Skip to content
New
Latest News || 27 October 2026: Dr Soumava Mukherjee's invited talk at Queen's University Belfast joined device-level hardware innovation with channel-level control. || 22 October 2026: I received a Certificate of Appreciation for serving as a reviewer for IEEE IMAS 2026. || 15 October 2026: I presented a poster on beam-domain ML fingerprinting for RIS-aided mmWave localisation at the Centre for Wireless Innovation's annual symposium. || 15 October 2026: CWI's annual symposium brought research and industry perspectives on 6G, ISAC, optical wireless, AI, and future networks to Queen's University Belfast. || 14 October 2026: Erik Nordell of Ericsson visited our Keysight Lab for a live RIS-assisted video streaming demonstration.
Knowledge Hubwireless communication basics1. Spectrum, Frequency, Wavelength, and Bandwidth

wireless communication basics learning note

1. Spectrum, Frequency, Wavelength, and Bandwidth

Build the foundation of wireless communication: spectrum, carrier frequency, wavelength, bandwidth, channel capacity, and the difference between bandwidth and data rate.

At a glance

  • Frequency sets physical scaleWavelength is c/f; carrier frequency locates a band, while bandwidth measures its width.
  • State what stays fixedAt fixed signal power and noise density, increasing bandwidth reduces SNR; the capacity gain is not linear.
  • Capacity is an ideal limitDelivered throughput also depends on coding, spatial layers, resource allocation and overhead.

Why These Concepts Come First

Every wireless system begins with a signal occupying part of the electromagnetic spectrum. Before studying antennas, MIMO, beamforming, or RIS, it is necessary to understand four connected ideas:

  • Frequency tells us how fast a sinusoidal wave oscillates.

  • Wavelength tells us the physical distance covered by one cycle.

  • Spectrum describes how signal energy is distributed across frequency.

  • Bandwidth measures the frequency interval used or supported by a signal or system.

  • Why they matter: antenna size, propagation, data rate, radio hardware, and spectrum allocation all depend on them.

References for this section3

Frequency and Wavelength

For an electromagnetic wave travelling in free space,

c=fλc = f\lambda
  • cc: free-space wave speed, about 3×1083\times10^8 m/s.

  • ff: carrier frequency in hertz

  • λ\lambda: wavelength in metres.

  • 3 GHz: wavelength about 10 cm.

  • 30 GHz: wavelength about 1 cm.

  • Implication: smaller wavelengths permit compact elements and arrays; path loss, blockage, and hardware loss still need checking.

FrequencyApproximate wavelengthTypical context
900 MHz33.3 cmWide-area cellular and IoT
3.5 GHz8.6 cmMid-band 5G
28 GHz10.7 mmmmWave access, sensing, RIS research
100 GHz3 mmSub-THz research and very wide bandwidths

How long is one wave? Frequency sets the size of everything

1. Formula and symbols
λ=cf,Ldipole≈λ2,Aiso=λ24π\lambda=\frac{c}{f},\qquad L_{\rm dipole}\approx\frac\lambda2,\qquad A_{\rm iso}=\frac{\lambda^2}{4\pi}
FSPL=20log⁡104πdλ dB\mathrm{FSPL}=20\log_{10}\frac{4\pi d}{\lambda}\ \text{dB}
  • cc: Speed of light, 3 × 10⁸ m/s
  • ff: Carrier frequency
  • λ\lambda: Wavelength: distance the wave travels in one cycle
  • LdipoleL_{\rm dipole}: Length of a half-wave dipole antenna
  • AisoA_{\rm iso}: Effective area of an ideal isotropic antenna
  • dd: Link distance
  • FSPL\mathrm{FSPL}: Free-space path loss between two isotropic antennas

FSPL grows with frequency only because the isotropic antenna’s area λ²/4π shrinks. A fixed-size antenna array gains that area back as directivity.

One wave on a 40 cm ruler (37 cycles)0 cm10 cm20 cm30 cm40 cmλ = 1.1 cmDrawn to the same scale:half-wave dipole8 × 8 array, λ/2 spacing
2. Work with the current inputs
λ=3×10828.00×109=1.1 cm,λ2=5.35 mm\lambda=\frac{3\times10^8}{28.00\times10^9}=1.1\ \mathrm{cm},\qquad\frac\lambda2=5.35\ \mathrm{mm}
8×8 array side=8⋅λ2=4.3 cm,Aiso=0.0912 cm28\times8\ \text{array side}=8\cdot\tfrac\lambda2=4.3\ \mathrm{cm},\qquad A_{\rm iso}=0.0912\ \mathrm{cm^2}
FSPL=20log⁡104π(100)0.01071=101.4 dB\mathrm{FSPL}=20\log_{10}\frac{4\pi(100)}{0.01071}=101.4\ \mathrm{dB}
3. Result
Wavelength λ1.1 cm
Half-wave dipole5.35 mm8 × 8 array: 4.3 cm square
Free-space loss at 100 m101.4 dB
The wave is drawn to scale on a 40 cm ruler. At 900 MHz one cycle is longer than the ruler; at 28 GHz it is about 1 cm, so a 64-element array fits on a coin. The ideal isotropic antenna’s capture area λ²/4π shrinks with frequency, which is where the extra free-space loss of high bands comes from.
References for this section3

What Bandwidth Means

If a system passes frequencies between a lower cut-off fLf_{\mathrm{L}} and an upper cut-off fHf_{\mathrm{H}}, its nominal bandwidth is

B=fH−fLB = f_{\mathrm{H}} - f_{\mathrm{L}}
  • BB: frequency interval width in hertz.

  • fHf_{\mathrm{H}}: upper boundary

  • fLf_{\mathrm{L}}: lower boundary, both in hertz.

  • Do not mix units: 20 MHz is a frequency width, not a bit rate.

  • Actual bit rate: also depends on modulation, coding, SNR, channel conditions, overhead, and receiver design.

Several bandwidth definitions appear in practice:

  • Absolute bandwidth: the complete interval between the lowest and highest non-zero spectral components.
  • Occupied bandwidth: the interval containing a stated percentage, commonly 99%, of the transmitted power.
  • Channel bandwidth: the spectrum assigned to a radio channel by a standard or regulator.
  • Transmission bandwidth: the frequency range that a device or channel can pass with acceptable distortion.
  • Effective bandwidth: a context-dependent measure of the portion that contributes meaningfully to system performance.

The definitions should not be mixed. For example, an OFDM waveform can be allocated a nominal channel bandwidth while guard bands and inactive subcarriers reduce the bandwidth that carries user data.

References for this section3

Baseband, Carrier, and Passband

Information normally begins as a baseband signal. Modulation shifts that information around a radio-frequency carrier so it can be radiated by an antenna.

Information→Baseband waveform→Carrier modulation→RF transmission
  • Carrier frequency: centre location of the radio signal in the spectrum.
  • Bandwidth: width occupied around that carrier.
  • Example: 20 MHz at 2.4 GHz and 20 MHz at 28 GHz have the same width but different wavelengths and propagation.
References for this section3

Bandwidth and the Ideal Capacity Limit

For an additive white Gaussian noise channel, Shannon's capacity formula is

C=Blog⁡2 ⁣(1+SN)C = B\log_{2}\!\left(1+\frac{S}{N}\right)
  • CC: idealized channel capacity in bit/s under the AWGN model.
  • BB: bandwidth in hertz.
  • SS: signal power in the common linear unit.
  • NN: noise power in that same linear unit.
  • S/NS/N is linear SNR, not dB.

Suppose B=20 MHzB=20\ \mathrm{MHz} and SNR=10 dB\mathrm{SNR}=10\ \mathrm{dB}. First convert 10 dB to a linear ratio: S/N=10S/N=10. The ideal capacity is then approximately

C=20×106log⁡2(11)≈69.2 Mbit/s\begin{aligned} C &= 20\times10^{6}\log_{2}(11) \\ &\approx 69.2\ \text{Mbit/s} \end{aligned}
  • 20×10620\times10^6: example bandwidth in hertz
  • 11=1+1011=1+10 for 10 dB SNR converted to linear ratio 10.
  • CC: ideal upper bound, about 69.2 Mbit/s here.
  • Reality: pilots, guards, control, interference, retransmissions, and implementation losses reduce delivered rate.

The formula exposes two useful facts:

  1. Capacity grows linearly with bandwidth when SNR is fixed.
  2. Capacity grows logarithmically with SNR, so repeatedly increasing transmit power gives diminishing returns.

Fixed power is a different comparison. With fixed received signal power and white-noise density, noise power grows with bandwidth: N=N0BN=N_0B. If linear SNR at 20 MHz is γ20\gamma_{20}, then at bandwidth BB in MHz, γ(B)=γ20(20/B)\gamma(B)=\gamma_{20}(20/B). The lab lets you compare this case with fixed SNR. Maintaining fixed SNR as bandwidth grows requires proportionally more received signal power.

Does wider bandwidth always multiply the rate?

1. Formula and symbols
λ=c/f,N=kTBF\lambda=c/f,\quad N=kTB F
γ=Pr/N,C=Blog⁡2(1+γ)\gamma=P_r/N,\qquad C=B\log_2(1+\gamma)
  • cc: wave speed
  • ff: carrier
  • λ\lambda: wavelength
  • kk: Boltzmann constant
  • TT: noise temperature
  • BB: bandwidth
  • FF: linear noise factor
  • PrP_r: received power
  • γ\gamma: linear SNR
  • CC: ideal capacity
  • NN: Receiver noise power
0.0100.0200.0Ideal capacity (Mbit/s) versus bandwidth1 MHz100 MHz
Bandwidth comparison
2. Work with the current inputs
λ=2997924583×109=99.93 mm\lambda=\frac{299792458}{3\times10^9}=99.93\ \mathrm{mm}
γ=1010/102020=10.000\gamma=10^{10/10}\frac{20}{20}=10.000
C=Blog⁡2(1+γ)=20log⁡2(1+10.000)=69.19 Mbit/sC=B\log_2(1+\gamma)=20\log_2(1+10.000)=69.19\ \mathrm{Mbit/s}
3. Result
Wavelength c/f99.93 mm
SNR at chosen bandwidth10.00 dB
Ideal capacity69.19 Mbit/s
Ideal AWGN Shannon limit, not measured throughput. In fixed-power mode, noise density and received signal power stay fixed, so SNR falls as bandwidth grows. The SNR slider specifies the 20 MHz reference. Fixed-SNR mode requires signal power to grow with bandwidth. Carrier frequency sets free-space wavelength, not this ideal rate.
References for this section3

Bandwidth Does Not Equal Throughput

It is common to hear that “more bandwidth means more speed.” That is directionally correct, but incomplete. Four terms should be separated:

  • Bandwidth BB is frequency-domain width in hertz.
  • Symbol rate RsR_{\mathrm{s}} is the number of transmitted modulation symbols per second.
  • Data rate or PHY bit rate is the number of coded or uncoded bits carried per second at the physical layer.
  • Throughput is the rate successfully delivered across a chosen interface; goodput counts only the useful application payload received correctly.

For pulse-shaped single-carrier signalling, occupied bandwidth and symbol rate are connected by the pulse shape. With a raised-cosine roll-off factor α\alpha, a common passband approximation is

B≈(1+α)RsB\approx(1+\alpha)R_{\mathrm{s}}
  • BB: approximate occupied passband width in hertz under the stated raised-cosine convention.
  • RsR_{\mathrm{s}}: modulation symbol rate in symbols/s.
  • α\alpha: roll-off factor, typically between 0 and 1.

If each symbol carries log⁡2(M)\log_2(M) coded bits using MM-ary modulation, the gross modulated bit rate is

Rgross=Rslog⁡2(M)R_{\mathrm{gross}} = R_{\mathrm{s}}\log_2(M)
  • RgrossR_{\mathrm{gross}}: uncoded gross modulated bit rate in bit/s.
  • RsR_{\mathrm{s}}: symbols/s
  • MM: number of modulation constellation points.
  • log⁡2M\log_2M: bits per symbol for an ideal MM-ary mapping.

Forward-error-correction code rate rcr_{\mathrm{c}}, the fraction of usable time-frequency resources ηRE\eta_{\mathrm{RE}}, and the number of spatial layers NlayerN_{\mathrm{layer}} give the useful PHY-rate approximation

RPHY≈B ηSE Nlayer ηRER_{\mathrm{PHY}} \approx B\,\eta_{\mathrm{SE}}\,N_{\mathrm{layer}}\,\eta_{\mathrm{RE}}
  • RPHYR_{\mathrm{PHY}}: approximate physical-layer payload rate in bit/s under this convention.
  • BB: bandwidth in hertz
  • ηSE\eta_{\mathrm{SE}}: per-layer spectral efficiency in bit/s/Hz, already reflecting modulation and coding.
  • NlayerN_{\mathrm{layer}}: usable spatial layers
  • ηRE\eta_{\mathrm{RE}}: fraction of resource elements available after stated overhead.

Scheduling, headers, and retransmissions can reduce delivered throughput further:

Rthroughput≈RPHYηscheduleηprotocol(1−PreTx)R_{\mathrm{throughput}} \approx R_{\mathrm{PHY}} \eta_{\mathrm{schedule}} \eta_{\mathrm{protocol}} (1-P_{\mathrm{reTx}})
  • RthroughputR_{\mathrm{throughput}}: delivered rate in bit/s at the stated interface.
  • RPHYR_{\mathrm{PHY}}: physical-layer rate from the previous estimate.
  • ηschedule\eta_{\mathrm{schedule}}: usable scheduled-resource fraction.
  • ηprotocol\eta_{\mathrm{protocol}}: protocol-efficiency fraction.
  • PreTxP_{\mathrm{reTx}}: simplified lost fraction to retransmission, not a standards-compliant universal formula.
  • Goodput: can fall further after transport and application headers.
Bandwidth→Symbols per second→PHY data rate→User throughput→Application goodput

User throughput can also be summarized as

Ruser≈B ηSE ηresourceR_{\mathrm{user}} \approx B\,\eta_{\mathrm{SE}}\,\eta_{\mathrm{resource}}
  • RuserR_{\mathrm{user}}: compact user-rate estimate in bit/s.
  • BB: bandwidth in hertz
  • ηSE\eta_{\mathrm{SE}}: spectral efficiency in bit/s/Hz.
  • ηresource\eta_{\mathrm{resource}}: usable fraction of resources under this compact model.
  • Example: 100 MHz at 1 bit/s/Hz can carry less than 40 MHz at 6 bit/s/Hz before other differences.

For example, suppose a 100 MHz carrier achieves 4 bit/s/Hz over two spatial layers, while 70% of resources carry scheduled user data and combined protocol/retransmission efficiency is 85%. An illustrative throughput is

Rthroughput≈100×106×4×2×0.70×0.85=476 Mbit/sR_{\mathrm{throughput}} \approx 100\times10^6\times4\times2\times0.70\times0.85 =476\ \mathrm{Mbit/s}
  • 100×106100\times10^6: bandwidth in hertz
  • 44: example per-layer bit/s/Hz
  • 22: spatial layers.
  • 0.700.70: scheduled usable resource fraction
  • 0.850.85: example protocol/retransmission efficiency.
  • RthroughputR_{\mathrm{throughput}}: illustrative 476 Mbit/s, not a guaranteed device speed.
References for this section3

Coherence Bandwidth and Frequency Selectivity

  • Delayed copies: multipath makes parts of a signal arrive at different times.
  • Signal narrower than coherence bandwidth: channel response is approximately flat across it.
  • Signal wider than coherence bandwidth: different frequencies can fade differently.

OFDM handles this by dividing a wideband channel into many narrow subcarriers. Each subcarrier experiences an approximately flat channel, making equalization simpler.

Flat or frequency-selective? Compare signal bandwidth with coherence bandwidth

1. Formula and symbols
H(f)=∑ℓaℓ e−j2πfτℓ,στ=∑ℓPℓ(τℓ−τˉ)2∑ℓPℓH(f)=\sum_\ell a_\ell\,e^{-j2\pi f\tau_\ell},\qquad\sigma_\tau=\sqrt{\frac{\sum_\ell P_\ell(\tau_\ell-\bar\tau)^2}{\sum_\ell P_\ell}}
Bc≈15στ(frequency correlation about 0.5)B_c\approx\frac{1}{5\sigma_\tau}\quad\text{(frequency correlation about 0.5)}
  • H(f)H(f): Channel frequency response
  • aℓ, τℓa_\ell,\ \tau_\ell: Amplitude and delay of path ℓ (three paths here)
  • PℓP_\ell: Path power |a_ℓ|²
  • τˉ\bar\tau: Power-weighted mean delay
  • στ\sigma_\tau: RMS delay spread
  • BcB_c: Coherence bandwidth: rough width over which H(f) stays similar
  • BB: Signal bandwidth (shaded band)

The 1/(5σ) rule is an approximate rule of thumb (Rappaport), not an exact law; the plotted |H(f)| is exact for the three-path model.

050100−30−150frequency offset (MHz)|H(f)|² (dB)width of B_c ≈ 2.53 MHzshaded: signal band B = 20 MHz
2. Work with the current inputs
τ=(0, 100, 230) ns,P=(1, 0.5, 0.22)⇒τˉ=58.5 ns, στ=79.1 ns\tau=(0,\ 100,\ 230)\ \mathrm{ns},\quad P=(1,\ 0.5,\ 0.22)\Rightarrow\bar\tau=58.5\ \mathrm{ns},\ \sigma_\tau=79.1\ \mathrm{ns}
Bc≈15×79.1×10−9=2.53 MHz,B/Bc=7.91B_c\approx\frac1{5\times79.1\times10^{-9}}=2.53\ \mathrm{MHz},\qquad B/B_c=7.91
gain variation inside B: max⁡∣H∣2−min⁡∣H∣2=23.5 dB\text{gain variation inside }B:\ \max|H|^2-\min|H|^2=23.5\ \mathrm{dB}
3. Result
RMS delay spread79.1 ns
Coherence bandwidth ≈2.53 MHz
Gain variation in band23.5 dBfrequency-selective
Three echoes with fixed relative powers (0, −3 and −6.6 dB); the slider stretches their delays. The curve is the exact channel gain across 100 MHz; the shaded band is your signal. When the band is much narrower than the coherence bandwidth the gain inside it is nearly flat; when it is wider, some frequencies sit in deep fades. OFDM avoids the problem by using many narrow subcarriers.
References for this section3

Practical Design Questions

When evaluating a wireless link, ask:

  • What is the carrier frequency and corresponding wavelength?
  • What channel bandwidth is allocated?
  • How much of it carries payload rather than guards and pilots?
  • What SNR and interference level are available?
  • Is the channel flat or frequency selective across the waveform?
  • What spectral efficiency can the chosen modulation and coding sustain?
References for this section3

Takeaway

  • Bandwidth: frequency-space available to the waveform.
  • Delivered rate: depends on SNR, spectral efficiency, overhead, and propagation.
  • Frequency/wavelength: also determine antenna scale and propagation behaviour.

Continue with Propagation, Link Budgets, and Cellular Infrastructure to build on these ideas.

References for this section3

Check your understanding

0 of 4 answered

  1. Using c ≈ 3 × 10⁸ m/s, what is the wavelength at 3 GHz?

  2. What is ideal AWGN capacity at 20 MHz and SNR = 10 dB?

  3. At fixed received signal power and noise density, bandwidth doubles. What happens to linear SNR?

  4. A 100 MHz carrier has 4 bit/s/Hz per layer, 2 layers, usable resource fraction 0.70 and remaining efficiency 0.85. What rate results?

Complete references and further reading