Computer Science – Information Theory
Scientific paper
2010-08-10
Computer Science
Information Theory
Scientific paper
We examine the Gaussian interference channel and the erasure relay channel. We focus on codes that are non-capacity-achieving ("bad") over appropriate point-to-point (two-terminal) channels. Over Gaussian point-to-point channels, for example, such codes require greater SNR than "good" ones to achieve reliable communications, but often exhibit lower estimation errors whenever the SNR is below the Shannon limit. Over multi-terminal channels, this advantage of "bad" codes at lower SNRs can be exploited by strategies that apply estimation, at various network nodes, to achieve partial decoding. Such strategies include soft partial interference cancelation (soft-IC) and soft decode-and-forward (soft-DF). We develop variants of these two approaches, which are susceptible to rigorous analysis. We focus on applications of "bad" LDPC codes. We develop analysis tools for soft-DF, including simultaneous density evolution (sim-DE), and use standard density evolution to analyze soft-IC. We apply our analysis to the design simple-structured "bad" codes that outperform more complex "good" ones.
Bennatan Amir
Calderbank Robert A.
Shamai Shlomo
No associations
LandOfFree
In Praise of Bad Codes for Multi-Terminal Communications does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with In Praise of Bad Codes for Multi-Terminal Communications, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and In Praise of Bad Codes for Multi-Terminal Communications will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-584473