Network Coding and Wireless Physical-layer ... - Jacobs University
Network Coding and Wireless Physical-layer ... - Jacobs University
Network Coding and Wireless Physical-layer ... - Jacobs University
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
40 Chapter 4: Unequal Erasure Protection (UEP) in <strong>Network</strong> <strong>Coding</strong><br />
Definition 4.2 A symbol m k belonging to the scalable message M = [m 1 , m 2 , ..., m ω ] has<br />
a dependency level of λ(m k ) = j if its significance depends on the successful recovery of<br />
the symbols having the dependency level of j − 1 but not on those with larger dependency<br />
level. A symbol of which significance does not depend on any symbol has the dependency<br />
level of 1. [3, 5]<br />
Definition 4.3 A functional vector U k (S) = [u k (s 1 ), u k (s 2 ), ..., u k (s ω )], k = 1, 2, ..., N,<br />
where N is the number of sink nodes in the network-coded multicast, is called the cumulative<br />
utility vector assigned by the sink node k to the set of scalable data S described in<br />
Definition 4.1 if each element u k (s i ), 1 ≤ i ≤ ω, is a non-negative real number representing<br />
the private utility that the sink node k assigns to the scalable data s i . [3]<br />
Definition 4.4 A functional vector ∆U k (S) is called the marginal utility vector assigned<br />
by the sink node k to the set of scalable data S described in Definition 4.1 if<br />
∆U k (S) = [∆ 1 , ∆ 2 , ..., ∆ ω ] (4.3)<br />
= [u k (s 1 ) − u k (s 0 ), u k (s 2 ) − u k (s 1 ), ...<br />
..., u k (s ω ) − u k (s ω−1 )], (4.4)<br />
where u k (s 0 ) = 0 <strong>and</strong> u k (s i ), 1 ≤ i ≤ ω, represents the element in the cumulative value<br />
vector U k (S) defined in Definition 4.3. [3]<br />
Definition 4.5 The set of scalable data S is said to be ordered if <strong>and</strong> only if the two<br />
following conditions are fulfilled.<br />
λ(m u ) ≥ λ(m v ), 1 ≤ v < u ≤ ω (4.5)<br />
∆ i−1 ≥ ∆ i , 1 < i ≤ ω (4.6)<br />
where λ(m j ) <strong>and</strong> ∆ i denote the dependency level of the symbol m j <strong>and</strong> the i th element in<br />
the marginal utility vector of S, respectively. A message M corresponding to an ordered<br />
set S of scalable data is called an ordered scalable message. [3]