Dive

The package uses two different measurement strategies depending on the platform. It always picks the most accurate method available, with an automatic fallback.

Android — /proc/net/dev Byte Counter Diff

On Android, the Linux kernel exposes a virtual file at /proc/net/dev that contains cumulative byte counters for every network interface on the device (WiFi, mobile data, ethernet, etc.). This file is updated by the kernel in real time and requires no special permissions to read.

A sample of what the file looks like:

Inter-|   Receive                                                |  Transmit
 face |bytes    packets errs drop fifo frame compressed multicast|bytes    packets errs drop fifo colls carrier compressed
  lo:  123456     100    0    0    0     0          0         0   123456     100    0    0    0     0       0          0
wlan0: 45678901   9000   0    0    0     0          0         0  2345678    4500    0    0    0     0       0          0

The package reads this file every tick using dart:io's File.readAsStringSync() inside a compute() isolate (so it never blocks the UI thread), then:

1. Parses out the rx_bytes (column 0) and tx_bytes (column 8) for every non-loopback interface
2. Sums them across all interfaces to get total bytes in/out
3. Subtracts the previous tick's reading to get the delta (bytes transferred in the last interval)
4. Divides by elapsed time to get bytes per second
5. Converts to Kbps using: (deltaBytes × 8) ÷ elapsedSeconds ÷ 1000

This approach measures actual real app traffic — every byte your app (and the OS) sends or receives is counted, not just a synthetic test download.

Automatic fallback: If /proc/net/dev returns zero (some Android manufacturers restrict access), the package silently switches to the HTTP probe method described below.

iOS / macOS / Windows / Linux / Web — HTTP Probe

On platforms where kernel traffic counters aren't accessible from pure Dart, the package performs a lightweight HTTP probe each interval:

1. Opens an HttpClient connection to https://speed.cloudflare.com/__down?bytes=10000 (configurable)
2. Streams the response body, counting every byte received
3. Starts a Stopwatch when the request begins and stops it when the response completes
4. Calculates download speed: (bytesReceived × 8) ÷ elapsedSeconds ÷ 1000
5. Estimates upload speed from outgoing request header size (~400 bytes)

The probe runs within 85% of the configured interval duration to ensure it completes before the next tick fires.

Quality Classification

Every snapshot is labelled with a NetworkQuality enum value based on the download speed in Kbps:

Quality	Condition	Typical use case
offline	0 Kbps	No network / probe failed
poor	< 512 Kbps	2G / very weak signal
good	512 Kbps – 10 Mbps	3G / average WiFi
excellent	≥ 10 Mbps	4G/5G / strong WiFi

Thresholds are fully configurable via QualityThresholds.

Stream Architecture

NetworkSpeedMonitor uses a broadcast StreamController internally, which means:

• Multiple widgets or listeners can subscribe to the same monitor instance simultaneously
• No data is buffered — each subscriber only receives snapshots emitted after they subscribe
• The stream stays open until dispose() is called

---

📄 License

MIT