Guard Band
Understanding Guard Bands
Guard bands are the necessary waste in spectrum allocation, a tax paid for the imperfection of real-world RF hardware. If filters had infinitely sharp roll-off, oscillators never drifted, and amplifiers were perfectly linear, guard bands would be zero. In practice, every wireless standard must specify guard bands wide enough to prevent interference between adjacent operators while keeping them narrow enough to maximize spectral efficiency.
Guard Band Sizing
BWguard = BWtransition/filter_order
Spectral efficiency impact:
η = BWuseful/(BWuseful+BWguard)
OFDM guard subcarriers:
Nguard = NFFT−Ndata−Npilot
Guard Band by Standard
| Standard | Channel BW | Guard Band | Percentage | Modulation |
|---|---|---|---|---|
| GSM | 200 kHz | ~10 kHz | 5% | GMSK |
| LTE 20 MHz | 20 MHz | 2 MHz | 10% | OFDM |
| 5G NR 100 MHz | 100 MHz | 4.05 MHz | 8.1% | OFDM |
| Wi-Fi 6 80 MHz | 80 MHz | ~6 MHz | 7.5% | OFDM |
| DVB-T2 8 MHz | 8 MHz | 0.36 MHz | 4.5% | OFDM |
Key Equations
Power: dB = 10log(P2/P1)
Voltage: dB = 20log(V2/V1)
dBm to watts:
P(W) = 10(dBm−30)/10
0 dBm = 1 mW, +30 dBm = 1 W
Wavelength:
λ = c/f = 300/f(MHz) meters
Comparison
| System | Guard BW | % overhead | Filter order | Purpose |
|---|---|---|---|---|
| LTE 20M | 2×0.5 MHz | 5% | N/A | Adjacent ch |
| 5G NR 100M | 2×0.85 MHz | 1.7% | N/A | Adjacent ch |
| WiFi 20M | 2×1.5 MHz | 15% | N/A | Channel edge |
| FDD duplex | 10–80 MHz | varies | Duplexer | TX-RX iso |
| TV channels | 0.25–1 MHz | 3–5% | Cavity | Adjacent ch |
Frequently Asked Questions
Why needed?
Real filters have finite roll-off. Oscillators drift (10 ppm = 35 kHz at 3.5 GHz). PA nonlinearity causes spectral regrowth. Doppler shift at 500 km/h = 1.6 kHz. Guard bands absorb all these imperfections.
5G NR spec?
3GPP defines per SCS and BW. 30 kHz SCS, 100 MHz: 4.05 MHz guard (8.1%). Narrower than LTE (10%) due to better baseband filtering. Guard = (channel BW - occupied BW)/2.
Trade-off?
Wider guard = less ACI but wasted spectrum. Licensed spectrum costs $1-5/MHz/pop. 8% waste on 100 MHz = $8-40M/operator. Drives development of sharper filters and better DPD linearization.