mobilegpueavesdropping: Update info

pittisl · Aug 31, 2023 · f1aa576 · f1aa576
1 parent 04a105f
commit f1aa576
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 0 deletions.
diff --git a/assets/media/2022-mobile-gpu-eavesdropping/mobilegpu-ioctl-call-map.png b/assets/media/2022-mobile-gpu-eavesdropping/mobilegpu-ioctl-call-map.png
diff --git a/assets/media/2022-mobile-gpu-eavesdropping/mobilegpu-result-keyboard.png b/assets/media/2022-mobile-gpu-eavesdropping/mobilegpu-result-keyboard.png
diff --git a/assets/media/2022-mobile-gpu-eavesdropping/mobilegpu-targeted-app.png b/assets/media/2022-mobile-gpu-eavesdropping/mobilegpu-targeted-app.png
diff --git a/content/publication/2022-mobile-gpu-eavesdropping/index.md b/content/publication/2022-mobile-gpu-eavesdropping/index.md
@@ -80,8 +80,28 @@ with certain GPU performance counter value changes.
 
 ### System Overview
 
+Our work designs the whole attack with an offline training phase and online attacking phase.
+In the offline phase, the attacker emulates all key
+presses over different device models and configurations to collect
+a sufficient amount of GPU PC (Performance Counter) data.
+In the online phase, the attacking application will spawn a moni-
+toring process, which runs as an Android service in background.
+If a target application is launched, the monitoring process will start
+reading the selected GPU PCs. These readings
+will be first used to recognize the current device model and configuration,
+and then applied to the corresponding classification model
+for eavesdropping.
+
 ![Eavesdropping System Overview](2022-mobile-gpu-eavesdropping/system-overview.png)
 
+### Reading GPU PC information from OS
+
+We make use of customized `ioctl()` system call on GPU device file in the Android
+OS in order to circumvent privilege limitation in system OpenGL library and directly
+query system GPU performance counter values.
+
+![Utilizing GPU device file in Android OS](2022-mobile-gpu-eavesdropping/mobilegpu-ioctl-call-map.png)
+
 ### Handling Noises and System Factors
 
 To reduce impact from noises and system factors, we use a distance-based
@@ -98,6 +118,13 @@ For random text string inputs, the average inference accuracy is 81.3%.
 
 ![Accuracy for Individual Key Presses](2022-mobile-gpu-eavesdropping/mobilegpu-overall-accuracy.png)
 
+The inference accuracy is shown to be not significantly affected by
+the choice of targeted App or the on-screen keyboard software.
+
+![Accuracy with different targeted App](2022-mobile-gpu-eavesdropping/mobilegpu-targeted-app.png)
+
+![Accuracy with different on-screen keyboard](2022-mobile-gpu-eavesdropping/mobilegpu-result-keyboard.png)
+
 ### Follow-up Security Fixes
 
 The vulnerabilities found in this paper is designated with CVE number