# Overview

The files: 

    sha1sum                                   filename
    b330ba5b72b194bb172f1ae6665c716bb93f43cc  cellular_localization_gpx.tar.xz
    aa23fffaa555bd9d1314faff73c630b28c055796  cellular_localization_pcap_A.tar.xz
    ee6b4f83930c8e1a189c2a773faccbaa9de792f2  cellular_localization_pcap_B.tar.xz
    c8737e8198e7ece133994ffcf39da1d3ce011773  cellular_localization_pcap_C.tar.xz
    0a15c4aecece2737b9c9d43192ad24a3e3fe95aa  cellular_localization_pcap_D.tar.xz

as referenced in the following paper:

    @inproceedings{Soroush:2013,
        Author = { Hamed Soroush and Keen Sung and Erik Learned-Miller and Brian Neil Levine and Marc Liberatore},
        Booktitle = {Proc. Privacy Enhancing Technologies Symposium (PETS)},
        Keywords = {Privacy; Cellular; Remote Inference},
        Month = {July},
        Sponsors = {CNS-0905349},
        Title = {{Disabling GPS is Not Enough: Cellular location leaks over the Internet}},
        Year = {2013}
    }

contain a collection of trimmed TCP traces streamed to phones as
they travelled along given paths, and of location data and signal
strength as observed on the phones. The traces were collected between
April and November 2012 along four paths around Amherst, MA.

If you use these data, please cite the above paper.

# Structure

As described in the paper, two files were collected for each
measurement:

 - `.pcap` files contain the TCP data as seen on the server-side
   during the measurement. The files are filtered by IP and port (no
   extraneous packets are present), and each packet is truncated
   at 96 bytes (the `-s 96` option was passed to `tcpdump`).

 - `.gpx` files contain the GPS and signal data for each trace, and
   were recorded on the phone.

Files are named are in the following format:

    {origin}2{destination}-{yyyymmdd}[-tracenum].{phoneID}
    
 - {origin} and {destination} correspond to locations in the paper
   as follows:
   
    A = Sunderland (sunderland)
    B = Northampton (noho)
    C = Belchertown (belch)
    D = Orange (airport)
    X = Amherst (lab|umass)    
    
 - {yyyymmdd} is the date the trace was collected,

 - [-tracenum] is an optional field, disambiguating multiple same-day
   traces (and occasionally omitted when unneeded), and
 
 - {phoneid} is a unique identifier corresponding to a given phone (it
   is usually the last quad of the phone's IP address).

All `gpx` files are in a single compressed tarball.

The `pcap` files are split into separate compressed tarballs, based
upon the non-Amherst endpoint.

Downloading and unpacking all the tarballs will result in the
following file structure:

    cellular_localization_traces/
      pcap/[non-Amherst endpoint]/{filename}.pcap
      gpx/[non-Amherst endpoint]/{filename}.gpx


# History

First written by Keen Sung <ksung@cs.umass.edu> on 2013-05-02.

Minor updates by Marc Liberatore <liberato@cs.umass.edu> on 2013-05-02.