Network Coding and Wireless Physical-layer ... - Jacobs University
Network Coding and Wireless Physical-layer ... - Jacobs University
Network Coding and Wireless Physical-layer ... - Jacobs University
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Chapter 5: <strong>Network</strong> <strong>Coding</strong> with LT Codes as Erasure Codes 59<br />
5.1 Introduction <strong>and</strong> the System Model of Relay<br />
<strong>Network</strong> Multicast with LT <strong>and</strong> <strong>Network</strong> Codes<br />
LT codes or other rateless codes as a forward error correction (FEC) scheme are arguably<br />
most useful in multicast applications in which its one-to-many nature makes the acknowledgement<br />
scheme very unpleasant. LT codes can potentially generate an infinite stream<br />
of encoded output symbols such that, even when some symbols are erased, the receiver<br />
can recover k original symbols using any K received encoded symbols when K is only<br />
slightly larger than k.<br />
Like LT codes, network codes find their first <strong>and</strong> simplest application in multicast.<br />
However, both types of codes have opposite aims. While LT codes increase redundancy<br />
in the networks to compensate for erasures, network codes decrease it by means of coding<br />
at the bottlenecks.<br />
According to Fig. 5.1, where A wants to multicast two binary symbols b 1 <strong>and</strong> b 2 to D<br />
<strong>and</strong> E, we can see that the node F encodes b 1 <strong>and</strong> b 2 together to achieve the multicast<br />
rate of 2 bits per unit time, if each edge represents the capacity of one bit per unit time. D<br />
receives b 1 from the path ABD <strong>and</strong> can recover b 2 from the symbol b 1 ⊕b 2 from ACF GD,<br />
whereas E receives b 2 from the path ACE <strong>and</strong> recovers b 1 from the symbol b 1 ⊕ b 2 from<br />
ABF GD. Had network coding not been there, only either b 1 or b 2 would have been able<br />
to pass the bottleneck F G in one unit time, i.e., one receiver would have been unable to<br />
use one of its possible transmission paths.<br />
In this chapter, we assume that the source node generates LT-coded symbols which<br />
are subsequently network coded by intermediate nodes along their ways to the receivers.<br />
To make things more concrete, Fig. 5.2 displays a block diagram showing all encoding<br />
<strong>and</strong> decoding processes as well as the buffer structure from the source to the destinations<br />
in accordance with the network in Fig. 5.1.<br />
One can observe a switch placed after the LT-encoder block <strong>and</strong> prior to two buffers.<br />
While traditional network coding only requires that the switch turns to each buffer half of