Mathematical Foundations of Communication
Every communication system — from early telegraph to 5G cellular — rests on mathematical principles that define what is possible and what is not.
1. Fourier Transform
Joseph Fourier discovered that any periodic signal can be decomposed into a sum of sinusoidal waves. This insight, formalized as the Fourier Transform, is the single most important mathematical tool in all of communication engineering. Every radio, every modem, every digital filter uses Fourier analysis.
Continuous Fourier Transform
The Fourier Transform converts a time-domain signal x(t) into its frequency-domain representation X(f):
┌─────────────────────────────────────────────────────┐ │ │ │ Forward Transform: │ │ │ │ X(f) = ∫ x(t) · e^(-j2πft) dt │ │ -∞ │ │ │ │ Inverse Transform: │ │ │ │ x(t) = ∫ X(f) · e^(j2πft) df │ │ -∞ │ │ │ │ Where: │ │ j = √(-1) (imaginary unit) │ │ f = frequency in Hz │ │ t = time in seconds │ │ e = Euler's number (2.71828...) │ │ │ └─────────────────────────────────────────────────────┘
Euler's Formula
The complex exponential bridges sinusoidal and exponential representations:
┌─────────────────────────────────────────────────────┐ │ │ │ e^(jθ) = cos(θ) + j·sin(θ) │ │ │ │ cos(θ) = [e^(jθ) + e^(-jθ)] / 2 │ │ │ │ sin(θ) = [e^(jθ) - e^(-jθ)] / 2j │ │ │ └─────────────────────────────────────────────────────┘
Key Properties
┌─────────────────────────────────────────────────────┐ │ │ │ Linearity: │ │ a·x₁(t) + b·x₂(t) ⟺ a·X₁(f) + b·X₂(f) │ │ │ │ Time Shift: │ │ x(t - t₀) ⟺ X(f) · e^(-j2πft₀) │ │ │ │ Frequency Shift (Modulation): │ │ x(t) · e^(j2πf₀t) ⟺ X(f - f₀) │ │ │ │ Convolution Theorem: │ │ x(t) * h(t) ⟺ X(f) · H(f) │ │ (time-domain convolution = frequency-domain │ │ multiplication) │ │ │ │ Parseval's Theorem: │ │ ∫|x(t)|² dt = ∫|X(f)|² df │ │ (total energy is preserved) │ │ │ │ Time-Frequency Duality: │ │ If x(t) ⟺ X(f) │ │ Then X(t) ⟺ x(-f) │ │ │ └─────────────────────────────────────────────────────┘
Why It Matters for Communication
- Any signal = sum of sinusoids: A square wave, a pulse, a voice signal—all can be expressed as sums of sine waves at different frequencies
- Bandwidth measurement: The Fourier Transform reveals how much frequency spectrum a signal occupies
- Filter design: Filters are defined by how they modify the frequency spectrum—multiply X(f) by H(f) in frequency domain
- Modulation: Shifting a signal to a carrier frequency is simply multiplication by e^(j2πf₀t)—a frequency shift
- Channel modeling: Multipath channels are characterized by their frequency response H(f)
Discrete Fourier Transform (DFT)
For digital systems, we work with sampled data. The DFT computes frequency components of N discrete samples:
┌─────────────────────────────────────────────────────┐ │ │ │ DFT: X[k] = Σ x[n] · e^(-j2πkn/N) │ │ n=0..N-1 │ │ │ │ IDFT: x[n] = (1/N) Σ X[k] · e^(j2πkn/N) │ │ k=0..N-1 │ │ │ │ FFT (Fast Fourier Transform): │ │ Cooley-Tukey algorithm computes DFT in │ │ O(N log N) instead of O(N²) │ │ │ │ Frequency resolution: Δf = fs / N │ │ │ └─────────────────────────────────────────────────────┘
2. Sampling Theorem
The Nyquist-Shannon Sampling Theorem bridges the continuous analog world and the discrete digital world. It tells us exactly how fast we must sample to perfectly reconstruct a signal.
The Theorem
┌─────────────────────────────────────────────────────┐ │ │ │ Nyquist-Shannon Sampling Theorem: │ │ │ │ A bandlimited signal x(t) with maximum │ │ frequency B Hz can be perfectly reconstructed │ │ from samples taken at rate fs if and only if: │ │ │ │ fs ≥ 2B │ │ │ │ The minimum rate fs = 2B is called the │ │ Nyquist rate. │ │ │ │ Reconstruction formula (sinc interpolation): │ │ │ │ x(t) = Σ x(nT) · sinc[(t - nT) / T] │ │ n │ │ │ │ Where: │ │ T = 1/fs (sampling period) │ │ sinc(x) = sin(πx) / (πx) │ │ │ └─────────────────────────────────────────────────────┘
Aliasing
When the sampling rate is too low (fs < 2B), high-frequency components fold back into lower frequencies, creating distortion called aliasing:
┌─────────────────────────────────────────────────────┐ │ │ │ Aliased frequency: │ │ │ │ f_alias = |f_signal - k · fs| │ │ │ │ Where k is the integer that minimizes the │ │ absolute value. │ │ │ │ Example: │ │ Signal frequency: 8 kHz │ │ Sampling rate: 10 kHz │ │ Aliased frequency: |8 - 1×10| = 2 kHz │ │ (8 kHz tone appears as 2 kHz!) │ │ │ │ Anti-aliasing filter: │ │ Low-pass filter with cutoff = B before │ │ sampling removes frequencies above B │ │ │ └─────────────────────────────────────────────────────┘
ADC Quantization
Analog-to-digital conversion combines sampling (in time) with quantization (in amplitude). The Effective Number of Bits (ENOB) relates to signal-to-noise ratio:
┌─────────────────────────────────────────────────────┐ │ │ │ ENOB = (SNR - 1.76) / 6.02 │ │ │ │ Where: │ │ SNR = signal-to-noise ratio in dB │ │ ENOB = effective number of bits │ │ │ │ Quantization noise power: │ │ q = V_pp / 2^n (quantization step) │ │ P_q = q² / 12 (uniform quantization noise) │ │ │ │ Dynamic range: DR = 6.02·n + 1.76 dB │ │ │ │ Example: 16-bit ADC │ │ DR = 6.02×16 + 1.76 = 98.08 dB │ │ ENOB ≈ 15.7 (practical limit) │ │ │ └─────────────────────────────────────────────────────┘
Standard Sampling Rates
┌─────────────────────────────────────────────────────┐ │ │ │ Application Rate Bandwidth │ │ ───────────────────────────────────────────── │ │ Telephony (G.711) 8 kHz 3.4 kHz │ │ Wideband voice 16 kHz 7 kHz │ │ CD Audio (Red Book) 44.1 kHz 20 kHz │ │ Professional Audio 48 kHz 20 kHz │ │ DVD Audio 96 kHz 40 kHz │ │ Ultra Hi-Res 192 kHz 80 kHz │ │ WiFi (802.11ac) 160 MHz 80 MHz │ │ 5G NR (sub-6) Up to 400 MHz │ │ │ └─────────────────────────────────────────────────────┘
3. Shannon-Hartley Theorem
Claude Shannon and Ralph Hartley established the fundamental limit on the rate at which information can be transmitted over a noisy channel. This theorem defines the absolute ceiling of communication.
The Theorem
┌─────────────────────────────────────────────────────┐ │ │ │ Shannon-Hartley Theorem: │ │ │ │ C = B · log₂(1 + S/N) │ │ │ │ Where: │ │ C = channel capacity (bits per second) │ │ B = channel bandwidth (Hz) │ │ S/N = signal-to-noise ratio (linear, not dB) │ │ │ │ In dB form: │ │ │ │ SNR(dB) = 10 · log₁₀(S/N) │ │ S/N = 10^(SNR_dB / 10) │ │ │ │ Therefore: │ │ C = B · log₂(1 + 10^(SNR_dB/10)) │ │ │ └─────────────────────────────────────────────────────┘
Capacity Examples
┌─────────────────────────────────────────────────────┐ │ │ │ Example 1: ADSL Phone Line │ │ B = 1.1 MHz, SNR = 40 dB (S/N = 10000) │ │ C = 1.1×10⁶ × log₂(1 + 10000) │ │ C = 1.1×10⁶ × 13.29 = 14.6 Mbps │ │ (Theoretical max—actual ADSL ≈ 8 Mbps) │ │ │ │ Example 2: WiFi 5 GHz Channel │ │ B = 80 MHz, SNR = 30 dB (S/N = 1000) │ │ C = 80×10⁶ × log₂(1 + 1000) │ │ C = 80×10⁶ × 9.97 = 797 Mbps │ │ (Theoretical max—actual WiFi ≈ 400 Mbps) │ │ │ │ Example 3: 5G mmWave │ │ B = 400 MHz, SNR = 20 dB (S/N = 100) │ │ C = 400×10⁶ × log₂(1 + 100) │ │ C = 400×10⁶ × 6.66 = 2.66 Gbps │ │ (Theoretical max—before MIMO gains) │ │ │ └─────────────────────────────────────────────────────┘
Practical Implications
┌─────────────────────────────────────────────────────┐ │ │ │ Why systems approach but never reach Shannon: │ │ │ │ 1. Finite block length: │ │ Shannon assumes infinite block length. │ │ Practical codes use finite blocks, │ │ introducing a gap of 0.5-2 dB. │ │ │ │ 2. Coding overhead: │ │ Error correction codes add redundancy. │ │ Turbo codes: ~0.7 dB from Shannon limit │ │ LDPC codes: ~0.04 dB from Shannon limit │ │ Polar codes: achieve capacity asymptotically │ │ │ │ 3. Non-Gaussian noise: │ │ Shannon assumes AWGN. Real channels have │ │ fading, interference, burst errors. │ │ │ │ 4. Hardware limitations: │ │ ADC/DAC resolution, oscillator phase noise, │ │ amplifier nonlinearity all reduce capacity. │ │ │ │ Bandwidth vs Power Tradeoff: │ │ C = B·log₂(1 + S/N) │ │ │ │ Doubling bandwidth: C_new ≈ C + B·log₂(2) │ │ ≈ C + B (additive gain) │ │ │ │ Doubling power: C_new ≈ B·log₂(1 + 2·S/N) │ │ (logarithmic gain) │ │ │ │ Bandwidth is always more valuable than power! │ │ │ └─────────────────────────────────────────────────────┘
Spectral Efficiency
┌─────────────────────────────────────────────────────┐ │ │ │ Spectral Efficiency: │ │ η = Rb / B (bits per second per Hz) │ │ │ │ Maximum: η_max = log₂(1 + S/N) (Shannon) │ │ │ │ System Comparison: │ │ ───────────────────────────────────────────── │ │ AM Radio: η ≈ 0.05 bps/Hz │ │ FM Radio: η ≈ 0.5 bps/Hz │ │ 2G GSM: η ≈ 1.35 bps/Hz │ │ 3G UMTS: η ≈ 2.4 bps/Hz │ │ 4G LTE: η ≈ 5 bps/Hz │ │ 5G NR: η ≈ 10 bps/Hz │ │ Theoretical: η ≈ 15 bps/Hz (with massive MIMO)│ │ │ └─────────────────────────────────────────────────────┘
4. Link Budget Calculation
A link budget accounts for all gains and losses from transmitter to receiver. It determines whether a communication link is feasible and how much margin exists.
Basic Link Equation
┌─────────────────────────────────────────────────────┐ │ │ │ Pr = Pt + Gt - L + Gr │ │ │ │ Where (all values in dB or dBW/dBm): │ │ Pr = received power │ │ Pt = transmitter output power │ │ Gt = transmitter antenna gain │ │ L = total path loss │ │ Gr = receiver antenna gain │ │ │ └─────────────────────────────────────────────────────┘
Effective Isotropic Radiated Power (EIRP)
┌─────────────────────────────────────────────────────┐ │ │ │ EIRP = Pt + Gt │ │ │ │ Where: │ │ Pt = transmitter power (dBW or dBm) │ │ Gt = transmit antenna gain (dBi) │ │ │ │ Example: │ │ Pt = 30 dBm (1 W), Gt = 12 dBi │ │ EIRP = 30 + 12 = 42 dBm (15.8 W EIRP) │ │ │ │ Regulatory limits: │ │ US WiFi 2.4 GHz: 36 dBm EIRP (4 W) │ │ US WiFi 5 GHz: 36 dBm EIRP (4 W) │ │ EU WiFi 2.4 GHz: 20 dBm EIRP (100 mW) │ │ LoRa 915 MHz: 30 dBm (1 W) │ │ │ └─────────────────────────────────────────────────────┘
Free Space Path Loss (FSPL)
┌─────────────────────────────────────────────────────┐ │ │ │ FSPL(dB) = 20·log₁₀(d) + 20·log₁₀(f) + 32.45 │ │ │ │ Where: │ │ d = distance in km │ │ f = frequency in MHz │ │ │ │ Derivation from inverse square law: │ │ FSPL = (4πdf/c)² │ │ FSPL(dB) = 20·log₁₀(4πdf/c) │ │ = 20·log₁₀(d) + 20·log₁₀(f) │ │ + 20·log₁₀(4π/c) │ │ = 20·log₁₀(d) + 20·log₁₀(f) │ │ + 32.45 │ │ │ │ Examples: │ │ WiFi at 10m, 2.4 GHz: │ │ FSPL = 20·log₁₀(0.01) + 20·log₁₀(2400) │ │ + 32.45 = 40.0 dB │ │ │ │ Cellular at 1 km, 800 MHz: │ │ FSPL = 20·log₁₀(1) + 20·log₁₀(800) │ │ + 32.45 = 90.5 dB │ │ │ │ Satellite at 36000 km, 12 GHz: │ │ FSPL = 20·log₁₀(36000) + 20·log₁₀(12000) │ │ + 32.45 = 205.2 dB │ │ │ └─────────────────────────────────────────────────────┘
Received Power and Link Margin
┌─────────────────────────────────────────────────────┐ │ │ │ Pr(dBm) = EIRP(dBm) - FSPL(dB) + Gr(dBi) │ │ │ │ Link Margin = Pr - sensitivity │ │ │ │ Example WiFi Link: │ │ Pt = 20 dBm (100 mW) │ │ Gt = 5 dBi (omnidirectional) │ │ EIRP = 25 dBm │ │ FSPL at 50m, 5 GHz = 20·log₁₀(0.05) │ │ + 20·log₁₀(5000) + 32.45 = 76.0 dB │ │ Gr = 3 dBi (laptop antenna) │ │ Pr = 25 - 76.0 + 3 = -48.0 dBm │ │ Sensitivity = -70 dBm (for 54 Mbps 802.11a) │ │ Link Margin = -48.0 - (-70) = 22.0 dB │ │ │ └─────────────────────────────────────────────────────┘
Noise Figure and C/N₀
┌─────────────────────────────────────────────────────┐ │ │ │ Noise Figure: │ │ F = SNR_in / SNR_out (linear ratio) │ │ F(dB) = SNR_in(dB) - SNR_out(dB) │ │ │ │ Cascaded Noise Figure (Friis formula): │ │ │ │ F_total = F₁ + (F₂-1)/G₁ + (F₃-1)/(G₁·G₂) │ │ + ... │ │ │ │ Where: │ │ F₁, F₂, ... = noise figures of stages │ │ G₁, G₂, ... = gains of preceding stages │ │ │ │ Key insight: First stage dominates! │ │ A low-noise amplifier (LNA) at the antenna │ │ minimizes total system noise figure. │ │ │ │ Noise Power: │ │ Pn = kTB │ │ k = 1.38 × 10⁻²³ J/K (Boltzmann constant) │ │ T = temperature in Kelvin (290 K typical) │ │ B = bandwidth in Hz │ │ │ │ Noise spectral density: │ │ N₀ = kT = -174 dBm/Hz at 290 K │ │ │ │ C/N₀ = Pr(dBm) - 10·log₁₀(B) + 174 (dB-Hz) │ │ │ └─────────────────────────────────────────────────────┘
5. Modulation Theory Mathematics
Modulation is the process of encoding information onto a carrier wave. Each modulation scheme has distinct mathematical properties that determine its bandwidth, noise performance, and implementation complexity.
Amplitude Modulation (AM)
┌─────────────────────────────────────────────────────┐ │ │ │ AM Signal: │ │ s(t) = Ac · [1 + m · x(t)] · cos(2πfct) │ │ │ │ Where: │ │ Ac = carrier amplitude │ │ m = modulation index (0 ≤ m ≤ 1) │ │ x(t) = normalized message signal (-1 to +1) │ │ fc = carrier frequency │ │ │ │ Modulation index: │ │ m = ΔAc / Ac = (max-min)/(max+min) │ │ │ │ Bandwidth: BW = 2·fm │ │ (fm = highest message frequency) │ │ │ │ Power relationship: │ │ P_total = Pc · (1 + m²/2) │ │ Pc = carrier power = Ac²/2 │ │ Psidebands = Pc · m²/2 │ │ │ │ Efficiency: │ │ η = Psidebands / P_total = m²/(2 + m²) │ │ Maximum efficiency (m=1): η = 33.3% │ │ │ └─────────────────────────────────────────────────────┘
Frequency Modulation (FM)
┌─────────────────────────────────────────────────────┐ │ │ │ FM Signal: │ │ s(t) = Ac · cos(2πfct + β·sin(2πfmt)) │ │ │ │ Where: │ │ β = modulation index = Δf / fm │ │ Δf = peak frequency deviation │ │ fm = modulating frequency │ │ │ │ Instantaneous frequency: │ │ f(t) = fc + Δf · cos(2πfmt) │ │ │ │ Carson's Rule (bandwidth estimation): │ │ │ │ BW ≈ 2(Δf + fm) = 2fm(β + 1) │ │ │ │ SNR Improvement over AM: │ │ │ │ SNR_FM = SNR_AM × 3β²(m+1)/m² │ │ │ │ For β >> 1 (wideband FM): │ │ SNR improvement ≈ 3β² │ │ (FM trades bandwidth for improved SNR) │ │ │ │ Pre-emphasis and de-emphasis: │ │ H_pre(f) = 1 + j·2πf·τ │ │ τ = 75 μs (US), 50 μs (Europe) │ │ │ └─────────────────────────────────────────────────────┘
Phase Modulation (PM)
┌─────────────────────────────────────────────────────┐ │ │ │ PM Signal: │ │ s(t) = Ac · cos(2πfct + φ(t)) │ │ │ │ Where: │ │ φ(t) = kp · x(t) (phase deviation) │ │ kp = phase sensitivity (rad/volt) │ │ │ │ Relationship to FM: │ │ PM with x(t) = FM with x(t) integrated │ │ FM with x(t) = PM with x(t) differentiated │ │ │ │ Peak phase deviation: │ │ Δφ = kp · max|x(t)| │ │ │ │ Bandwidth: BW ≈ 2(Δφ + 1)·fm │ │ (similar to Carson's rule for FM) │ │ │ └─────────────────────────────────────────────────────┘
6. Digital Modulation Mathematics
Digital modulation maps bits onto symbols transmitted over the air. The choice of modulation directly impacts data rate, spectral efficiency, and error performance.
BPSK (Binary Phase Shift Keying)
┌─────────────────────────────────────────────────────┐ │ │ │ BPSK Signal: │ │ s(t) = A·cos(2πfct + φ) │ │ φ = 0° for bit 1 │ │ φ = 180° for bit 0 │ │ │ │ Constellation: │ │ │ │ Q (Imaginary) │ │ │ │ │ │ │ │ ────●────────────●──── I (Real) │ │ -1 +1 │ │ (bit 0) (bit 1) │ │ │ │ BER = 0.5 · erfc(√(Eb/N₀)) │ │ │ │ Where: │ │ Eb = energy per bit │ │ N₀ = noise spectral density │ │ erfc = complementary error function │ │ │ │ For BER = 10⁻⁶: Eb/N₀ = 10.5 dB required │ │ │ └─────────────────────────────────────────────────────┘
QPSK (Quadrature Phase Shift Keying)
┌─────────────────────────────────────────────────────┐
│ │
│ QPSK Signal: │
│ s(t) = I·cos(2πfct) - Q·sin(2πfct) │
│ I, Q ∈ {-1, +1} │
│ │
│ Constellation: │
│ │
│ Q │
│ │ ● (+1,+1) ● (-1,+1) │
│ │ bit=11 bit=01 │
│ │ │
│ ────┼────────────────────────── I │
│ │ │
│ │ ● (+1,-1) ● (-1,-1) │
│ │ bit=10 bit=00 │
│ │ │
│ │
│ Same BER as BPSK for same Eb/N₀: │
│ BER = 0.5 · erfc(√(Eb/N₀)) │
│ │
│ But: 2 bits per symbol → half the bandwidth │
│ Rs = Rb/2 (symbol rate = bit rate / 2) │
│ │
│ Spectral efficiency: η = 2 bps/Hz │
│ │
└─────────────────────────────────────────────────────┘16-QAM (16-Quadrature Amplitude Modulation)
┌─────────────────────────────────────────────────────┐ │ │ │ 16-QAM Constellation (4×4 grid): │ │ │ │ Q │ │ │ │ │ +3 ┤ ● ● ● ● │ │ +1 ┤ ● ● ● ● │ │ -1 ┤ ● ● ● ● │ │ -3 ┤ ● ● ● ● │ │ └──┴────┴────┴────┴── I │ │ -3 -1 +1 +3 │ │ │ │ 4 bits per symbol (log₂(16) = 4) │ │ │ │ Symbol rate: Rs = Rb/4 │ │ Spectral efficiency: η = 4 bps/Hz │ │ │ │ BER ≈ (3/2)·erfc(√(Eb/(5N₀))) │ │ (approximation for large SNR) │ │ │ │ Eb/N₀ required for BER = 10⁻⁶: ~14.4 dB │ │ (3.9 dB more than BPSK) │ │ │ │ Tradeoff: Higher throughput, requires more power │ │ │ └─────────────────────────────────────────────────────┘
Symbol Rate vs Bit Rate
┌─────────────────────────────────────────────────────┐ │ │ │ Fundamental relationships: │ │ │ │ Rs = Rb / log₂(M) │ │ Rb = Rs × log₂(M) │ │ η = Rb / B = log₂(M) × (Rs/B) │ │ │ │ Where: │ │ Rs = symbol rate (baud) │ │ Rb = bit rate (bps) │ │ M = constellation size (number of symbols) │ │ B = bandwidth (Hz) │ │ η = spectral efficiency (bps/Hz) │ │ │ │ Modulation M bits/sym η (bps/Hz) │ │ ───────────────────────────────────────── │ │ BPSK 2 1 ~1 │ │ QPSK 4 2 ~2 │ │ 16-QAM 16 4 ~4 │ │ 64-QAM 64 6 ~6 │ │ 256-QAM 256 8 ~8 │ │ 1024-QAM 1024 10 ~10 │ │ │ └─────────────────────────────────────────────────────┘
BER vs Eb/N₀ for Key Modulations
┌─────────────────────────────────────────────────────┐ │ │ │ BER Performance (AWGN channel): │ │ │ │ Modulation Formula Eb/N₀ @ 10⁻⁶ │ │ ───────────────────────────────────────────── │ │ BPSK 0.5·erfc(√(Eb/N₀)) 10.5 dB │ │ QPSK 0.5·erfc(√(Eb/N₀)) 10.5 dB │ │ 8-PSK ≈ 2/(3log₂8)·erfc(… 14.0 dB │ │ 16-QAM (3/2)·erfc(√(Eb/5N₀)) 14.4 dB │ │ 64-QAM ≈ (7/12)·erfc(… 18.8 dB │ │ 256-QAM ≈ (15/32)·erfc(… 22.5 dB │ │ │ │ Key insight: Each doubling of M costs ~3-4 dB │ │ in Eb/N₀ but doubles spectral efficiency. │ │ │ └─────────────────────────────────────────────────────┘
7. Probability and Noise
Noise is the fundamental enemy of communication. Understanding its statistical properties is essential for predicting system performance and designing robust receivers.
Gaussian (Normal) Noise Model
┌─────────────────────────────────────────────────────┐ │ │ │ Probability Density Function (PDF): │ │ │ │ 1 -(x-μ)² │ │ p(x) = ─────── · e^(──────────) │ │ σ√(2π) 2σ² │ │ │ │ Where: │ │ μ = mean (expected value) │ │ σ = standard deviation │ │ σ² = variance (noise power) │ │ │ │ Properties: │ │ ∫ p(x) dx = 1 (total probability = 1) │ │ P(μ-σ < x < μ+σ) = 68.3% │ │ P(μ-2σ < x < μ+2σ) = 95.4% │ │ P(μ-3σ < x < μ+3σ) = 99.7% │ │ │ │ Why Gaussian? │ │ Central Limit Theorem: sum of many independent │ │ noise sources approaches Gaussian distribution │ │ │ └─────────────────────────────────────────────────────┘
Thermal Noise (Johnson-Nyquist Noise)
┌─────────────────────────────────────────────────────┐ │ │ │ Thermal noise power in a resistor: │ │ │ │ Pn = kTB │ │ │ │ Where: │ │ k = 1.380649 × 10⁻²³ J/K (Boltzmann) │ │ T = temperature in Kelvin (290 K = 17°C) │ │ B = bandwidth in Hz │ │ │ │ Noise power spectral density: │ │ N₀ = kT = 1.38×10⁻²³ × 290 = 4.00×10⁻²¹ │ │ = -204 dBW/Hz = -174 dBm/Hz │ │ │ │ Practical noise floor calculations: │ │ BW = 1 Hz: -174 dBm │ │ BW = 1 kHz: -144 dBm │ │ BW = 1 MHz: -114 dBm │ │ BW = 20 MHz: -101 dBm (WiFi channel) │ │ BW = 100 MHz: -94 dBm (5G NR channel) │ │ │ └─────────────────────────────────────────────────────┘
Noise Figure (Cascaded Systems)
┌─────────────────────────────────────────────────────┐ │ │ │ Friis Formula for Cascaded Noise Figure: │ │ │ │ (F₂-1) (F₃-1) │ │ F = F₁ + ────── + ─────── + ... │ │ G₁ G₁·G₂ │ │ │ │ Where: │ │ F₁, F₂, F₃ = noise factors (linear) │ │ G₁, G₂ = power gains (linear) │ │ │ │ In dB: │ │ NF(dB) = 10·log₁₀(F) │ │ │ │ Example: Typical WiFi Receiver Chain │ │ ┌─────────┬──────────┬──────────┐ │ │ │ Stage │ NF (dB) │ Gain (dB)│ │ │ ├─────────┼──────────┼──────────┤ │ │ │ Antenna │ 0 │ -1 │ │ │ │ LNA │ 2 │ 15 │ │ │ │ Filter │ 1 │ -1 │ │ │ │ Mixer │ 8 │ -3 │ │ │ │ IF Amp │ 3 │ 20 │ │ │ └─────────┴──────────┴──────────┘ │ │ │ │ F = 1 + (1.58-1)/0.794 │ │ + (1.26-1)/(0.794×31.6) │ │ + (6.31-1)/(0.794×31.6×0.794) │ │ + (2.0-1)/(0.794×31.6×0.794×0.501) │ │ F ≈ 1.83 (2.6 dB total) │ │ │ │ Key: LNA at front dominates system performance! │ │ │ └─────────────────────────────────────────────────────┘
Marcum Q-Function (Radar Detection)
┌─────────────────────────────────────────────────────┐ │ │ │ Marcum Q-function: │ │ │ │ Q_M(a,b) = ∫ x·(x/a)^(M-1)·exp(-(x²+a²)/2) │ │ b │ │ ·I_(M-1)(ax) dx │ │ │ │ Where: │ │ I_(M-1) = modified Bessel function of │ │ order M-1 │ │ M = number of pulses integrated │ │ a = signal parameter (SNR related) │ │ b = detection threshold │ │ │ │ Probability of detection: │ │ Pd = Q_M(√(2·SNR), √(2·Fn)) │ │ │ │ Probability of false alarm: │ │ Pfa = Q_M(0, √(2·Fn)) = e^(-Fn/2) │ │ · Σ (Fn/2)^k / k! (for k=0 to M-1) │ │ │ │ For M=1 (single pulse): │ │ Pd = Q₁(√(2·SNR), √(2·Fn)) │ │ Pfa = e^(-Fn/2) │ │ │ │ Used to determine: │ │ - Required SNR for given Pd and Pfa │ │ - Maximum detection range of radar │ │ - Effect of pulse integration on detection │ │ │ └─────────────────────────────────────────────────────┘
Signal-to-Noise Ratio in Practice
┌─────────────────────────────────────────────────────┐ │ │ │ SNR Definitions: │ │ │ │ SNR = S/N (linear ratio) │ │ SNR(dB) = 10·log₁₀(S/N) │ │ │ │ Eb/N₀ (energy per bit to noise density ratio): │ │ Eb/N₀ = SNR × (B/Rb) │ │ Eb/N₀(dB) = SNR(dB) - 10·log₁₀(Rb/B) │ │ │ │ CNR (carrier-to-noise ratio): │ │ CNR = C/N (carrier power to noise power) │ │ │ │ Required SNR for common systems: │ │ ───────────────────────────────────────────── │ │ Analog AM: 30-40 dB │ │ Analog FM: 10-15 dB │ │ BPSK digital: 10-12 dB (BER 10⁻⁶) │ │ QPSK digital: 10-12 dB (BER 10⁻⁶) │ │ 64-QAM: 24-28 dB (BER 10⁻⁶) │ │ OFDM (WiFi): 15-25 dB │ │ LTE (QPSK): 5 dB │ │ 5G (256-QAM): 28 dB │ │ │ └─────────────────────────────────────────────────────┘
Summary: Key Equations Reference
┌─────────────────────────────────────────────────────┐ │ │ │ FOURIER TRANSFORM │ │ X(f) = ∫x(t)e^(-j2πft)dt │ │ Convolution theorem: time conv = freq multiply │ │ │ │ SAMPLING THEOREM │ │ fs ≥ 2B (Nyquist rate) │ │ ENOB = (SNR - 1.76) / 6.02 │ │ │ │ SHANNON-HARTLEY │ │ C = B·log₂(1 + S/N) │ │ │ │ LINK BUDGET │ │ FSPL(dB) = 20log₁₀(d) + 20log₁₀(f) + 32.45 │ │ Pr = EIRP - FSPL + Gr │ │ Pn = kTB (thermal noise floor) │ │ │ │ MODULATION │ │ AM: s(t) = Ac[1+m·x(t)]cos(2πfct) │ │ FM: BW = 2(Δf + fm) (Carson's rule) │ │ PM: s(t) = Ac·cos(2πfct + kp·x(t)) │ │ │ │ DIGITAL MODULATION │ │ Rs = Rb/log₂(M) │ │ BER(BPSK) = 0.5·erfc(√(Eb/N₀)) │ │ η = Rb/B (spectral efficiency) │ │ │ │ NOISE │ │ p(x) = (1/σ√(2π))e^(-x²/2σ²) (Gaussian) │ │ Pn = kTB = -174 dBm/Hz + 10log₁₀(B) │ │ F_total = F₁ + (F₂-1)/G₁ + ... (Friis) │ │ │ └─────────────────────────────────────────────────────┘
Contributions of These Foundations
These seven mathematical pillars form the complete theoretical framework for communication engineering. From Fourier's insight that signals are sums of sinusoids, through Shannon's hard limits on channel capacity, to the noise models that define receiver sensitivity—every communication system is built on these equations.
The elegance of these foundations lies in their universality. The same equations that govern a 1920s AM radio station also govern 5G mmWave massive MIMO. The math doesn't change—only the parameters do. Higher frequencies, wider bandwidths, more antennas, better codes—but the same Fourier transforms, the same Shannon limits, the same Gaussian noise.
“If I have seen further, it is by standing on the shoulders of giants.”
— Isaac Newton, 1675
Dedicated to Nikola Tesla (1856–1943), whose pioneering work in alternating current, radio, and wireless transmission laid the practical groundwork upon which these mathematical foundations were built.
Timeline
1822Joseph Fourier publishes Théorie analytique de la chaleurFoundation of Fourier analysis for signal decomposition
1843Ohm's law and early signal mathematicsGeorg Ohm applies Fourier analysis to acoustics
1928Harry Nyquist publishes sampling theoryNyquist-Shannon sampling theorem foundations
1948Claude Shannon publishes A Mathematical Theory of CommunicationShannon-Hartley theorem and channel capacity limits
1949Shannon and Hartley formalize C = B·log₂(1 + S/N)Fundamental limit of communication channels
1949Shannon publishes sinc interpolationReconstruction of bandlimited signals from samples
1960Bell Labs develops link budget methodologySystematic approach to RF link design
1966Cooley-Tukey FFT algorithm publishedEfficient computation of discrete Fourier transform
1974QAM modulation formalized for digital communicationQuadrature amplitude modulation becomes standard
1993Turbo codes approach Shannon limitError correction codes within 0.7 dB of theoretical capacity
2007LDPC codes widely adopted in Wi-Fi and LTENear-Shannon-limit performance in practical systems
20205G NR uses advanced LDPC and polar codesModern systems push closer to theoretical limits
Sources & Further Reading
- Shannon, C.E. (1948). A Mathematical Theory of Communication. Bell System Technical Journal, 27(3), 379-423.
- Haykin, S. Communication Systems (5th Edition). Wiley, 2009.
- Proakis, J.G. Digital Communications (5th Edition). McGraw-Hill, 2008.
- Fourier Transform - Wikipedia
- Nyquist-Shannon Sampling Theorem - Wikipedia
- Shannon-Hartley Theorem - Wikipedia
- Link Budget - Wikipedia
- Thermal Noise - Johnson-Nyquist Noise - Wikipedia