OFDM & 5G NR
OFDM is the modulation underlying WiFi, LTE, and 5G NR. Understanding subcarrier orthogonality, cyclic prefix, PAPR, and the 5G numerology framework is essential for any RF engineer working on modern wireless systems.
Why OFDM?
OFDM solves inter-symbol interference (ISI) in wideband channels by splitting the channel into hundreds of narrow subcarriers, each experiencing flat (non-selective) fading — allowing simple one-tap equalisation per subcarrier.
→ each subcarrier sees flat fading → trivial 1-tap equalisation
Example: Urban τrms=1 μs → Bcoh=200 kHz
5G NR 15 kHz SCS (200 kHz ≫ 15 kHz) → each SC is flat faded ✓
Subcarrier Orthogonality
Subcarriers are orthogonal when spaced at Δf = 1/Tu — at the peak of each subcarrier, all others are exactly zero. This is implemented using IFFT at the transmitter and FFT at the receiver.
IFFT / FFT Implementation
RX: X̂[k] = Σ y[n]·e−j2πkn/N (FFT, N-point)
Δf = 1/Tu = BW/N · Tu = N/BW = useful symbol duration
Cyclic Prefix
The CP copies the last N_cp samples of each OFDM symbol and prepends them. As long as T_cp ≥ τ_max (maximum channel delay spread), there is zero ISI and zero ICI. The CP converts linear convolution into circular convolution — exactly what the FFT assumes.
5G NR Normal CP (15 kHz SCS): Tcp = 4.69 μs → 6.7% overhead
Extended CP: Tcp = 16.67 μs — for large cells / high delay spread
CP Length vs Environment
| Environment | Typical τmax | 5G NR SCS |
|---|---|---|
| Indoor office | 0.1–0.3 μs | 60 kHz (T_cp=1.17 μs) ✓ |
| Urban macro | 1–3 μs | 15 kHz (T_cp=4.69 μs) ✓ |
| Rural / large cell | 3–10 μs | 15 kHz Extended CP ✓ |
| mmWave indoor | 0.01–0.05 μs | 120 kHz (T_cp=0.29 μs) ✓ |
The PAPR Problem
When N subcarriers align in phase, their voltages add — peak power is N× average. This forces PAs to back off from P1dB, reducing efficiency.
Statistical PAPR (99.9% CCDF): ≈ 8–12 dB for large N
LTE/5G NR: PAPR ≈ 10–12 dB · Single carrier QPSK: PAPR ≈ 0–3.5 dB
PAPR Reduction Techniques
| Technique | PAPR Reduction | Used In |
|---|---|---|
| Clipping & Filtering | 3–5 dB | WiFi APs, small cells |
| DPD (Digital Pre-Distortion) | Allows 3–5 dB less back-off | 5G base stations |
| DFT-s-OFDM (SC-FDMA) | PAPR ≈ 3–4 dB vs 10 dB OFDM | 5G NR UL, LTE UL |
5G NR Numerology (μ)
5G NR uses scalable subcarrier spacing: Δf = 2μ × 15 kHz. Different μ values are optimised for different frequency ranges and environments.
| μ | SCS (kHz) | Tu (μs) | Normal CP (μs) | Slots/subframe | Use Case |
|---|---|---|---|---|---|
| 0 | 15 | 66.7 | 4.69 | 1 | FR1, large cells (like LTE) |
| 1 | 30 | 33.3 | 2.34 | 2 | FR1 standard urban |
| 2 | 60 | 16.7 | 1.17 | 4 | FR1 and NR-U unlicensed |
| 3 | 120 | 8.33 | 0.59 | 8 | FR2 mmWave |
| 4 | 240 | 4.17 | 0.29 | 16 | FR2 SSB reference signals only |
Resource Grid
RB = 12 subcarriers × 1 slot in frequency
Slot = 14 OFDM symbols (normal CP)
Subframe = 1 ms = 2μ slots · Frame = 10 ms
100 MHz at μ=1 (30 kHz SCS): 275 RBs → 3300 active SCs
Peak DL (4 layers, 256-QAM): ≈ 3.2 Gbps
FR1 vs FR2
| Feature | FR1 (sub-6 GHz) | FR2 (mmWave, 24–52 GHz) |
|---|---|---|
| Max channel BW | 100 MHz | 400 MHz |
| SCS | 15, 30, 60 kHz | 60, 120 kHz |
| Antenna | 2–8 port MIMO | 64–256+ element massive MIMO |
| Peak DL speed | ~1–2 Gbps | ~4–10 Gbps |
5G RF Chain Requirements
| Parameter | LTE 20 MHz | 5G NR FR1 100 MHz | 5G NR FR2 400 MHz |
|---|---|---|---|
| Noise floor (NF=7 dB) | −94 dBm | −87 dBm | −81 dBm |
| ADC sample rate | ~30.72 Msps | ~245.76 Msps | ~983 Msps |
| LO phase noise @ 1 MHz | −135 dBc/Hz | −130 dBc/Hz | −110 dBc/Hz |
Massive MIMO & Beamforming
Beam HPBW ≈ 0.886·λ/(N·d) radians
64 elements: gain = +18 dB — offsets 18 dB mmWave path loss vs 3.5 GHz
HPBW at 28 GHz, N=64, d=λ/2: 2.8° pencil beam
OFDM / 5G Thumb Rules
| # | Rule | Consequence |
|---|---|---|
| 1 | Subcarrier spacing Δf = 1/Tu | Wider SCS = shorter symbol = better for high Doppler and mmWave |
| 2 | CP must exceed max delay spread τmax | Under-sized CP → ICI and ISI, uncorrectable |
| 3 | OFDM PAPR ≈ 10–12 dB | PA back-off required; UL uses DFT-s-OFDM (~3 dB PAPR) |
| 4 | 10× wider BW = 10 dB more noise | 5G 100 MHz is 7 dB noisier than LTE 20 MHz |
| 5 | N elements → 10·log(N) dB array gain | 64 elements = 18 dB — makes mmWave viable |
| 6 | 5G NR 1 RB = 12 subcarriers × 14 symbols | 275 RBs in 100 MHz at 30 kHz SCS |
| 7 | EVM budget: phase noise + nonlinearity + IQ imbalance | For 256-QAM (3.5%): allocate ≤1.5% each to main contributors |
OFDM and 5G NR Physical Layer Fundamentals
Orthogonal Frequency Division Multiplexing (OFDM) is the foundation of every modern wireless standard — WiFi 4/5/6/6E/7, LTE, and 5G NR all use OFDM or a derivative. Splitting a wide channel into many narrow orthogonal subcarriers eliminates inter-symbol interference without complex time-domain equalisers.
5G NR Numerology — Why Scalable SCS Matters
5G NR introduces flexible numerology where subcarrier spacing scales as 15×2ᵘ kHz. At sub-6 GHz (FR1), 15 kHz and 30 kHz SCS provide the long symbol duration needed for urban macro cells. At mmWave (FR2), 120 kHz SCS gives shorter symbols — needed because mmWave delay spreads are shorter (indoor τ_max < 100 ns) but Doppler and phase noise are worse.
PAPR — The RF Engineer's Main OFDM Concern
OFDM's high PAPR (10–12 dB) forces PAs to operate at large back-off, drastically reducing efficiency. Digital pre-distortion (DPD) linearises the PA, allowing 3–5 dB less back-off. The uplink avoids this with DFT-spread OFDM (SC-FDMA), reducing UE handset PAPR to approximately 3 dB.