Chapter Summary

Chapter Summary

Key Points

  • 1.

    OTFS transmitter has three stages. (1) Place QAM symbols on the DD grid XDDX_{DD}. (2) ISFFT precoder produces TF-grid symbols XTFX_{TF}. (3) Heisenberg transform (OFDM-like modulator with CP) generates the time-domain waveform s(t)s(t). The total compute is O(MNlog(MN))O(MN\log(MN)), dominated by the 2D FFT of the ISFFT and the NN length-MM IFFTs of the Heisenberg transform.

  • 2.

    Heisenberg and Wigner transforms are unitary adjoints. Under critical-lattice bi-orthogonal pulses, WigHeis=Id\text{Wig} \circ \text{Heis} = \text{Id} on the TF grid. No energy is lost or created; the transmit and receive chains invert cleanly. Unitarity also means the DD-domain noise is i.i.d. CN(0,σ2)\mathcal{CN}(0, \sigma^2) — a clean Gaussian channel for the detector.

  • 3.

    Cyclic prefix LCPlmaxL_{CP} \geq l_{\max} is required. The doubly-circular DD input-output relation holds exactly only when CP absorbs the channel memory. OTFS inherits OFDM's CP dimensioning directly; at 5G NR numerology the standard CP (normal or extended) suffices for most deployment scenarios. Frame-level (reduced) CP schemes can save overhead at the cost of more advanced detection.

  • 4.

    End-to-end DD-to-DD chain realizes the 2D convolution. Under ideal pulses and sufficient CP, the full OTFS chain produces X^DD=hXDD+WDD\hat{X}_{DD} = h \star\star X_{DD} + W_{DD} — exactly the Chapter 4 DD input-output relation. The detector (Chapter 8) inverts this sparse 2D convolution to recover the data symbols. Pulse-shape and CP imperfections introduce bounded cross-talk that can be controlled by design choices (Chapter 20).

  • 5.

    OTFS is precoded OFDM. The entire modulator chain is ISFFT → OFDM transmitter, and the demodulator chain is OFDM receiver → SFFT. The OFDM blocks are unchanged from 5G NR; only the ISFFT/SFFT precoder/postcoder are new. This deploys as software/firmware on existing 5G hardware, not as a silicon-level redesign — a concrete argument for 6G standardization feasibility.

  • 6.

    Zak-OTFS is the conceptually cleaner alternative. The CommIT Zak-OTFS formulation (Mohammed-Hadani-Chockalingam-Caire 2022) constructs the transmit waveform directly from the discrete Zak transform, producing the same waveform as Hadani-Rakib OTFS under idealized pulses. Zak-OTFS is preferred in the recent literature for its cleaner treatment of pulse-shape effects and fractional Doppler.

Looking Ahead

With the OTFS transceiver fully specified, we now turn to practical questions: how to estimate the channel at the receiver (Chapter 7), how to detect the data symbols (Chapter 8), and how much performance OTFS delivers (Chapter 9). Chapter 7 — embedded pilot estimation — contains the first of the CommIT contributions in Part II of this book: the pilot design that recovers PP path parameters with a single pilot impulse and minimal guard region. Chapter 8 develops the DD-domain detectors (MP, LMMSE, LCD) that exploit the sparsity and structure we built here. Chapter 9 proves the diversity and performance advantages quantitatively.