Awesome

Modified Wehe Command Line Client for MobileAtlas

This repository contains a modified version of the Wehe command-line client used to detect economic differentiation in mobile networks via traffic replaying using the MobileAtlas measurement platform.

A brief overview of both the Wehe and MobileAtlas projects can be found in the repository of the modified Wehe server.

Changes to the Client

Removal of Analysis Retrieval Logic

Since we are interested in economic differentiation and not in the throttling of network traffic, we do not use the analysis server provided by Wehe and thus removed the logic responsible for requesting/retrieving results from it.

Removing Randomised Test Runs

Wehe compares measured throughput information between two replays: One where recorded traffic is replayed unaltered and one where payloads are randomised. However, for our work on economic differentiation, we do not need these randomised replays since we only need to retrieve billing information from the MNO.

Support for multiple TCP and UDP conversations per test

Initially, Wehe supported only one TCP/UDP conversation per Test. Our changes make it possible to create tests from PCAP files that contain many TCP and UDP streams and thus allow tests to better reflect the behaviour of the actual applications.

Removal of Websocket Dependency/M-Lab

Since we do not use M-Lab for our tests we were able to remove the dependency on the Tyrus Websocket library which was only used in supporting M-Lab.

Adding new Tests

New tests can be created by following these steps:

1. Follow the instructions found in the wehe-server repository to generate the necessary files.
2. Move the generated file called <testname>.client_all.json into the res/ folder
3. Add a new entry in the apps_list.json file:

   {
     "name": "<testname>",
     "time": <time before test times out in seconds>,
     "image": "<testname>",
     "category": "",
     "randomdatafile": "",
     "datafile": "<testname>.pcap_client_all.json"
   }
   

About the Code

• The code is based off of the Wehe Android Client.
• Building the source code requires an additional library: JSON (Maven repo).
• The src/ directory contains the source files.
• The res/ directory contains the app and port client traffic for the tests.
• The code runs one test for each shell command.

Running the client

There are two ways to run the code:

1. Run the jar file (wehe-cmdline.jar)
2. Build from source (use Java 11)
3. Use docker to build a container image

A compiled jar comes with the repo (wehe-cmdline.jar). This jar was compiled using Java 11.

Usage

Usage: java -jar wehe-cmdline.jar -s <SERVER_IPS> -n <TEST_NAME> [OPTION]...

Example: java -jar wehe-cmdline.jar -s 192.0.2.15,2001:db8::15 -n snapchat_misc -r results/ -l info

Options

-n TEST_NAME - The name of the test to run. Argument required. See below for list of test names.

-s SERV_NAME - The hostname or IP of server to run the tests. Default: wehe4.meddle.mobi.

-t RESR_ROOT - The resources root containing apps_list.json and the tests. No need to change this if using the jar or if you don't move the res/ directory. Default: res/.

-r RSLT_ROOT - The results root containing the output logs and info. Default: test_results/.

-l LOG_LEVEL - The level of logs and above that should be printed to console (all levels will be saved to logs on disk regardless of the level printed to console). Choose from wtf, error, warn, info, or debug. Default: none of these, only UI logs (what the user would see on screen when running the Android app) will be printed to the console.

-h - Print the help message.

-v - Print the version number.

Provided Tests

Output

Output is contained in RSLT_ROOT:

• By default, the name of RSLT_ROOT is res/ (change the directory name using the -r option).
• info.txt - This file contains the user's random ID and the current history count.
• logs/ - This directory contains the logs that would be printed to the Android logs (think of Android's Log class).
  • Log files are the in the format logs_<randomID>_<historyCount>_<exitCode>.txt.
• ui/ - This directory contains the text that a user running the Android client would see on his/her screen.
  • Log files are in the format ui_<randomID>_<historyCount>_<exitCode>.txt.

Exit Codes

Each error exit code corresponds to a specific spot in the code where execution of the tests was terminated. CTRL+F the exit status on the code base to find where each of these locations are located.

NOTE: If the Exit Code is negative, then two errors occurred: the number in the Exit Code and ERR_WR_LOGS. For example, -24 means no result was found and the logs could not be written to disk.

Funding

This project is funded through the NGI0 PET Fund, a fund established by NLnet with financial support from the European Commission's Next Generation Internet programme, under the aegis of DG Communications Networks, Content and Technology under grant agreement No 825310.