Chapter Summary

Chapter Summary

Key Points

• 1.

  Cooperative diversity creates virtual MIMO. Two single-antenna users cooperating via decode-and-forward achieve diversity order 2, equivalent to a $2 \times 1$ MISO system. Dynamic DF, which adapts the listening duration to the channel realization, achieves the optimal DMT $d^*(r) = 2(1-r)$ without the rate penalty of static protocols.

• 2.

  Coded cooperation integrates cooperation into the code design. By partitioning each user's codeword into a systematic part (transmitted directly) and a parity part (relayed by the partner), the effective code spans two independent fading paths, achieving full cooperative diversity with standard code constructions.

• 3.

  Cloud-RAN centralizes processing with fronthaul-limited APs. When each AP quantizes and forwards its observation (compress-and-forward), the system achieves the cooperative MIMO capacity minus a penalty determined by the quantization noise $\sigma_q^2 = (\|\mathbf{g}_l\|^2 P + \sigma^2)/(2^{2C_{\text{fh}}} - 1)$. Even moderate fronthaul capacity ($\sim$5 bits/use) is sufficient to approach the ideal centralized performance.

• 4.

  Oblivious vs channel-aware processing is the key design choice in C-RAN. Oblivious processing keeps APs simple (no CSI needed) at the cost of higher fronthaul requirements. Channel-aware processing reduces fronthaul but requires CSI acquisition at the APs.

• 5.

  Cell-free massive MIMO eliminates cell boundaries. With $L$ distributed APs serving $K \ll L$ users via matched-filter combining, the per-user rate scales as $\Theta(\log L)$ — the same as co-located massive MIMO with $L$ antennas. The macro-diversity gain of spatial distribution compensates for the loss of coherent array gain.

• 6.

  User-centric serving sets make cell-free scalable. By assigning each user to its $N_{\text{AP}}$ nearest APs rather than all $L$ APs, the fronthaul load scales with $N_{\text{AP}}$ rather than $L$, enabling practical deployment at the cost of reduced macro-diversity.