27.06.2013 Views

6th European Conference - Academic Conferences

6th European Conference - Academic Conferences

6th European Conference - Academic Conferences

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.

Hind Al Falasi and Liren Zhang<br />

therefore, creating a covert channel diminishes. The service time at the high security network is the<br />

only factor under the control of the high security network. The other elements are controlled by the<br />

physical environment of the network. On the other hand, if the distance between them is small, we<br />

estimate that the i th packet is deleted from the buffer at α1 + T1 + T2 + 1/ µ2.<br />

Another element to consider is the high security network service time, which affects the SAFP<br />

queuing time. We are considering this element because it leads to the establishment of a timing<br />

covert channel between the high security network and the low security network. A slow service time<br />

eventually leads to a full buffer at the SAFP. In other words, packets from the low security network are<br />

lost; therefore, no acknowledgments are sent from the SAFP to the low security network. From there,<br />

the high security network can control the SAFP buffer; subsequently, it can control the rate of the<br />

acknowledgments from the SAFP to the low security network. Therefore, it can use the delays to<br />

signal messages to the low security network. The SAFP buffer is modeled using the M/M/1/K model<br />

as it has a finite capacity where the maximum number of packets in the buffer is K. A packet enters<br />

the queue if it finds fewer than K packets in the buffer and is lost otherwise. The probability of a full<br />

buffer = blocking probability = probability of a covert channel. An illustration of the above scenario is<br />

presented in Figure 2.<br />

Figure 2: Communication representation between low security network, SAFP and high security<br />

network<br />

5. Analysis of the system using Markov chain model<br />

Using the state transition diagram (see Figure 3), we found the blocking probability of the SAFP buffer<br />

(PK):<br />

Solving the equations in terms of P0:<br />

p<br />

p<br />

p<br />

(<br />

1<br />

2<br />

k<br />

1<br />

2<br />

p<br />

1<br />

0<br />

p<br />

K<br />

1<br />

2<br />

1<br />

2<br />

1<br />

2<br />

2<br />

p<br />

0<br />

2<br />

k<br />

2<br />

) p<br />

1<br />

p<br />

p<br />

k<br />

p<br />

1<br />

p<br />

0<br />

0<br />

K 1<br />

1<br />

p<br />

k 1<br />

k<br />

k<br />

P<br />

K<br />

0<br />

1<br />

2<br />

p<br />

1<br />

2<br />

k 1<br />

K<br />

p<br />

0<br />

k<br />

1<br />

k<br />

0<br />

k<br />

K<br />

K<br />

1<br />

11<br />

(1)<br />

(2)<br />

(3)

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

Saved successfully!

Ooh no, something went wrong!