Signal bandwidth is the span of frequencies occupied by a signal, typically measured in hertz (Hz) between defined lower and upper frequency limits. The exact definition of those limits varies by context: common conventions include the 3 dB (half-power) points, the null-to-null width of a main lobe, or a regulatory mask that bounds occupied spectral power.
In practice
In analog front-end design, bandwidth determines the minimum cutoff frequency required of anti-aliasing filters and amplifiers in the signal chain. If the signal bandwidth exceeds half the ADC sample rate, aliasing folds out-of-band energy back into the passband, corrupting the digitized signal. For a baseband signal centered at DC, the required sample rate is at least twice the one-sided bandwidth (Nyquist). For bandpass signals centered at a carrier frequency, undersampling techniques can be used legitimately if the signal band fits within a single Nyquist zone, as covered in "Sampling bandpass signals."
In digital communications, signal bandwidth is coupled to data rate and modulation scheme, along with coding rate, pulse shaping, and spectral efficiency. A narrower channel bandwidth limits symbol rate and therefore throughput; higher-order modulations (e.g., 256-QAM) pack more bits per symbol to recover data rate without widening the spectrum. The relationship between received signal power, noise power in the occupied bandwidth, and bit error rate is central to link budget analysis, as explored in "Understanding and Relating Eb/No, SNR, and other Power Efficiency Metrics."
In RF and wireless embedded systems, bandwidth also has a regulatory dimension. Transmitters must generally confine emissions within assigned channel bandwidths (e.g., 20/40/80/160 MHz channels in Wi-Fi, or narrowband channels in standards such as LoRa and Zigbee, though exact allocations are standard- and region-dependent). Exceeding the mask causes interference to adjacent channels and can result in certification failure. "RF in Slow Motion: Sonifying a Wi-Fi 7 Packet" illustrates how a wide-bandwidth Wi-Fi 7 packet occupies its channel over time.
When working at complex baseband, bandwidth is measured relative to DC rather than to the original RF carrier. A 10 MHz bandpass signal centered at 2.4 GHz, when downconverted and represented as a complex (I/Q) baseband signal, occupies roughly -5 MHz to +5 MHz around DC. This is central to software-defined radio (SDR) and modem design; "Generating Complex Baseband and Analytic Bandpass Signals" and "Model Signal Impairments at Complex Baseband" go into practical detail on that representation.
Discussed on DSPRelated
Frequently asked
What is the difference between signal bandwidth and channel bandwidth?
Signal bandwidth describes the frequency span actually occupied by a specific signal's energy. Channel bandwidth is an allocated slot defined by a standard or regulator (e.g., a 5 MHz LTE channel, a 125 kHz LoRa channel). The signal bandwidth must fit within the channel bandwidth, but the two are not the same number and the signal often occupies less than the full channel.
How does signal bandwidth relate to the required ADC sample rate?
For a
baseband signal, the
Nyquist criterion requires the
sample rate to be at least twice the highest frequency component present. For a bandpass signal, the required rate depends on where the band sits relative to Nyquist zones: if the band is confined to a single Nyquist zone, a sample rate as low as twice the signal bandwidth can in principle suffice, even if the carrier frequency is much higher, though in practice guard bands and anti-alias filtering requirements mean the usable minimum is somewhat higher. Some margin beyond the theoretical minimum is always applied.
Why does bandwidth matter for noise and SNR?
Thermal noise power scales linearly with bandwidth: P_noise = kTB, where k is Boltzmann's constant, T is temperature in Kelvin, and B is bandwidth in Hz. A wider signal occupies a wider noise floor, reducing
SNR for the same signal power. This trade-off between bandwidth and SNR is fundamental to receiver design and is quantified in metrics like
Eb/No.
What does 3 dB bandwidth mean, and are there other definitions?
The 3 dB bandwidth is the frequency range over which a signal's spectral level or a system's
frequency response stays within 3 dB (half power) of its peak or reference level. Other definitions are used depending on context: null-to-null bandwidth marks the first spectral nulls on either side of a main lobe; occupied bandwidth is defined by regulators as the band containing a specified percentage (often 99%) of total signal power; noise-equivalent bandwidth (NEBW) is used in filter and receiver analysis. The right definition depends on the application.
Is signal bandwidth the same as data rate or throughput?
No, but they are related. Shannon's channel capacity theorem sets an upper bound on data rate as C = B * log2(1 +
SNR), where B is bandwidth in Hz. A wider signal bandwidth allows a higher maximum data rate, but actual throughput also depends on SNR, modulation order, and coding rate. Doubling bandwidth can double capacity only if SNR is maintained.
Differentiators vs similar concepts
Signal bandwidth is often confused with two adjacent concepts. First, system or hardware bandwidth (e.g., an amplifier's -3 dB frequency or an
ADC's analog input bandwidth) describes the
frequency response of a component, not the spectral occupancy of a signal. Second, data rate or bit rate is a measure of information throughput in bits per second, not a frequency span. The two are related through modulation and Shannon's theorem but are distinct quantities with different units.
