The Physics of Multipath Propagation

In a pristine, theoretical vacuum, a radio wave travels from the transmitter to the receiver in a perfectly straight line, experiencing only Free Space Path Loss. In a real-world terrestrial environment—such as a dense urban city—the physical reality is vastly different. The signal transmitted by the base station interacts violently with the physical environment before reaching the mobile receiver.

The electromagnetic waves bounce off skyscrapers (Reflection), bend around the sharp edges of buildings (Diffraction), and scatter off rough surfaces like foliage or vehicles (Scattering). Consequently, the receiver does not capture a single, clean signal. It captures a massive composite of dozens or hundreds of distinct “multipath components,” each arriving from a different angle, with a different amplitude, and after a different physical delay.

Because these multipath components travel different distances, they arrive at the receiving antenna with completely different phases. If two strong multipath components arrive exactly 180 degrees out of phase, they destructively interfere, completely cancelling each other out. The total received signal power plummets to near zero. This phenomenon is known as a Deep Fade. Because the mobile user is constantly moving, these phase relationships change continuously, causing the signal amplitude to fluctuate wildly and unpredictably. This chaotic fluctuation is known as Multipath Fading.

Statistical Models of Fading Channels

Because it is impossible to deterministically map every single reflection in a city, engineers model fading statistically using Probability Density Functions (PDFs). The two most critical models are Rayleigh and Rician fading.

Rayleigh Fading (Non-Line-of-Sight)

Rayleigh Fading models the worst-case scenario in a dense urban environment (like downtown Manhattan).

  • The Assumption: It assumes that the direct Line-of-Sight (LOS) path between the transmitter and receiver is completely blocked by buildings. The receiver only captures the scattered, reflected multipath components.
  • The Mathematics: Because these components arrive from random angles with random phases, their sum is modeled statistically as a complex Gaussian random variable with a mean of zero. The amplitude envelope of this received signal follows the Rayleigh PDF.
  • The Consequence: In a Rayleigh channel, deep fades are frequent and severe. The Bit Error Rate (BER) for a digital modulation scheme like BPSK drops only linearly (inversely proportional to SNR, $P_b \approx \frac{1}{4\gamma}$). No matter how high the transmitter power is increased, deep fades still mathematically occur, dominating the overall error rate. Overcoming Rayleigh fading requires diversity techniques (spatial, frequency, or time) to bypass the fades, rather than simply wasting raw transmitter power.

Rician Fading (Line-of-Sight)

Rician Fading models a more favorable environment (like a rural area or a microcell) where a dominant path exists.

  • The Assumption: It assumes the presence of a strong, deterministic Line-of-Sight (LOS) path in addition to the scattered multipath components.
  • The Mathematics: Because this dominant path is deterministic, the complex Gaussian random variable modeling the signal no longer has a zero mean. The amplitude envelope follows the mathematically more complex Rician PDF.
  • The K-Factor: The severity of Rician fading is quantified by the K-factor, defined as the ratio of the power in the dominant LOS path to the variance (power) of the scattered multipath components ($K = P_{LOS}/P_{NLOS}$). The K-factor dictates the shape of the PDF. If $K$ approaches infinity, the channel approaches a pristine AWGN channel (perfect LOS, zero scattering). If $K = 0$, it implies there is absolutely zero power in the LOS path. Mathematically, substituting $K=0$ into the Rician PDF equations causes the Bessel functions to collapse, reducing the equation perfectly to the Rayleigh PDF. Rayleigh is simply a special, worst-case instance of Rician fading.

Delay Spread and Frequency Selective Fading

Multipath propagation not only affects signal amplitude (fading) but also causes time dispersion, severely limiting the maximum data rate of the channel.

Delay Spread and Coherence Bandwidth

When a transmitter sends a sharp pulse, the primary LOS path arrives first. The reflections arrive microseconds later, “smearing” the pulse in the time domain.

  • Delay Spread ($\sigma_\tau$): This metric measures the time difference between the arrival of the very first Line-of-Sight signal component and the arrival of the last significant reflected multipath component.
  • Coherence Bandwidth ($B_c$): This is the frequency-domain counterpart to Delay Spread, and they are inversely proportional ($B_c \approx 1/(5\sigma_\tau)$). Coherence Bandwidth defines a specific frequency range over which the channel response is “flat” (all frequencies within it experience identical fading).

Frequency Selective Fading vs. Flat Fading

The relationship between the signal bandwidth ($B_s$) and the Coherence Bandwidth ($B_c$) determines the type of fading the signal experiences.

  • Flat Fading: Occurs when $B_s \ll B_c$. The signal is very narrow. All frequency components of the signal experience the exact same fading amplitude and phase shift. The signal shape is preserved, it simply gets weaker or stronger overall.
  • Frequency Selective Fading: Occurs when $B_s > B_c$. The signal is so wide that different frequency components of the signal experience completely different fading characteristics. Some parts of the signal might be deeply faded while others are amplified, violently distorting the signal shape.

Inter-Symbol Interference (ISI)

Frequency Selective Fading inevitably causes Inter-Symbol Interference (ISI), a catastrophic distortion. Because $B_s > B_c$, the reciprocal relationship means the duration of a single transmitted symbol ($T_s$) is strictly less than the Delay Spread ($\sigma_\tau$). Consequently, the transmitter fires Symbol 2 before the delayed multipath reflections of Symbol 1 have finished arriving at the receiver. The delayed echoes of Symbol 1 physically overlap and smash into the primary LOS path of Symbol 2. The receiver cannot distinguish between the symbols, destroying data integrity. Overcoming ISI requires complex, battery-draining equalizers or the adoption of Orthogonal Frequency Division Multiplexing (OFDM).

Doppler Spread and Fast Fading

While Delay Spread is caused by the physical environment, Doppler Spread is caused entirely by the relative velocity of the mobile receiver.

The Doppler Effect

When a mobile phone moves toward a base station, the received radio waves are physically compressed, increasing the apparent frequency. When it moves away, the waves stretch, decreasing the frequency. This frequency shift is the Doppler Effect ($f_d = \frac{v}{\lambda} \cos(\theta)$). Because multipath components arrive from all angles simultaneously, they all experience different Doppler shifts. Some are shifted higher, some lower. This broadens the transmitted spectral line into a wider bandwidth known as the Doppler Spread ($B_D$).

Coherence Time

Doppler Spread is inversely proportional to Coherence Time ($T_c$). Coherence Time is the statistical measure of the time duration over which the channel impulse response is considered invariant (i.e., how long the channel remains stable before it changes).

Fast Fading vs. Slow Fading

The relationship between the symbol duration ($T_s$) and the Coherence Time ($T_c$) determines how fast the fading occurs relative to the data.

  • Slow Fading: Occurs when $T_s \ll T_c$. The mobile is moving slowly (e.g., walking). The channel changes very slowly compared to the data rate. A single deep fade might wipe out thousands of contiguous bits in a massive burst error, requiring complex interleaving to fix.
  • Fast Fading: Occurs when $T_s > T_c$. The mobile is moving at extreme velocity (e.g., on a bullet train). The Doppler spread is massive, and the channel changes violently within the duration of a single symbol. The phase of the channel coefficient rotates rapidly and unpredictably.

The Failure of Coherent Detection

Fast Fading completely destroys Coherent Detection schemes (like standard QAM). Coherent detection requires the receiver to accurately estimate the absolute phase shift introduced by the channel using pilot tones. If the channel phase rotates wildly within a single symbol duration due to high velocity, maintaining an accurate, absolute channel estimate becomes physically impossible. The receiver guesses wrong, and the BER skyrockets to 50%.

To survive Fast Fading, networks must deploy Differential Phase Shift Keying (DPSK). DPSK abandons channel estimation entirely. It encodes the data in the difference in phase between the current symbol and the previous symbol. It only assumes the channel remains relatively constant across two consecutive symbols, making it highly robust to rapid, chaotic phase rotations where Coherent systems collapse, albeit at the cost of a 3 dB SNR penalty.