1. Frequency-Selective Fading vs.
Flat Fading
Flat Fading Channel
Occurs when the channel's bandwidth is larger than the signal's bandwidth.
All frequency components of the signal experience the same level of fading.
The channel can be represented as a single gain (amplitude scaling) and phase shift.
Example: Narrowband signals in environments with short delay spreads.
Frequency-Selective Fading Channel
Occurs when the signal's bandwidth is larger than the channel's coherence bandwidth.
Different frequency components experience different levels of fading.
This happens because the channel introduces frequency-dependent fading due to varying delays from multipath
propagation.
The signal might experience intersymbol interference (ISI) due to the overlapping of symbols caused by delayed
paths.
Example: Broadband signals in environments with long delay spreads.
Key Difference: Flat fading affects all frequencies uniformly, while frequency-selective fading affects them differently or
selectively.
2. Zero-Forcing Equalizer
A zero-forcing (ZF) equalizer is a signal processing technique used to mitigate the effects of fading, particularly for
frequency-selective channels.
How It Works
The equalizer inverts the channel's frequency response to remove the distortion caused by the channel.
It attempts to force the combined channel and equalizer response to be flat (ideal).
Limitation: It amplifies noise when the channel response is weak (low SNR), potentially degrading performance
in such cases.
Modeling
Modeling a Frequency-Selective Channel
1. Impulse Response: Represent the channel as a finite impulse response (FIR) filter with multiple taps, where each
tap corresponds to a different path with a specific delay and attenuation.
2. Frequency Response: Compute the frequency response of the channel from the impulse response FFT using fast
Fourier transform (FFT).
3. Effect on Signal: Pass the transmitted signal through the channel filter to observe how different frequencies are
affected by multiplicating both in frequency domain.
4.
Modeling a Zero-Forcing Equalizer
1. Channel Estimation: Estimate the channel's impulse response or frequency response (e.g., using a known pilot
signal).
2. Inverse Filter Design: Design an inverse filter to compensate for the channel's effect.
3. Application: Apply the inverse filter to the received signal to recover the original transmitted signal.
Comparison with Flat Fading
For flat fading:
Use a single complex scalar to represent the channel effect.
No need for a sophisticated equalizer like ZF, as all frequency components are uniformly affected.
Modeling the Impulse Response of a Frequency-Selective Channel
The impulse response of a frequency-selective channel represents the multipath propagation of the
transmitted signal. It is typically modeled as a discrete-time random process. Simply put, the channel
is modeled as an FIR filter with L taps (or paths), where L corresponds to the number of significant
multipath components. These taps are modeled as complex Gaussian random variables.
Frequency-Selective Fading and OFDM Subcarriers
Coherence Bandwidth and Subcarrier Spacing
In a frequency-selective fading channel:
o The coherence bandwidth ( Bc ) of the channel defines the range of frequencies over which the
channel's frequency response can be considered approximately constant.
o Subcarriers in OFDM are spaced closely in frequency (subcarrier spacing Δf =1/T , where T is
the OFDM symbol duration).
Relation Between Bc and Δf
If the subcarrier spacing Δf is much smaller than the coherence bandwidth ( Δf ≪ B c):
o Several adjacent subcarriers will experience approximately the same channel gain.
o This is a common scenario in practical OFDM systems, ensuring that the channel is locally flat
across small frequency ranges.
If Δf is comparable to or larger than Bc :
o Each subcarrier could experience a significantly different channel gain, leading to complex
equalization requirements.
Practical Assumption for OFDM
In practice:
The frequency-selective fading is modeled as a piecewise constant function over the OFDM
subcarriers.
The channel is typically constant across a group of subcarriers that lie within the coherence
bandwidth.
This grouping simplifies channel estimation and equalization since the channel does not need to be
individually estimated for every single subcarrier.
