25.04.2014 Views

TITRE Adaptive Packet Video Streaming Over IP Networks - LaBRI

TITRE Adaptive Packet Video Streaming Over IP Networks - LaBRI

TITRE Adaptive Packet Video Streaming Over IP Networks - LaBRI

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

packets. If the number of lost packets is not more than h i , then the decoder will be able to recover<br />

Ui. Otherwise, Ui is completely lost. Figure 4-16 shows the format of packets sent on the <strong>IP</strong><br />

network.<br />

d bytes<br />

<strong>IP</strong>/UDP/RTP<br />

header<br />

4 bytes<br />

FEC header<br />

seq n k<br />

RTP Payload<br />

Figure 4-16: Header information and packet format<br />

UEP augments the amount of the traffic sent in the network. To correctly control the volume<br />

of transmitted data, we have to provide a certain ratio of traffic overhead called r, for each level of<br />

priority score. We assume that moving from one priority score to other increases by a 10 percent<br />

ratio. Then, the ratio r can be defined by:<br />

r = 0 . 1×<br />

p<br />

(Eq.3)<br />

Therefore, the traffic overhead is limited to 10 percent (i.e., r=0.1) for the data flow of priority<br />

score 1, to 20 percent (i.e., r=0.2) for the data flow of priority 2, and so on. In order to find the<br />

efficient value n i for the (n i , k ) i RS code, we proceed as below:<br />

∗<br />

Let ς<br />

U i<br />

be the reserved byte-budget for error protection. It depends on the number of bytes<br />

used to send U i when no error protection is performed. It is given by:<br />

( k ⋅ d m )<br />

ς (Eq.4)<br />

∗<br />

U<br />

= r ⋅<br />

i i<br />

+<br />

i<br />

Where d is the packet header size (i.e, when RTP/UDP/<strong>IP</strong> is used with the proposed UEP, d<br />

= (20+8+12+4) = 44 bytes). The relation between the real byte-budget spent on error protection,<br />

ς<br />

U i<br />

, and the RS code to be used can be stated as follows:<br />

The error margin between<br />

( n − k ) ⋅ ( t d )<br />

ς (Eq.5)<br />

U i<br />

=<br />

i i i<br />

+<br />

ς<br />

U i<br />

and ∗ Ui<br />

∧<br />

∗<br />

ς is ς ς U<br />

− ς<br />

=<br />

i Ui<br />

, that can be positive or negative. It<br />

cumulates along the data access unit arrivals. Using the formula (Eq.4) and (Eq.5), the fluctuation<br />

of the error margin can be written as:<br />

∧<br />

ς<br />

( n )<br />

i<br />

⎧<br />

⎪<br />

= ⎨<br />

⎪<br />

⎩<br />

0 , i = 1<br />

∧<br />

⎪( r + 1) ⋅ m − n ⋅ t + ( r − n + k ) ⋅ d + ς ( n )<br />

i<br />

i<br />

i<br />

i<br />

i<br />

i−1 , i<br />

> 1<br />

(Eq.6)<br />

To respect the constraint given on traffic overhead, the best value n i is the one that provides<br />

the smallest error margin in formula (Eq.6). Then, n i is obtained by:<br />

min<br />

ni<br />

∧<br />

( ni<br />

) such that ni<br />

∈ℵ,<br />

ni<br />

≥ ki<br />

, ∀mi<br />

, ki<br />

, ti<br />

ς (Eq.7)<br />

With the proposed UEP, the RS code evolves dynamically so the network bandwidth is<br />

correctly controlled according to video application requirement.<br />

86

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

Saved successfully!

Ooh no, something went wrong!