Chapter Summary

Chapter Summary

Key Points

• 1.

  MIMO-OTFS channel is a 3D tensor in delay, Doppler, and angle. For $N_t \times N_r$ antennas and $MN$ DD cells, the channel tensor $\mathcal{H}_{\mathrm{DD}} \in \mathbb{C}^{N_r \times N_t \times MN}$ has multilinear rank $\leq P$ where $P$ is the number of resolvable paths. The sparsity is the structural lever for every efficient MIMO-OTFS algorithm.

• 2.

  Beamspace-DD gives 4D ultra-sparsity. Applying DFT beamspace transforms to both ends yields a tensor with only $P$ nonzero entries out of $MN N_t N_r$ — a compression factor of $\sim 10^5$ in realistic scenarios. Enables compressed-sensing channel estimation with $\mathcal{O}(P \log N)$ pilot measurements.

• 3.

  MIMO-MP detection is the standard algorithm. Message passing on the DD-spatial factor graph achieves near-MMSE with $\mathcal{O}(MN \cdot P \cdot N_s \cdot N_r)$ per iteration, converging in 5-10 iterations. Damping factor $\alpha \in [0.5, 0.9]$ stabilizes on dense channels. Implemented in commercial prototypes.

• 4.

  Diversity multiplies by $P$. The MIMO-OTFS DMT is $d^*(r) = (N_t - r)(N_r - r) P$. A $4 \times 4$ system with $P = 8$ paths gets $d^*(0) = 128$ — vastly beyond MIMO-OFDM's fixed $d^*(0) = N_t N_r$. The reliability gain is the path-count multiplier that OTFS's DD-domain processing captures.

• 5.

  Hybrid beamforming matches the sparse-path geometry. For mmWave arrays with $N_t \gg P$, the optimal precoder lies in the span of steering vectors. $N_{RF} = P$ RF chains suffice for full digital capacity. 5-10× hardware savings; standard for mmWave 5G, native for 6G MIMO-OTFS.

• 6.

  MIMO-OFDM hits a SINR ceiling at high Doppler. SINR saturates at $\sim 3/(\pi^2 (\nu/\Delta f)^2)$ above the ICI threshold. MIMO-OTFS has no such ceiling — the DD-domain processing handles arbitrary Doppler. Capacity gap at 300 km/h: $\sim 13$ dB.

• 7.

  Deployment rule: MIMO-OTFS when $\nu T_{\text{frame}} > 0.05$. Below this threshold, MIMO-OFDM wins on simplicity and standardization. Above, MIMO-OTFS wins on capacity and reliability. 5G NR mmWave at $> 60$ km/h: use MIMO-OTFS. Indoor WiFi: stay with MIMO-OFDM.

• 8.

  Compressed-sensing pilot reduction: $\mathcal{O}(P \log N)$ pilots vs $\mathcal{O}(MN N_t)$ for dense estimation. For $P = 10$, $MN N_t = 10^6$: $\sim 140$ pilots vs $\sim 10^6$ — four orders of magnitude saved. Standardization in 6G (Rel. 21+).