DFT-s-OFDM vs CP-OFDM

Where OFDM's peaks come from

An OFDM symbol is the sum of hundreds of independently-modulated subcarriers. Most of the time they interfere every which way and the envelope is modest — but occasionally they line up in phase and add coherently, producing a huge instantaneous peak. The ratio of that peak to the average power is the PAPR, and it forces the power amplifier to back off (waste headroom) to avoid clipping. Add subcarriers and watch the peaks grow.

Two transmit chains

CP-OFDM maps each modulation symbol straight onto a subcarrier. DFT-s-OFDM inserts an M-point DFT first, spreading every symbol across all M allocated subcarriers — so the time-domain signal behaves like a single carrier. Everything downstream (IFFT, CP) is identical.

How a symbol is actually built — OFDM & DFT-s-OFDM

An OFDM symbol is literally a sum of sine waves — one subcarrier each, every one carrying a data symbol. Press build and watch the subcarriers stack into the time-domain waveform. Both sides use the same data: plain OFDM lets each subcarrier go its own way (peaks form when they align), while DFT-s-OFDM runs a DFT first so the very same sum comes out flat. That flatness is the whole point.

The envelope, side by side

Same bits, two waveforms. CP-OFDM swings violently; DFT-s-OFDM stays flat. A flatter envelope means the PA can run closer to saturation without clipping — directly buying transmit power and coverage.

PAPR CCDF

The CCDF answers "what fraction of OFDM symbols exceed a given peak-to-average ratio?" CP-OFDM's curve sits well to the right — big peaks are common. DFT-s-OFDM (and especially π/2-BPSK) is shifted left, so the PA rarely sees large peaks. Read off the 0.1% point — the value engineers design the PA back-off around.

Spectrum & PA efficiency

PAPR isn't just about clipping the wanted signal — when a high-PAPR signal is pushed through a non-linear PA, the distortion splatters energy into adjacent channels (worse ACLR). The low-PAPR DFT-s-OFDM signal tolerates a hotter PA operating point with cleaner spectral regrowth, so the UE both reaches further and interferes less.

Receiver complexity & noise

Nothing is free. Because DFT-s-OFDM spreads each symbol over all subcarriers, the receiver must de-spread with an IDFT after frequency-domain equalisation — and that spreading also couples the per-subcarrier noise, giving slightly worse performance in highly frequency-selective channels (a little noise enhancement). CP-OFDM equalises each subcarrier independently and pairs naturally with MIMO.

Full comparison

When to use which

Move the slider from cell-centre to cell-edge and watch the recommendation flip.

How it's configured

Test yourself