08.08.2013 Views

Chapter 3 Time-to-live Covert Channels - CAIA

Chapter 3 Time-to-live Covert Channels - CAIA

Chapter 3 Time-to-live Covert Channels - CAIA

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Error rate<br />

1e−02<br />

1e−04<br />

1e−06<br />

1e−08<br />

MED0<br />

MED<br />

AMI<br />

DED<br />

MEI0<br />

MEI<br />

1 2 3 4 5 6 7<br />

Amplitude A<br />

DEI<br />

DUB<br />

CHAPTER 3. TIME-TO-LIVE COVERT CHANNELS<br />

Error rate<br />

1e−02<br />

5e−03<br />

1e−03<br />

5e−04<br />

1e−04<br />

5e−05<br />

1e−05<br />

MED0<br />

MED<br />

AMI<br />

DED<br />

MEI0<br />

MEI<br />

1 2 3 4 5 6 7<br />

Amplitude A<br />

Figure 3.12: Error rate for different modulation schemes and amplitudes for the <strong>CAIA</strong> trace<br />

(left) and the Leipzig trace (right) (y-axis is logarithmic)<br />

3.5.2 Error rate<br />

Figure 3.12 shows the error rate for the <strong>CAIA</strong> and Leipzig datasets depending on the<br />

different modulation schemes and A (results for other traces are in Appendix B.3). Overall<br />

the error rates for A ≤ 2 are similar <strong>to</strong> the average rate of TTL changes (see Section 3.1).<br />

A notable exception is DUB for the <strong>CAIA</strong> trace. Since this trace contains a large number<br />

of very long flows the probability for TTL wrap-arounds is much higher than for the other<br />

traces. The error rate for MEI and DUB actually increases for increasing A because the<br />

probability of wrap-arounds increases.<br />

MEI0 and MED0 have higher error rates than MEI and MED. This is because often<br />

the probability that the first special zero bit is wrong is higher than the average error<br />

rate, since TTL changes occur more frequently at the start of flows (see Section 3.1).<br />

Both direct schemes and mapped schemes perform equally well for A = 1 as predicted<br />

by the error probabilities (see Appendix B.2). In general the error rate does not decrease<br />

proportionally with increasing A because the empirical error distributions have long tails<br />

as shown in Section 3.1.5.<br />

Figure 3.13 compares the performance of the different modulation schemes averaged<br />

across all traces. For A = 1 the error rate varies between 0.1% and 1%, and MED performs<br />

best, followed by MEI, AMI and the direct schemes. For larger amplitudes MED and AMI<br />

outperform all other schemes. MEI and MED clearly outperform MEI0 and MED0.<br />

We investigated if the error rate for mapped and differential schemes can be reduced<br />

by using hop count differences instead of TTL differences. The receiver converts all TTL<br />

values <strong>to</strong> hop counts. This eliminates errors when the TTL was changed by middleboxes<br />

but the hop count is the same (see Section 3.1.3). For example, the TTLs 56 and 120 are<br />

different but the hop count is 8 in both cases assuming the usual initial TTL values. Our<br />

60<br />

DEI<br />

DUB

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

Saved successfully!

Ooh no, something went wrong!