Chapter Summary

Chapter Summary

Key Points

• 1.

  The cache-aided block-fading BC extends Chapter 5's static MIMO BC with a coherence block length $T_c$ and per-block pilot overhead $\tau$. CSIT quality determines the realizable spatial multiplexing gain $L$.

• 2.

  CSIT scales the spatial DoF, not the caching DoF. With estimation error $\sigma_e^2/\sigma^2$, $\mathrm{DoF} = t + L(1 - \sigma_e^2/\sigma^2)_+$. Caching gain $t$ is CSIT-independent; spatial gain is not. On CSIT-poor channels (mmWave, high mobility), caching is disproportionately valuable.

• 3.

  Pilot overhead caps spatial DoF at $T_c/4$. With $\tau = L$ pilots per $T_c$-length block, effective spatial DoF = $L(1 - L/T_c)$, maximized at $L^* = T_c/2$ for $T_c/4$ spatial DoF. Caching gain $t$ is pilot-free and adds on top.

• 4.

  Blind interference management. In the no-CSIT regime, pure MIMO DoF collapses to 1 but cache-aided delivery achieves $\mathrm{DoF} = t + 1$. Cached side information substitutes for CSIT; XOR cancellation works without channel knowledge.

• 5.

  Outage analysis: group size matters. The multicast group size $t + L$ determines the worst-user outage penalty. Larger groups yield higher DoF but worse outage rate. Optimal group sizing requires trade-off analysis at the target outage level.

• 6.

  High-mobility scenarios benefit most from caching. At $T_c = 100$, $L = 32$, $\mu = 0.2$ and 20+ users, the caching gain contributes roughly half the deliverable DoF. Without caching, the spatial pipeline is capped by the pilot wall.

• 7.

  Design implication. Cache-aided designs should be preferred over CSIT-hungry massive MIMO in CSIT-poor environments. For vehicular, aerial, and mmWave deployments, coded caching is a primary DoF resource, not an afterthought.