Chapter Summary

Chapter Summary

Key Points

• 1.

  The relay channel model. The relay channel is the simplest network where cooperation helps: a source communicates to a destination with the help of a relay, subject to a strictly causal constraint ($X_{r,i}$ depends only on $Y_r^{i-1}$). Despite over fifty years of research, the capacity of the general relay channel remains an open problem — known exactly only for the degraded case.

• 2.

  The cut-set bound as the universal benchmark. The cut-set bound $C \leq \max_{p(x,x_r)} \min\{I(X,X_r;Y), I(X;Y,Y_r|X_r)\}$ provides the tightest general upper bound. It arises from two bottlenecks: the broadcast cut (source to relay+destination) and the multiple-access cut (source to destination with relay side information). For many practical channels, the gap to achievability is small.

• 3.

  Decode-and-forward: optimal when the relay has a good view. DF achieves $R_{\text{DF}} = \max \min\{I(X;Y_r|X_r), I(X,X_r;Y)\}$ via block-Markov encoding and backward decoding. It achieves capacity for the degraded relay channel. The key limitation: the rate is bottlenecked by the weakest hop (source-to-relay).

• 4.

  Compress-and-forward: exploiting noisy observations as side information. CF does not require the relay to decode — it compresses $Y_r$ using Wyner-Ziv coding and forwards the compressed version. CF excels when the source-relay link is weak but the relay-destination link is strong. The achievable rate depends on the compression-forwarding trade-off.

• 5.

  Compute-and-forward: structured codes for function decoding. Using nested lattice codes, the relay can decode integer linear combinations of source codewords — bridging channel coding and network coding. CaF achieves high rates when channel coefficients align with integer vectors, connecting information theory to number theory.

• 6.

  The diamond network and multi-relay extensions. The diamond network (source → two relays → destination) introduces four cuts and foreshadows noisy network coding. Quantize-map-and-forward achieves within a constant gap of the cut-set bound, showing that simple compression-based strategies are near-optimal for general relay networks.