DFT-s-OFDM

DFT-s-OFDM is architecturally a single-carrier transmission scheme despite using OFDM-style processing. The key is the precoding DFT step before the conventional IFFT. By spreading each data symbol across all allocated subcarriers in the frequency domain, the time-domain signal effectively becomes a single-carrier waveform. This means the transmit signal does not suffer from the superposition of multiple independently modulated subcarriers, which is the root cause of high Peak-to-Average Power Ratio (PAPR) in CP-OFDM. The result is a waveform that combines the robustness of single-carrier modulation with the flexible frequency-domain equalization and resource allocation benefits of OFDM.

The primary motivation for DFT-s-OFDM is PAPR reduction. In CP-OFDM, the instantaneous power can spike dramatically when multiple subcarriers constructively interfere. DFT-s-OFDM avoids this by ensuring each time-domain sample is a weighted sum of all data symbols, not just one per subcarrier. This produces a signal with an amplitude distribution closer to a single-carrier waveform.

• Typical PAPR Improvement: 2-3 dB lower than CP-OFDM for the same constellation.
• Benefit: Allows the power amplifier to operate closer to its saturation point, dramatically improving power efficiency.
• Critical for Uplink: This is why it was chosen for the LTE uplink and remains an option for 5G NR uplink, where user equipment (UE) battery life is paramount.

The output of the M-point DFT precoder is mapped to a subset of N available IFFT subcarriers (where M < N). This mapping defines the multiple access scheme:

• Localized FDMA (LFDMA): The M DFT outputs are mapped to a contiguous block of subcarriers. This provides frequency-domain scheduling gain, allowing the base station to assign users to the best parts of the channel.
• Distributed FDMA (DFDMA): The M DFT outputs are spread across the entire bandwidth with zeros inserted between them. This creates a comb-like spectrum, providing frequency diversity even without channel knowledge, at the cost of higher sensitivity to frequency offset.

LFDMA is the standard choice in LTE and 5G NR uplink due to its compatibility with channel-dependent scheduling.

A defining advantage of DFT-s-OFDM is its compatibility with low-complexity Frequency-Domain Equalization (FDE). Because a cyclic prefix is inserted, the linear convolution with the multipath channel becomes circular convolution. At the receiver, after CP removal and an N-point FFT, a simple single-tap equalizer can be applied to each subcarrier to compensate for the channel. This is far less computationally intensive than the time-domain equalizers required for traditional single-carrier systems. The equalized symbols are then transformed back via an M-point IDFT to recover the data. This architecture is known as Single-Carrier Frequency Domain Equalization (SC-FDE).

To enable coherent demodulation and channel estimation, Demodulation Reference Signals (DMRS) must be multiplexed with DFT-s-OFDM data. In the LTE uplink, DMRS occupies the center symbol of each slot (SC-FDMA symbol 3). Critically, the DMRS is inserted in the time domain, after the M-point DFT precoding. This means the DMRS is not precoded and maintains a constant amplitude zero autocorrelation (CAZAC) property in the frequency domain, enabling accurate channel estimation across the allocated bandwidth. The 5G NR uplink supports a more flexible DMRS configuration, including front-loaded patterns for low-latency decoding.

In 5G NR, DFT-s-OFDM is officially termed transform precoding. It is an optional feature for the uplink, enabled via higher-layer signaling. When enabled, the standard CP-OFDM chain is modified by inserting the DFT spreader. 5G NR also introduces DFT-s-OFDM with frequency-domain spectral shaping (FDSS) using a raised-cosine filter in the frequency domain. This further reduces PAPR and out-of-band emissions, making it particularly suitable for millimeter-wave (mmWave) and power-limited devices. The choice between CP-OFDM and DFT-s-OFDM in NR allows the network to optimize for either MIMO spatial multiplexing (CP-OFDM) or coverage and power efficiency (DFT-s-OFDM).

The defining metric that DFT-s-OFDM is engineered to minimize. PAPR is the ratio of the peak signal power to the average power, expressed in dB. High PAPR forces power amplifiers to operate with significant back-off to avoid non-linear distortion, drastically reducing efficiency.

• CP-OFDM PAPR: Typically 10-13 dB, requiring expensive, inefficient amplifiers.
• DFT-s-OFDM PAPR: Reduced to 4-7 dB, mimicking single-carrier behavior.
• Impact: Lower PAPR directly extends battery life in user equipment (UE) and reduces thermal dissipation requirements.

The receiver-side signal processing technique that makes DFT-s-OFDM practical. By inserting a Cyclic Prefix (CP) and performing equalization in the frequency domain, the receiver can combat multipath fading with low computational complexity.

• Mechanism: The CP transforms linear convolution with the channel into circular convolution, enabling simple single-tap equalization per subcarrier.
• Advantage: FDE provides the robustness of OFDM against frequency-selective channels while preserving the single-carrier waveform's low PAPR.
• Contrast: Traditional single-carrier systems require complex time-domain equalizers.

The multiple-access variant of DFT-s-OFDM used in the LTE uplink. SC-FDMA assigns different users to non-overlapping sets of subcarriers at the IDFT input, ensuring intra-cell orthogonality.

• Localized FDMA (LFDMA): Assigns contiguous subcarriers to a user, preserving the low PAPR property.
• Distributed FDMA (DFDMA): Assigns interleaved subcarriers, sacrificing some PAPR reduction for frequency diversity.
• LTE Choice: 3GPP selected LFDMA for the LTE uplink to maximize power amplifier efficiency.

The specific name for the DFT operation in the 3GPP specifications (TS 38.211). Transform precoding is an optional step in the 5G NR uplink physical layer that spreads modulated symbols across allocated subcarriers before the OFDM modulator.

• Enable/Disable: The network can dynamically configure transform precoding via RRC signaling. When disabled, the UE falls back to CP-OFDM.
• DFT Size: Must be a multiple of 2, 3, and 5 to allow efficient FFT implementation.
• Purpose: Explicitly defined to maintain compatibility with the LTE SC-FDMA waveform for power-limited devices.

A key 5G NR enhancement allowing dynamic switching between DFT-s-OFDM and CP-OFDM on a per-transmission basis. This enables the gNB to optimize the uplink for coverage or capacity.

• Coverage-Limited Scenarios: The network schedules DFT-s-OFDM to maximize UE transmit power efficiency at the cell edge.
• Capacity-Limited Scenarios: The network switches to CP-OFDM to enable UL MIMO spatial multiplexing for higher data rates.
• RRC Configuration: The transformPrecoder parameter in the PUSCH configuration determines the waveform.