Prerequisites & Notation

Before You Begin

This chapter combines coded caching with multi-antenna broadcast. The reader should be comfortable with the MAN scheme and with basic MIMO-BC information theory. Prior exposure to zero-forcing beamforming and degrees-of-freedom analysis helps but is not strictly required.

  • The MAN scheme and the rate formula (Ch 2)(Review ch02)

    Self-check: Can you state the delivery structure for integer tt and compute the rate for K=10K = 10, t=2t = 2?

  • Index coding perspective on coded caching (Ch 4)(Review ch04)

    Self-check: Can you explain why the MAN conflict graph admits a tight fractional-chromatic coloring?

  • MIMO broadcast channel capacity and DoF(Review ch15)

    Self-check: Can you state the DoF of an LL-antenna Gaussian BC with KK single-antenna users (for L≀KL \leq K)?

  • Zero-forcing and dirty paper coding(Review ch17)

    Self-check: Can you sketch how ZF beamforming achieves min⁑(L,K)\min(L, K) DoF?

  • Complex Gaussian random variables and noise models

    Self-check: Are you comfortable with w∼CN(0,Οƒ2I)\mathbf{w} \sim \mathcal{CN}(\mathbf{0}, \sigma^2 \mathbf{I})?

  • Interference alignment basics(Review ch26)

    Self-check: Can you sketch the interference alignment idea for a 3-user interference channel?

Notation for This Chapter

New symbols for the multi-antenna setting. In this chapter LL denotes the number of transmit antennas (CC-specific convention); cross-book chapters may use NtN_t instead.

SymbolMeaningIntroduced
LLNumber of transmit antennas (single transmitter)s01
KKNumber of single-antenna userss01
hk∈CL\mathbf{h}_k \in \mathbb{C}^LChannel vector from the LL-antenna transmitter to user kks01
x∈CL\mathbf{x} \in \mathbb{C}^LTransmitted vector per channel uses01
yky_kReceived scalar at user kk: yk=hkHx+wky_k = \mathbf{h}_k^H \mathbf{x} + w_ks01
wk,CN(0,Οƒ2)\mathbf{w}_{k}, \mathcal{CN}(0, \sigma^2)Additive noise at user kk, complex Gaussians01
SNR\text{SNR}Signal-to-noise ratio (transmit SNR)s01
DoF\mathrm{DoF}Degrees of Freedom: lim⁑SNRβ†’βˆžR/log⁑2SNR\lim_{\text{SNR} \to \infty} R/\log_2 \text{SNR}s02
ttCoded caching gain t=KM/Nt = KM/Ns02
Sβ€²\mathcal{S}'(t+L)(t + L)-subset of [K][K]: Lampiris-Caire delivery groups03