Chapter Summary

Chapter Summary

Key Points

• 1.

  Coded D2D combines MAN-style placement with D2D delivery. Users partitioned into clusters of size $K_g$; MAN delivery within each cluster; spatial reuse across clusters.

• 2.

  Ji-Caire-Molisch 2015 result (CommIT): The scaling order of per-user throughput is $\Theta(\mu)$ for both uncoded and coded D2D. The two mechanisms — spatial reuse and coded multicasting — do not cumulate at the scaling-law level.

• 3.

  Coded scheme wins on the constant. $c_{\text{coded}}/ c_{\text{uncoded}} \approx 1 + K_g M/N$. For $K_g M/N = 2$-$5$ (typical), the coded scheme is 3-6× better in practical throughput.

• 4.

  Shared resource constraint. Both mechanisms compete for the same airtime. Spatial reuse gives concurrent links; coded multicast gives more content per link. They don't multiplicatively compound.

• 5.

  Aggregate throughput scales linearly in $n$ for D2D (both coded and uncoded) — much better than cellular's $\log n$. D2D is the right delivery paradigm for dense networks.

• 6.

  Design lever: cluster size $K_g$. Larger $K_g$ improves coded-gain constant but explodes subpacketization $\binom{K_g} {K_g M/N}$. Practical sweet spot: $K_g \in [5, 20]$.

• 7.

  Deployment viability. 5G NR Sidelink supports the basic D2D primitives. Cache-aware protocols exist as 3GPP study items (Rel-19+). Coordinated MAN placement across devices is the main standardization gap.