01.01.2015 Views

Proceedings [PDF] - Measurement and Analysis of P2P Activity ...

Proceedings [PDF] - Measurement and Analysis of P2P Activity ...

Proceedings [PDF] - Measurement and Analysis of P2P Activity ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

International Conference Advances in the <strong>Analysis</strong> <strong>of</strong> Online Paedophile <strong>Activity</strong> Paris, France; 2-3 June, 2009<br />

3.2 Number <strong>of</strong> paedophile files<br />

Figure 2 presents, the evolution <strong>of</strong> the number <strong>of</strong> distincts<br />

paedophile file-id (vertical axis) observed during<br />

our measurements as a function <strong>of</strong> time (horizontal axis)<br />

represented by the number <strong>of</strong> sessions. We assumed<br />

that a file is a paedophile one if its name contains at<br />

least one paedophile keyword.<br />

4. ONGOING AND FUTURE WORK<br />

We are planning to perform a new experiment using<br />

multiple distributed clients using the PlatnetLab platform<br />

<strong>and</strong> with one session. These measures will give<br />

partial views <strong>of</strong> the network. We plan to use them<br />

to estimate the real value <strong>of</strong> several parameters such<br />

as : the number <strong>of</strong> paedophile files <strong>and</strong> the number<br />

<strong>of</strong> copies <strong>of</strong> a particular paedophile file. We will use<br />

the multiple-recapture model [3] for these estimations.<br />

The multiple-recapture model is used to estimate the<br />

unknown size <strong>of</strong> a population using multiple samples.<br />

Until now, we have installed our client on 94 PlanetLab<br />

machines. We have collected measures simultaneously<br />

from all these machines. These collected measures are<br />

under study. The next step is to analyse the results<br />

in order to determine if there is a positive or negative<br />

dependency between the results obtained by each client<br />

<strong>and</strong> to determine if the results are homogeneous.<br />

Figure 2: Evolution <strong>of</strong> the number <strong>of</strong> paedophile<br />

files found during the measurements.<br />

Here we found that this curve is growing like Figure<br />

1. Similary, we can’t discover all paedophile files<br />

during our measurement. Here, we detected 701 857<br />

paedophile files.<br />

3.3 Ages contained in filenames<br />

Figure 3 represents the distribution <strong>of</strong> ages found in<br />

the filenames obtained during our measurement. It describes<br />

the percentage <strong>of</strong> the occurences (vertical axis)<br />

for each age (horizontal axis) in the filenames viewed<br />

during our measurement. There are 77 030 filenames<br />

that contains an age.<br />

5. ACKNOWLEDGEMENT<br />

This work is supported by the European MAPAP<br />

(SIP-2006-PP-221003) <strong>and</strong> the French ANR/MAPE projects.<br />

6. REFERENCES<br />

[1] F. Aidouni, M. Latapy, <strong>and</strong> C. Magnien. Ten<br />

weeks in the life <strong>of</strong> an edonkey server. <strong>Proceedings</strong><br />

<strong>of</strong> Hot<strong>P2P</strong>’09, 2009.<br />

[2] E. Adar <strong>and</strong> B.A. Huberman. Free riding on<br />

gnutella. First Monday, vol. 5, 2000.<br />

[3] Z. Schnabel. The estimation <strong>of</strong> the total fish<br />

population <strong>of</strong> a lake. Am Math Monthly, 1938.<br />

Figure 3: Ages distribution in filenames.<br />

We observe that the interval <strong>of</strong> ages ranges from 0<br />

years old up to 20, with an important focus between 8<br />

<strong>and</strong> 15 years old <strong>and</strong> two peaks, at 9 <strong>and</strong> 12 years old.<br />

2<br />

92

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!