feat: 控制指令消息改为protobuf二进制

2026-07-15 17:03:58 +08:00
parent 92ad90cbeb
commit da1730ce19
21 changed files with 658 additions and 147 deletions
--- a/WebRTCControlled/app/build.gradle
+++ b/WebRTCControlled/app/build.gradle
@@ -1,5 +1,6 @@
 plugins {
    id 'com.android.application'
+    id 'com.google.protobuf'
 }

 def releaseTime() {
@@ -124,6 +125,19 @@ android {

 }

+protobuf {
+    protoc {
+        artifact = 'com.google.protobuf:protoc:3.25.1'
+    }
+    generateProtoTasks {
+        all().each { task ->
+            task.builtins {
+                java {}
+            }
+        }
+    }
+}
+
 dependencies {
    implementation 'androidx.appcompat:appcompat:1.6.1'
    implementation 'com.google.android.material:material:1.11.0'
@@ -137,6 +151,8 @@ dependencies {
 //    implementation 'org.webrtc:google-webrtc:1.0.32006'
    // OkHttp for WebSocket
    implementation 'com.squareup.okhttp3:okhttp:4.12.0'
-    // Gson for JSON
+    // Gson for JSON（信令消息仍使用 JSON）
    implementation 'com.google.code.gson:gson:2.10.1'
+    // Protobuf（DataChannel 控制指令二进制）
+    implementation 'com.google.protobuf:protobuf-java:3.25.1'
 }
--- a/WebRTCControlled/app/src/main/java/com/ttstd/controlled/input/InputCommandHandler.java
+++ b/WebRTCControlled/app/src/main/java/com/ttstd/controlled/input/InputCommandHandler.java
@@ -2,17 +2,18 @@ package com.ttstd.controlled.input;

 import android.util.Log;

-import com.google.gson.Gson;
-import com.google.gson.JsonObject;
+import com.google.protobuf.InvalidProtocolBufferException;
+import com.ttstd.control.Action;
+import com.ttstd.control.ControlMessage;

 /**
 * 处理控制端发来的输入指令（触摸、按键）。
- * 解析指令并委托给 {@link InputExecutor} 接口执行。
+ * 指令以 protobuf 二进制（{@link ControlMessage}）通过 DataChannel 传输，
+ * 解析后委托给 {@link InputExecutor} 接口执行。
 */
 public class InputCommandHandler {

    private static final String TAG = "InputCommandHandler";
-    private final Gson gson = new Gson();
    private final int screenWidth;
    private final int screenHeight;
    private final InputExecutor inputExecutor;
@@ -24,47 +25,47 @@ public class InputCommandHandler {
        this.inputExecutor = new SystemInputUtils();
    }

-    public void handleCommand(String commandJson) {
+    public void handleCommand(byte[] data) {
        try {
-            JsonObject command = gson.fromJson(commandJson, JsonObject.class);
-            String action = command.get("action").getAsString();
+            ControlMessage command = ControlMessage.parseFrom(data);
+            Action action = command.getAction();

            switch (action) {
-                case "TOUCH":
+                case TOUCH:
                    handleTouch(command);
                    break;
-                case "KEY":
+                case KEY:
                    handleKey(command);
                    break;
-                case "SWIPE":
+                case SWIPE:
                    handleSwipe(command);
                    break;
-                case "LONG_PRESS":
+                case LONG_PRESS:
                    handleLongPress(command);
                    break;
-                case "MOTION_EVENT":
+                case MOTION_EVENT:
                    handleMotionEvent(command);
                    break;
                default:
                    Log.w(TAG, "Unknown action: " + action);
            }
-        } catch (Exception e) {
-            Log.e(TAG, "Error handling command: " + e.getMessage(), e);
+        } catch (InvalidProtocolBufferException e) {
+            Log.e(TAG, "Error parsing command: " + e.getMessage(), e);
        }
    }

-    private void handleMotionEvent(JsonObject command) {
-        int action = command.get("motionAction").getAsInt();
-        float relX = command.get("x").getAsFloat();
-        float relY = command.get("y").getAsFloat();
+    private void handleMotionEvent(ControlMessage command) {
+        int action = command.getMotionAction();
+        float relX = (float) command.getX();
+        float relY = (float) command.getY();
        int x = (int) (relX * screenWidth);
        int y = (int) (relY * screenHeight);
        inputExecutor.injectMotionEvent(action, x, y);
    }

-    private void handleTouch(JsonObject command) {
-        float relX = command.get("x").getAsFloat();
-        float relY = command.get("y").getAsFloat();
+    private void handleTouch(ControlMessage command) {
+        float relX = (float) command.getX();
+        float relY = (float) command.getY();
        int x = (int) (relX * screenWidth);
        int y = (int) (relY * screenHeight);

@@ -72,18 +73,18 @@ public class InputCommandHandler {
        inputExecutor.injectTap(x, y);
    }

-    private void handleKey(JsonObject command) {
-        int keyCode = command.get("keyCode").getAsInt();
+    private void handleKey(ControlMessage command) {
+        int keyCode = command.getKeyCode();
        Log.d(TAG, "Key press: " + keyCode);
        inputExecutor.injectKeyEvent(keyCode);
    }

-    private void handleSwipe(JsonObject command) {
-        float relX1 = command.get("x1").getAsFloat();
-        float relY1 = command.get("y1").getAsFloat();
-        float relX2 = command.get("x2").getAsFloat();
-        float relY2 = command.get("y2").getAsFloat();
-        long duration = command.has("duration") ? command.get("duration").getAsLong() : 300;
+    private void handleSwipe(ControlMessage command) {
+        float relX1 = (float) command.getX1();
+        float relY1 = (float) command.getY1();
+        float relX2 = (float) command.getX2();
+        float relY2 = (float) command.getY2();
+        long duration = command.getDuration();

        int x1 = (int) (relX1 * screenWidth);
        int y1 = (int) (relY1 * screenHeight);
@@ -94,9 +95,9 @@ public class InputCommandHandler {
        inputExecutor.injectSwipe(x1, y1, x2, y2, duration);
    }

-    private void handleLongPress(JsonObject command) {
-        float relX = command.get("x").getAsFloat();
-        float relY = command.get("y").getAsFloat();
+    private void handleLongPress(ControlMessage command) {
+        float relX = (float) command.getX();
+        float relY = (float) command.getY();
        int x = (int) (relX * screenWidth);
        int y = (int) (relY * screenHeight);

--- a/WebRTCControlled/app/src/main/java/com/ttstd/controlled/service/ScreenCaptureService.java
+++ b/WebRTCControlled/app/src/main/java/com/ttstd/controlled/service/ScreenCaptureService.java
@@ -207,7 +207,7 @@ public class ScreenCaptureService extends Service {
        // 初始化 WebRTC
        webRtcClient = new WebRtcClient(this, wsClient, deviceId);
        webRtcClient.initialize(eglBase);
-        webRtcClient.setInputCallback(commandJson -> inputHandler.handleCommand(commandJson));
+        webRtcClient.setInputCallback(commandBytes -> inputHandler.handleCommand(commandBytes));

        // 初始化屏幕捕获
        screenCapturer = new ScreenCapturerAndroid(resultData, new MediaProjection.Callback() {
--- a/WebRTCControlled/app/src/main/java/com/ttstd/controlled/webrtc/WebRtcClient.java
+++ b/WebRTCControlled/app/src/main/java/com/ttstd/controlled/webrtc/WebRtcClient.java
@@ -59,7 +59,7 @@ public class WebRtcClient {
    private InputCommandCallback inputCallback;

    public interface InputCommandCallback {
-        void onCommandReceived(String commandJson);
+        void onCommandReceived(byte[] data);
    }

    public WebRtcClient(Context context, WebSocketClient wsClient, String deviceId) {
@@ -316,10 +316,9 @@ public class WebRtcClient {
            public void onMessage(DataChannel.Buffer buffer) {
                byte[] data = new byte[buffer.data.remaining()];
                buffer.data.get(data);
-                String message = new String(data, StandardCharsets.UTF_8);
-                Log.d(TAG, "DataChannel message: " + message);
+                Log.d(TAG, "DataChannel message received, bytes=" + data.length);
                if (inputCallback != null) {
-                    inputCallback.onCommandReceived(message);
+                    inputCallback.onCommandReceived(data);
                }
            }
        });
--- a/WebRTCControlled/app/src/main/proto/control_message.proto
+++ b/WebRTCControlled/app/src/main/proto/control_message.proto
@@ -0,0 +1,41 @@
+syntax = "proto3";
+
+package com.ttstd.control;
+
+option java_package = "com.ttstd.control";
+option java_multiple_files = true;
+
+// 控制指令类型，对应原 JSON 字段 action。
+enum Action {
+  ACTION_UNKNOWN = 0;
+  TOUCH = 1;
+  SWIPE = 2;
+  KEY = 3;
+  LONG_PRESS = 4;
+  MOTION_EVENT = 5;
+}
+
+// 通过 RTCDataChannel 传输的控制指令（protobuf 二进制）。
+// 坐标 x/y/x1/y1/x2/y2 均为相对屏幕的百分比，取值范围 0.0 ~ 1.0。
+message ControlMessage {
+  // 指令类型
+  Action action = 1;
+
+  // 单点坐标（TOUCH / LONG_PRESS / MOTION_EVENT）
+  double x = 2;
+  double y = 3;
+
+  // 滑动起止坐标（SWIPE）
+  double x1 = 4;
+  double y1 = 5;
+  double x2 = 6;
+  double y2 = 7;
+  int64 duration = 8;
+
+  // 按键（KEY）：key_action 0=按下 1=抬起（对应 Android KeyEvent ACTION_DOWN/UP）
+  int32 key_code = 9;
+  int32 key_action = 10;
+
+  // 原始动作（MOTION_EVENT）：对应 Android MotionEvent ACTION_*（0=DOWN 1=UP 2=MOVE）
+  int32 motion_action = 11;
+}
--- a/WebRTCControlled/build.gradle
+++ b/WebRTCControlled/build.gradle
@@ -1,5 +1,6 @@
 // Top-level build file
 plugins {
    id 'com.android.application' version '8.1.4' apply false
+    id 'com.google.protobuf' version '0.9.4' apply false
 }
 apply from: "config.gradle"
--- a/WebRTCController/app/build.gradle
+++ b/WebRTCController/app/build.gradle
@@ -1,5 +1,6 @@
 plugins {
    id 'com.android.application'
+    id 'com.google.protobuf'
 }

 def releaseTime() {
@@ -116,6 +117,19 @@ android {
    }
 }

+protobuf {
+    protoc {
+        artifact = 'com.google.protobuf:protoc:3.25.1'
+    }
+    generateProtoTasks {
+        all().each { task ->
+            task.builtins {
+                java {}
+            }
+        }
+    }
+}
+
 dependencies {
    implementation 'androidx.appcompat:appcompat:1.6.1'
    implementation 'com.google.android.material:material:1.11.0'
@@ -127,6 +141,8 @@ dependencies {
    // OkHttp for WebSocket
    implementation 'com.squareup.okhttp3:okhttp:4.12.0'

-    // Gson for JSON
+    // Gson for JSON（信令消息仍使用 JSON）
    implementation 'com.google.code.gson:gson:2.10.1'
+    // Protobuf（DataChannel 控制指令二进制）
+    implementation 'com.google.protobuf:protobuf-java:3.25.1'
 }
--- a/WebRTCController/app/src/main/java/com/ttstd/controller/MainActivity.java
+++ b/WebRTCController/app/src/main/java/com/ttstd/controller/MainActivity.java
@@ -16,6 +16,8 @@ import androidx.appcompat.app.AppCompatActivity;

 import com.google.gson.Gson;
 import com.google.gson.JsonObject;
+import com.ttstd.control.Action;
+import com.ttstd.control.ControlMessage;
 import com.ttstd.controller.signaling.SignalMessage;
 import com.ttstd.controller.signaling.WebSocketClient;
 import com.ttstd.controller.view.RemoteTouchView;
@@ -326,55 +328,60 @@ public class MainActivity extends AppCompatActivity {

    private void sendTouchCommand(float relX, float relY) {
        if (webRtcClient != null) {
-            JsonObject cmd = new JsonObject();
-            cmd.addProperty("action", "TOUCH");
-            cmd.addProperty("x", relX);
-            cmd.addProperty("y", relY);
-            webRtcClient.sendControlCommand(cmd.toString());
+            ControlMessage msg = ControlMessage.newBuilder()
+                    .setAction(Action.TOUCH)
+                    .setX(relX)
+                    .setY(relY)
+                    .build();
+            webRtcClient.sendControlCommand(msg);
        }
    }

    private void sendSwipeCommand(float relX1, float relY1, float relX2, float relY2, long duration) {
        if (webRtcClient != null) {
-            JsonObject cmd = new JsonObject();
-            cmd.addProperty("action", "SWIPE");
-            cmd.addProperty("x1", relX1);
-            cmd.addProperty("y1", relY1);
-            cmd.addProperty("x2", relX2);
-            cmd.addProperty("y2", relY2);
-            cmd.addProperty("duration", duration);
-            webRtcClient.sendControlCommand(cmd.toString());
+            ControlMessage msg = ControlMessage.newBuilder()
+                    .setAction(Action.SWIPE)
+                    .setX1(relX1)
+                    .setY1(relY1)
+                    .setX2(relX2)
+                    .setY2(relY2)
+                    .setDuration(duration)
+                    .build();
+            webRtcClient.sendControlCommand(msg);
        }
    }

    private void sendKeyCommand(int keyCode, int action) {
        if (webRtcClient != null) {
-            JsonObject cmd = new JsonObject();
-            cmd.addProperty("action", "KEY");
-            cmd.addProperty("keyCode", keyCode);
-            cmd.addProperty("keyAction", action);
-            webRtcClient.sendControlCommand(cmd.toString());
+            ControlMessage msg = ControlMessage.newBuilder()
+                    .setAction(Action.KEY)
+                    .setKeyCode(keyCode)
+                    .setKeyAction(action)
+                    .build();
+            webRtcClient.sendControlCommand(msg);
        }
    }

    private void sendLongPressCommand(float relX, float relY) {
        if (webRtcClient != null) {
-            JsonObject cmd = new JsonObject();
-            cmd.addProperty("action", "LONG_PRESS");
-            cmd.addProperty("x", relX);
-            cmd.addProperty("y", relY);
-            webRtcClient.sendControlCommand(cmd.toString());
+            ControlMessage msg = ControlMessage.newBuilder()
+                    .setAction(Action.LONG_PRESS)
+                    .setX(relX)
+                    .setY(relY)
+                    .build();
+            webRtcClient.sendControlCommand(msg);
        }
    }

    private void sendMotionEventCommand(int action, float relX, float relY) {
        if (webRtcClient != null) {
-            JsonObject cmd = new JsonObject();
-            cmd.addProperty("action", "MOTION_EVENT");
-            cmd.addProperty("motionAction", action);
-            cmd.addProperty("x", relX);
-            cmd.addProperty("y", relY);
-            webRtcClient.sendControlCommand(cmd.toString());
+            ControlMessage msg = ControlMessage.newBuilder()
+                    .setAction(Action.MOTION_EVENT)
+                    .setMotionAction(action)
+                    .setX(relX)
+                    .setY(relY)
+                    .build();
+            webRtcClient.sendControlCommand(msg);
        }
    }

--- a/WebRTCController/app/src/main/java/com/ttstd/controller/view/RemoteTouchView.java
+++ b/WebRTCController/app/src/main/java/com/ttstd/controller/view/RemoteTouchView.java
@@ -9,8 +9,8 @@ import android.view.SurfaceView;

 import androidx.annotation.NonNull;

-import com.google.gson.Gson;
-import com.google.gson.JsonObject;
+import com.ttstd.control.Action;
+import com.ttstd.control.ControlMessage;

 /**
 * 自定义SurfaceView，用于捕获触摸事件并转换为远端输入指令。
@@ -20,7 +20,6 @@ import com.google.gson.JsonObject;
 public class RemoteTouchView extends SurfaceView {

    private static final String TAG = "RemoteTouchView";
-    private final Gson gson = new Gson();
    private TouchEventListener listener;

    private float lastTouchX;
@@ -144,39 +143,39 @@ public class RemoteTouchView extends SurfaceView {
        return super.onTouchEvent(event);
    }

-    public String createTouchCommand(float relX, float relY) {
-        JsonObject cmd = new JsonObject();
-        cmd.addProperty("action", "TOUCH");
-        cmd.addProperty("x", relX);
-        cmd.addProperty("y", relY);
-        return cmd.toString();
+    public ControlMessage createTouchCommand(float relX, float relY) {
+        return ControlMessage.newBuilder()
+                .setAction(Action.TOUCH)
+                .setX(relX)
+                .setY(relY)
+                .build();
    }

-    public String createSwipeCommand(float relX1, float relY1, float relX2, float relY2, long duration) {
-        JsonObject cmd = new JsonObject();
-        cmd.addProperty("action", "SWIPE");
-        cmd.addProperty("x1", relX1);
-        cmd.addProperty("y1", relY1);
-        cmd.addProperty("x2", relX2);
-        cmd.addProperty("y2", relY2);
-        cmd.addProperty("duration", duration);
-        return cmd.toString();
+    public ControlMessage createSwipeCommand(float relX1, float relY1, float relX2, float relY2, long duration) {
+        return ControlMessage.newBuilder()
+                .setAction(Action.SWIPE)
+                .setX1(relX1)
+                .setY1(relY1)
+                .setX2(relX2)
+                .setY2(relY2)
+                .setDuration(duration)
+                .build();
    }

-    public String createLongPressCommand(float relX, float relY) {
-        JsonObject cmd = new JsonObject();
-        cmd.addProperty("action", "LONG_PRESS");
-        cmd.addProperty("x", relX);
-        cmd.addProperty("y", relY);
-        return cmd.toString();
+    public ControlMessage createLongPressCommand(float relX, float relY) {
+        return ControlMessage.newBuilder()
+                .setAction(Action.LONG_PRESS)
+                .setX(relX)
+                .setY(relY)
+                .build();
    }

-    public String createMotionEventCommand(int action, float relX, float relY) {
-        JsonObject cmd = new JsonObject();
-        cmd.addProperty("action", "MOTION_EVENT");
-        cmd.addProperty("motionAction", action);
-        cmd.addProperty("x", relX);
-        cmd.addProperty("y", relY);
-        return cmd.toString();
+    public ControlMessage createMotionEventCommand(int action, float relX, float relY) {
+        return ControlMessage.newBuilder()
+                .setAction(Action.MOTION_EVENT)
+                .setMotionAction(action)
+                .setX(relX)
+                .setY(relY)
+                .build();
    }
 }
--- a/WebRTCController/app/src/main/java/com/ttstd/controller/webrtc/WebRtcClient.java
+++ b/WebRTCController/app/src/main/java/com/ttstd/controller/webrtc/WebRtcClient.java
@@ -7,6 +7,7 @@ import com.google.gson.Gson;
 import com.google.gson.JsonObject;
 import com.ttstd.controller.signaling.SignalMessage;
 import com.ttstd.controller.signaling.WebSocketClient;
+import com.ttstd.control.ControlMessage;

 import org.webrtc.DataChannel;
 import org.webrtc.DefaultVideoDecoderFactory;
@@ -27,7 +28,6 @@ import org.webrtc.VideoEncoderFactory;
 import org.webrtc.VideoTrack;

 import java.nio.ByteBuffer;
-import java.nio.charset.StandardCharsets;
 import java.util.ArrayList;
 import java.util.List;

@@ -231,10 +231,10 @@ public class WebRtcClient {
        }
    }

-    public void sendControlCommand(String commandJson) {
+    public void sendControlCommand(ControlMessage message) {
        if (dataChannel != null && dataChannel.state() == DataChannel.State.OPEN) {
            DataChannel.Buffer buffer = new DataChannel.Buffer(
-                    ByteBuffer.wrap(commandJson.getBytes(StandardCharsets.UTF_8)), false);
+                    ByteBuffer.wrap(message.toByteArray()), false);
            dataChannel.send(buffer);
        }
    }
@@ -279,8 +279,7 @@ public class WebRtcClient {
            public void onMessage(DataChannel.Buffer buffer) {
                byte[] data = new byte[buffer.data.remaining()];
                buffer.data.get(data);
-                String message = new String(data, StandardCharsets.UTF_8);
-                Log.d(TAG, "DataChannel message: " + message);
+                Log.d(TAG, "DataChannel message received, bytes=" + data.length);
            }
        });
    }
--- a/WebRTCController/app/src/main/proto/control_message.proto
+++ b/WebRTCController/app/src/main/proto/control_message.proto
@@ -0,0 +1,41 @@
+syntax = "proto3";
+
+package com.ttstd.control;
+
+option java_package = "com.ttstd.control";
+option java_multiple_files = true;
+
+// 控制指令类型，对应原 JSON 字段 action。
+enum Action {
+  ACTION_UNKNOWN = 0;
+  TOUCH = 1;
+  SWIPE = 2;
+  KEY = 3;
+  LONG_PRESS = 4;
+  MOTION_EVENT = 5;
+}
+
+// 通过 RTCDataChannel 传输的控制指令（protobuf 二进制）。
+// 坐标 x/y/x1/y1/x2/y2 均为相对屏幕的百分比，取值范围 0.0 ~ 1.0。
+message ControlMessage {
+  // 指令类型
+  Action action = 1;
+
+  // 单点坐标（TOUCH / LONG_PRESS / MOTION_EVENT）
+  double x = 2;
+  double y = 3;
+
+  // 滑动起止坐标（SWIPE）
+  double x1 = 4;
+  double y1 = 5;
+  double x2 = 6;
+  double y2 = 7;
+  int64 duration = 8;
+
+  // 按键（KEY）：key_action 0=按下 1=抬起（对应 Android KeyEvent ACTION_DOWN/UP）
+  int32 key_code = 9;
+  int32 key_action = 10;
+
+  // 原始动作（MOTION_EVENT）：对应 Android MotionEvent ACTION_*（0=DOWN 1=UP 2=MOVE）
+  int32 motion_action = 11;
+}
--- a/WebRTCController/build.gradle
+++ b/WebRTCController/build.gradle
@@ -1,5 +1,6 @@
 // Top-level build file
 plugins {
    id 'com.android.application' version '8.1.4' apply false
+    id 'com.google.protobuf' version '0.9.4' apply false
 }
 apply from: "config.gradle"
--- a/webrtc_controller_flutter/lib/controller/remote_controller.dart
+++ b/webrtc_controller_flutter/lib/controller/remote_controller.dart
@@ -4,6 +4,7 @@ import 'dart:convert';
 import 'package:flutter_webrtc/flutter_webrtc.dart';

 import '../models/signal_message.dart';
+import '../proto/control_message.pb.dart';
 import '../signaling/signaling_client.dart';
 import '../webrtc/webrtc_controller.dart';

@@ -113,9 +114,9 @@ class RemoteController {
    });
  }

-  /// 发送控制指令（JSON 字符串）。
-  void sendControlCommand(String commandJson) {
-    _webRtc?.sendControlCommand(commandJson);
+  /// 发送控制指令（protobuf 二进制）。
+  void sendControlCommand(ControlMessage command) {
+    _webRtc?.sendControlCommand(command);
  }

  /// 断开连接并释放资源。
--- a/webrtc_controller_flutter/lib/proto/control_message.pb.dart
+++ b/webrtc_controller_flutter/lib/proto/control_message.pb.dart
@@ -0,0 +1,209 @@
+// This is a generated file - do not edit.
+//
+// Generated from control_message.proto.
+
+// @dart = 3.3
+
+// ignore_for_file: annotate_overrides, camel_case_types, comment_references
+// ignore_for_file: constant_identifier_names
+// ignore_for_file: curly_braces_in_flow_control_structures
+// ignore_for_file: deprecated_member_use_from_same_package, library_prefixes
+// ignore_for_file: non_constant_identifier_names, prefer_relative_imports
+
+import 'dart:core' as $core;
+
+import 'package:fixnum/fixnum.dart' as $fixnum;
+import 'package:protobuf/protobuf.dart' as $pb;
+
+import 'control_message.pbenum.dart';
+
+export 'package:protobuf/protobuf.dart' show GeneratedMessageGenericExtensions;
+
+export 'control_message.pbenum.dart';
+
+/// 通过 RTCDataChannel 传输的控制指令（protobuf 二进制）。
+/// 坐标 x/y/x1/y1/x2/y2 均为相对屏幕的百分比，取值范围 0.0 ~ 1.0。
+class ControlMessage extends $pb.GeneratedMessage {
+  factory ControlMessage({
+    Action? action,
+    $core.double? x,
+    $core.double? y,
+    $core.double? x1,
+    $core.double? y1,
+    $core.double? x2,
+    $core.double? y2,
+    $fixnum.Int64? duration,
+    $core.int? keyCode,
+    $core.int? keyAction,
+    $core.int? motionAction,
+  }) {
+    final result = create();
+    if (action != null) result.action = action;
+    if (x != null) result.x = x;
+    if (y != null) result.y = y;
+    if (x1 != null) result.x1 = x1;
+    if (y1 != null) result.y1 = y1;
+    if (x2 != null) result.x2 = x2;
+    if (y2 != null) result.y2 = y2;
+    if (duration != null) result.duration = duration;
+    if (keyCode != null) result.keyCode = keyCode;
+    if (keyAction != null) result.keyAction = keyAction;
+    if (motionAction != null) result.motionAction = motionAction;
+    return result;
+  }
+
+  ControlMessage._();
+
+  factory ControlMessage.fromBuffer($core.List<$core.int> data,
+          [$pb.ExtensionRegistry registry = $pb.ExtensionRegistry.EMPTY]) =>
+      create()..mergeFromBuffer(data, registry);
+  factory ControlMessage.fromJson($core.String json,
+          [$pb.ExtensionRegistry registry = $pb.ExtensionRegistry.EMPTY]) =>
+      create()..mergeFromJson(json, registry);
+
+  static final $pb.BuilderInfo _i = $pb.BuilderInfo(
+      _omitMessageNames ? '' : 'ControlMessage',
+      package:
+          const $pb.PackageName(_omitMessageNames ? '' : 'com.ttstd.control'),
+      createEmptyInstance: create)
+    ..aE<Action>(1, _omitFieldNames ? '' : 'action', enumValues: Action.values)
+    ..aD(2, _omitFieldNames ? '' : 'x')
+    ..aD(3, _omitFieldNames ? '' : 'y')
+    ..aD(4, _omitFieldNames ? '' : 'x1')
+    ..aD(5, _omitFieldNames ? '' : 'y1')
+    ..aD(6, _omitFieldNames ? '' : 'x2')
+    ..aD(7, _omitFieldNames ? '' : 'y2')
+    ..aInt64(8, _omitFieldNames ? '' : 'duration')
+    ..aI(9, _omitFieldNames ? '' : 'keyCode')
+    ..aI(10, _omitFieldNames ? '' : 'keyAction')
+    ..aI(11, _omitFieldNames ? '' : 'motionAction')
+    ..hasRequiredFields = false;
+
+  @$core.Deprecated('See https://github.com/google/protobuf.dart/issues/998.')
+  ControlMessage clone() => deepCopy();
+  @$core.Deprecated('See https://github.com/google/protobuf.dart/issues/998.')
+  ControlMessage copyWith(void Function(ControlMessage) updates) =>
+      super.copyWith((message) => updates(message as ControlMessage))
+          as ControlMessage;
+
+  @$core.override
+  $pb.BuilderInfo get info_ => _i;
+
+  @$core.pragma('dart2js:noInline')
+  static ControlMessage create() => ControlMessage._();
+  @$core.override
+  ControlMessage createEmptyInstance() => create();
+  @$core.pragma('dart2js:noInline')
+  static ControlMessage getDefault() => _defaultInstance ??=
+      $pb.GeneratedMessage.$_defaultFor<ControlMessage>(create);
+  static ControlMessage? _defaultInstance;
+
+  /// 指令类型
+  @$pb.TagNumber(1)
+  Action get action => $_getN(0);
+  @$pb.TagNumber(1)
+  set action(Action value) => $_setField(1, value);
+  @$pb.TagNumber(1)
+  $core.bool hasAction() => $_has(0);
+  @$pb.TagNumber(1)
+  void clearAction() => $_clearField(1);
+
+  /// 单点坐标（TOUCH / LONG_PRESS / MOTION_EVENT）
+  @$pb.TagNumber(2)
+  $core.double get x => $_getN(1);
+  @$pb.TagNumber(2)
+  set x($core.double value) => $_setDouble(1, value);
+  @$pb.TagNumber(2)
+  $core.bool hasX() => $_has(1);
+  @$pb.TagNumber(2)
+  void clearX() => $_clearField(2);
+
+  @$pb.TagNumber(3)
+  $core.double get y => $_getN(2);
+  @$pb.TagNumber(3)
+  set y($core.double value) => $_setDouble(2, value);
+  @$pb.TagNumber(3)
+  $core.bool hasY() => $_has(2);
+  @$pb.TagNumber(3)
+  void clearY() => $_clearField(3);
+
+  /// 滑动起止坐标（SWIPE）
+  @$pb.TagNumber(4)
+  $core.double get x1 => $_getN(3);
+  @$pb.TagNumber(4)
+  set x1($core.double value) => $_setDouble(3, value);
+  @$pb.TagNumber(4)
+  $core.bool hasX1() => $_has(3);
+  @$pb.TagNumber(4)
+  void clearX1() => $_clearField(4);
+
+  @$pb.TagNumber(5)
+  $core.double get y1 => $_getN(4);
+  @$pb.TagNumber(5)
+  set y1($core.double value) => $_setDouble(4, value);
+  @$pb.TagNumber(5)
+  $core.bool hasY1() => $_has(4);
+  @$pb.TagNumber(5)
+  void clearY1() => $_clearField(5);
+
+  @$pb.TagNumber(6)
+  $core.double get x2 => $_getN(5);
+  @$pb.TagNumber(6)
+  set x2($core.double value) => $_setDouble(5, value);
+  @$pb.TagNumber(6)
+  $core.bool hasX2() => $_has(5);
+  @$pb.TagNumber(6)
+  void clearX2() => $_clearField(6);
+
+  @$pb.TagNumber(7)
+  $core.double get y2 => $_getN(6);
+  @$pb.TagNumber(7)
+  set y2($core.double value) => $_setDouble(6, value);
+  @$pb.TagNumber(7)
+  $core.bool hasY2() => $_has(6);
+  @$pb.TagNumber(7)
+  void clearY2() => $_clearField(7);
+
+  @$pb.TagNumber(8)
+  $fixnum.Int64 get duration => $_getI64(7);
+  @$pb.TagNumber(8)
+  set duration($fixnum.Int64 value) => $_setInt64(7, value);
+  @$pb.TagNumber(8)
+  $core.bool hasDuration() => $_has(7);
+  @$pb.TagNumber(8)
+  void clearDuration() => $_clearField(8);
+
+  /// 按键（KEY）：key_action 0=按下 1=抬起（对应 Android KeyEvent ACTION_DOWN/UP）
+  @$pb.TagNumber(9)
+  $core.int get keyCode => $_getIZ(8);
+  @$pb.TagNumber(9)
+  set keyCode($core.int value) => $_setSignedInt32(8, value);
+  @$pb.TagNumber(9)
+  $core.bool hasKeyCode() => $_has(8);
+  @$pb.TagNumber(9)
+  void clearKeyCode() => $_clearField(9);
+
+  @$pb.TagNumber(10)
+  $core.int get keyAction => $_getIZ(9);
+  @$pb.TagNumber(10)
+  set keyAction($core.int value) => $_setSignedInt32(9, value);
+  @$pb.TagNumber(10)
+  $core.bool hasKeyAction() => $_has(9);
+  @$pb.TagNumber(10)
+  void clearKeyAction() => $_clearField(10);
+
+  /// 原始动作（MOTION_EVENT）：对应 Android MotionEvent ACTION_*（0=DOWN 1=UP 2=MOVE）
+  @$pb.TagNumber(11)
+  $core.int get motionAction => $_getIZ(10);
+  @$pb.TagNumber(11)
+  set motionAction($core.int value) => $_setSignedInt32(10, value);
+  @$pb.TagNumber(11)
+  $core.bool hasMotionAction() => $_has(10);
+  @$pb.TagNumber(11)
+  void clearMotionAction() => $_clearField(11);
+}
+
+const $core.bool _omitFieldNames =
+    $core.bool.fromEnvironment('protobuf.omit_field_names');
+const $core.bool _omitMessageNames =
+    $core.bool.fromEnvironment('protobuf.omit_message_names');
--- a/webrtc_controller_flutter/lib/proto/control_message.pbenum.dart
+++ b/webrtc_controller_flutter/lib/proto/control_message.pbenum.dart
@@ -0,0 +1,47 @@
+// This is a generated file - do not edit.
+//
+// Generated from control_message.proto.
+
+// @dart = 3.3
+
+// ignore_for_file: annotate_overrides, camel_case_types, comment_references
+// ignore_for_file: constant_identifier_names
+// ignore_for_file: curly_braces_in_flow_control_structures
+// ignore_for_file: deprecated_member_use_from_same_package, library_prefixes
+// ignore_for_file: non_constant_identifier_names, prefer_relative_imports
+
+import 'dart:core' as $core;
+
+import 'package:protobuf/protobuf.dart' as $pb;
+
+/// 控制指令类型，对应原 JSON 字段 action。
+class Action extends $pb.ProtobufEnum {
+  static const Action ACTION_UNKNOWN =
+      Action._(0, _omitEnumNames ? '' : 'ACTION_UNKNOWN');
+  static const Action TOUCH = Action._(1, _omitEnumNames ? '' : 'TOUCH');
+  static const Action SWIPE = Action._(2, _omitEnumNames ? '' : 'SWIPE');
+  static const Action KEY = Action._(3, _omitEnumNames ? '' : 'KEY');
+  static const Action LONG_PRESS =
+      Action._(4, _omitEnumNames ? '' : 'LONG_PRESS');
+  static const Action MOTION_EVENT =
+      Action._(5, _omitEnumNames ? '' : 'MOTION_EVENT');
+
+  static const $core.List<Action> values = <Action>[
+    ACTION_UNKNOWN,
+    TOUCH,
+    SWIPE,
+    KEY,
+    LONG_PRESS,
+    MOTION_EVENT,
+  ];
+
+  static final $core.List<Action?> _byValue =
+      $pb.ProtobufEnum.$_initByValueList(values, 5);
+  static Action? valueOf($core.int value) =>
+      value < 0 || value >= _byValue.length ? null : _byValue[value];
+
+  const Action._(super.value, super.name);
+}
+
+const $core.bool _omitEnumNames =
+    $core.bool.fromEnvironment('protobuf.omit_enum_names');
--- a/webrtc_controller_flutter/lib/proto/control_message.pbjson.dart
+++ b/webrtc_controller_flutter/lib/proto/control_message.pbjson.dart
@@ -0,0 +1,68 @@
+// This is a generated file - do not edit.
+//
+// Generated from control_message.proto.
+
+// @dart = 3.3
+
+// ignore_for_file: annotate_overrides, camel_case_types, comment_references
+// ignore_for_file: constant_identifier_names
+// ignore_for_file: curly_braces_in_flow_control_structures
+// ignore_for_file: deprecated_member_use_from_same_package, library_prefixes
+// ignore_for_file: non_constant_identifier_names, prefer_relative_imports
+// ignore_for_file: unused_import
+
+import 'dart:convert' as $convert;
+import 'dart:core' as $core;
+import 'dart:typed_data' as $typed_data;
+
+@$core.Deprecated('Use actionDescriptor instead')
+const Action$json = {
+  '1': 'Action',
+  '2': [
+    {'1': 'ACTION_UNKNOWN', '2': 0},
+    {'1': 'TOUCH', '2': 1},
+    {'1': 'SWIPE', '2': 2},
+    {'1': 'KEY', '2': 3},
+    {'1': 'LONG_PRESS', '2': 4},
+    {'1': 'MOTION_EVENT', '2': 5},
+  ],
+};
+
+/// Descriptor for `Action`. Decode as a `google.protobuf.EnumDescriptorProto`.
+final $typed_data.Uint8List actionDescriptor = $convert.base64Decode(
+    'CgZBY3Rpb24SEgoOQUNUSU9OX1VOS05PV04QABIJCgVUT1VDSBABEgkKBVNXSVBFEAISBwoDS0'
+    'VZEAMSDgoKTE9OR19QUkVTUxAEEhAKDE1PVElPTl9FVkVOVBAF');
+
+@$core.Deprecated('Use controlMessageDescriptor instead')
+const ControlMessage$json = {
+  '1': 'ControlMessage',
+  '2': [
+    {
+      '1': 'action',
+      '3': 1,
+      '4': 1,
+      '5': 14,
+      '6': '.com.ttstd.control.Action',
+      '10': 'action'
+    },
+    {'1': 'x', '3': 2, '4': 1, '5': 1, '10': 'x'},
+    {'1': 'y', '3': 3, '4': 1, '5': 1, '10': 'y'},
+    {'1': 'x1', '3': 4, '4': 1, '5': 1, '10': 'x1'},
+    {'1': 'y1', '3': 5, '4': 1, '5': 1, '10': 'y1'},
+    {'1': 'x2', '3': 6, '4': 1, '5': 1, '10': 'x2'},
+    {'1': 'y2', '3': 7, '4': 1, '5': 1, '10': 'y2'},
+    {'1': 'duration', '3': 8, '4': 1, '5': 3, '10': 'duration'},
+    {'1': 'key_code', '3': 9, '4': 1, '5': 5, '10': 'keyCode'},
+    {'1': 'key_action', '3': 10, '4': 1, '5': 5, '10': 'keyAction'},
+    {'1': 'motion_action', '3': 11, '4': 1, '5': 5, '10': 'motionAction'},
+  ],
+};
+
+/// Descriptor for `ControlMessage`. Decode as a `google.protobuf.DescriptorProto`.
+final $typed_data.Uint8List controlMessageDescriptor = $convert.base64Decode(
+    'Cg5Db250cm9sTWVzc2FnZRIxCgZhY3Rpb24YASABKA4yGS5jb20udHRzdGQuY29udHJvbC5BY3'
+    'Rpb25SBmFjdGlvbhIMCgF4GAIgASgBUgF4EgwKAXkYAyABKAFSAXkSDgoCeDEYBCABKAFSAngx'
+    'Eg4KAnkxGAUgASgBUgJ5MRIOCgJ4MhgGIAEoAVICeDISDgoCeTIYByABKAFSAnkyEhoKCGR1cm'
+    'F0aW9uGAggASgDUghkdXJhdGlvbhIZCghrZXlfY29kZRgJIAEoBVIHa2V5Q29kZRIdCgprZXlf'
+    'YWN0aW9uGAogASgFUglrZXlBY3Rpb24SIwoNbW90aW9uX2FjdGlvbhgLIAEoBVIMbW90aW9uQW'
+    'N0aW9u');
--- a/webrtc_controller_flutter/lib/utils/control_commands.dart
+++ b/webrtc_controller_flutter/lib/utils/control_commands.dart
@@ -1,57 +1,59 @@
-import 'dart:convert';
+import 'package:fixnum/fixnum.dart';
+import 'package:webrtc_controller_flutter/proto/control_message.pb.dart';

 /// 控制指令构造工具，对应 Android 端 RemoteTouchView 的指令格式。
 ///
-/// 指令通过 DataChannel 以 JSON 字符串发送，被控端（WebRTCControlled）
-/// 的 InputCommandHandler 负责解析执行。
+/// 指令通过 DataChannel 以 protobuf 二进制（[ControlMessage]）发送，
+/// 被控端（WebRTCControlled）的 InputCommandHandler 负责解析执行。
 class ControlCommands {
  ControlCommands._();

  /// 单击指令。坐标 x/y 为相对于屏幕的百分比（0.0 ~ 1.0）。
-  static String touch(double x, double y) => jsonEncode({
-        'action': 'TOUCH',
-        'x': x,
-        'y': y,
-      });
+  static ControlMessage touch(double x, double y) => ControlMessage(
+        action: Action.TOUCH,
+        x: x,
+        y: y,
+      );

  /// 滑动指令。坐标为相对百分比，duration 为毫秒。
-  static String swipe(
+  static ControlMessage swipe(
    double x1,
    double y1,
    double x2,
    double y2,
    int durationMs,
  ) =>
-      jsonEncode({
-        'action': 'SWIPE',
-        'x1': x1,
-        'y1': y1,
-        'x2': x2,
-        'y2': y2,
-        'duration': durationMs,
-      });
+      ControlMessage(
+        action: Action.SWIPE,
+        x1: x1,
+        y1: y1,
+        x2: x2,
+        y2: y2,
+        duration: Int64(durationMs),
+      );

  /// 按键指令。action: 0=按下, 1=抬起（对应 Android KeyEvent.ACTION_DOWN/UP）。
-  static String key(int keyCode, int action) => jsonEncode({
-        'action': 'KEY',
-        'keyCode': keyCode,
-        'keyAction': action,
-      });
+  static ControlMessage key(int keyCode, int action) => ControlMessage(
+        action: Action.KEY,
+        keyCode: keyCode,
+        keyAction: action,
+      );

  /// 长按指令。坐标 x/y 为相对于屏幕的百分比（0.0 ~ 1.0）。
  /// 对应 Android 端 RemoteTouchView.createLongPressCommand。
-  static String longPress(double x, double y) => jsonEncode({
-        'action': 'LONG_PRESS',
-        'x': x,
-        'y': y,
-      });
+  static ControlMessage longPress(double x, double y) => ControlMessage(
+        action: Action.LONG_PRESS,
+        x: x,
+        y: y,
+      );

  /// 原始 MotionEvent 指令，用于实现“实时跟手”。
  /// action: 0=DOWN, 1=UP, 2=MOVE（对应 Android MotionEvent.ACTION_XXX）。
-  static String motionEvent(int action, double x, double y) => jsonEncode({
-        'action': 'MOTION_EVENT',
-        'motionAction': action,
-        'x': x,
-        'y': y,
-      });
+  static ControlMessage motionEvent(int action, double x, double y) =>
+      ControlMessage(
+        action: Action.MOTION_EVENT,
+        motionAction: action,
+        x: x,
+        y: y,
+      );
 }
--- a/webrtc_controller_flutter/lib/webrtc/webrtc_controller.dart
+++ b/webrtc_controller_flutter/lib/webrtc/webrtc_controller.dart
@@ -3,6 +3,7 @@ import 'package:flutter_webrtc/flutter_webrtc.dart';

 import '../config/ice_servers.dart';
 import '../models/signal_message.dart';
+import '../proto/control_message.pb.dart';
 import '../signaling/signaling_client.dart';

 /// 封装 WebRTC 连接逻辑（对应 Android 端 WebRtcClient）。
@@ -105,8 +106,19 @@ class WebRtcController {
  void _setupDataChannel(RTCDataChannel channel) {
    channel.onMessage = (RTCDataChannelMessage message) {
      // 处理来自被控端的消息（本控制端主要发送，此处仅做日志记录）。
+      if (message.isBinary) {
+        try {
+          final command = ControlMessage.fromBuffer(message.binary);
          // ignore: avoid_print
-      print('DataChannel message: ${message.text}');
+          print('DataChannel message: ${command.action}');
+        } catch (e) {
+          // ignore: avoid_print
+          print('DataChannel message decode failed: $e');
+        }
+      } else {
+        // ignore: avoid_print
+        print('DataChannel text message: ${message.text}');
+      }
    };
    channel.onDataChannelState = (RTCDataChannelState state) {
      // ignore: avoid_print
@@ -171,11 +183,11 @@ class WebRtcController {
    await _pc?.addCandidate(candidate);
  }

-  /// 通过 DataChannel 发送控制指令（JSON 字符串）。
-  void sendControlCommand(String commandJson) {
+  /// 通过 DataChannel 发送控制指令（protobuf 二进制）。
+  void sendControlCommand(ControlMessage command) {
    if (_dataChannel?.state == RTCDataChannelState.RTCDataChannelOpen) {
-      debugPrint('WebRtcController: Sending command -> $commandJson');
-      _dataChannel!.send(RTCDataChannelMessage(commandJson));
+      debugPrint('WebRtcController: Sending command -> ${command.action}');
+      _dataChannel!.send(RTCDataChannelMessage.fromBinary(command.writeToBuffer()));
    }
  }

--- a/webrtc_controller_flutter/proto/control_message.proto
+++ b/webrtc_controller_flutter/proto/control_message.proto
@@ -0,0 +1,41 @@
+syntax = "proto3";
+
+package com.ttstd.control;
+
+option java_package = "com.ttstd.control";
+option java_multiple_files = true;
+
+// 控制指令类型，对应原 JSON 字段 action。
+enum Action {
+  ACTION_UNKNOWN = 0;
+  TOUCH = 1;
+  SWIPE = 2;
+  KEY = 3;
+  LONG_PRESS = 4;
+  MOTION_EVENT = 5;
+}
+
+// 通过 RTCDataChannel 传输的控制指令（protobuf 二进制）。
+// 坐标 x/y/x1/y1/x2/y2 均为相对屏幕的百分比，取值范围 0.0 ~ 1.0。
+message ControlMessage {
+  // 指令类型
+  Action action = 1;
+
+  // 单点坐标（TOUCH / LONG_PRESS / MOTION_EVENT）
+  double x = 2;
+  double y = 3;
+
+  // 滑动起止坐标（SWIPE）
+  double x1 = 4;
+  double y1 = 5;
+  double x2 = 6;
+  double y2 = 7;
+  int64 duration = 8;
+
+  // 按键（KEY）：key_action 0=按下 1=抬起（对应 Android KeyEvent ACTION_DOWN/UP）
+  int32 key_code = 9;
+  int32 key_action = 10;
+
+  // 原始动作（MOTION_EVENT）：对应 Android MotionEvent ACTION_*（0=DOWN 1=UP 2=MOVE）
+  int32 motion_action = 11;
+}
--- a/webrtc_controller_flutter/pubspec.lock
+++ b/webrtc_controller_flutter/pubspec.lock
@@ -98,7 +98,7 @@ packages:
    source: hosted
    version: "2.2.0"
  fixnum:
-    dependency: transitive
+    dependency: "direct main"
    description:
      name: fixnum
      sha256: b6dc7065e46c974bc7c5f143080a6764ec7a4be6da1285ececdc37be96de53be
@@ -323,6 +323,14 @@ packages:
      url: "https://pub.flutter-io.cn"
    source: hosted
    version: "2.1.8"
+  protobuf:
+    dependency: "direct main"
+    description:
+      name: protobuf
+      sha256: "75ec242d22e950bdcc79ee38dd520ce4ee0bc491d7fadc4ea47694604d22bf06"
+      url: "https://pub.flutter-io.cn"
+    source: hosted
+    version: "6.0.0"
  pub_semver:
    dependency: transitive
    description:
--- a/webrtc_controller_flutter/pubspec.yaml
+++ b/webrtc_controller_flutter/pubspec.yaml
@@ -43,6 +43,8 @@ dependencies:

  # 设备 ID 生成
  uuid: ^4.4.0
+  protobuf: ^6.0.0
+  fixnum: ^1.1.1

 dev_dependencies:
  flutter_test: