# Investigating Traffic Analysis Attacks on Apple iCloud Private Relay

## Publications: Ali Zohaib, Jade Sheffey, and Amir Houmansadr. Investigating Traffic Analysis Attacks on Apple iCloud Private Relay. In Proc. of ACM ASIA Conference on Computer and Communications Security 2023.

## Trace overview

We release our dataset used for our website fingerprinting and flow correlation experiments to benefit future works on traffic analysis, available at [UMass Trace Repository](https://traces.cs.umass.edu/)

## Layout

Layout of the folder:

```
.
├── flow-correlation
│ └── pcaps
│     ├── Here
│     └── There
└── website-fingerprinting
    ├── monitored
    └── unmonitored
```

## Directory discriptions

* `flow-correlation`, `website-fingerprinting` each contains the generated traces and collected traffic dataset of the corresponding experiments. All .pcaps are parsed as .csv files containing the relevant fields as columns.
  * In the `flow-correlation` directory, the `Here` and `There` subdirectories correspond to the client-side and server-side collection for the flows respectively. The `flows.json` file contains the combined flows.
  * In the `website-fingerprinting` directory, `monitored` traces are present under a numerically labelled subdirectory for the websites in the monitored set. The `unmonitored` traces are single instances also named numerically in the respective folder.

## Models
The following are the original implementations used for the experiments in the paper. Please refer to these links for downloading the respective codes:

* DeepCorr: <https://github.com/SPIN-UMass/DeepCorr>
* DeepFingerprinting: <https://github.com/deep-fingerprinting/df>
* Var-CNN: <https://github.com/sanjit-bhat/Var-CNN>


## Contact

Please do not hesitate to contact Ali Zohaib via azohaib at cs.umass.edu if you need any assistance with using these trace files.