Path Loss

Multipath Fading Multipath fading

Instructor: M.A. Ingram

ECE4823
Interference
Shadowing

Excess Delay Strength Variation

„ The propagation delay relative to that of the „ As the vehicle moves over a short distance, the
shortest path strength of each path varies because the surfaces are
complex

τ=8µs

τ=0
τ=47µs

Measured Data from

The Channel is a Filter Darmstadt, Germany [Molisch, ’01]

„ The multipath channel can be

represented as linear, time-varying
bandpass filter
Transmitted Received
Signal Signal
x(t) h(t ,τ ) y(t)
+∞
y (t ) = ∫ x(t − τ )h(t ,τ )dτ
−∞

1
 Baseband Impulse Response Path Model
„ More convenient to work with baseband „ The channel is assumed to comprise N
signals discrete paths of propagation (rays)
Each path has an amplitude α(t), a phase θ(t)
{ }
„

h(t ,τ ) = Re hb (t ,τ )e jω ct and a propagation delay τ

x(t ) = Re{ c(t )e } jω c t hb (t ,τ )

N =5
y (t ) = Re{ r (t )e } jω c t α 0 ( t ) e jθ ( t )
0

The factor of ½
α1 (t )e jθ ( t )α (t )e jθ
1
2 (t )
+∞ ensures that baseband 2
α 3 (t )e jθ ( tα) 4 (t )e jθ
1 4 (t )
3

average power
r (t ) = ∫ c(t − τ )hb (t ,τ )dτ equals passband
τ0 τ3 τ4
delay
2 −∞ average power τ1 τ2

Probing the Channel Pulse Width >> τmax

„ The channel may be probed or “sounded” by „ Suppose a pulse much wider than the length
transmitting a pulse p(t) and recording the of the impulse response is transmitted at time
response at the receiver t=0
p(t)
„ The response is the convolution of p(t) with γ
the channel impulse response
t
1 N −1 hb ( t , τ )
Tp
r (t ) = ∑α i (t )e jθi (t ) p(t − τ i )
2 i =0
τ 0 τ1 τ2 τ3 τ 4 delay

Time Variation of the Probe

Instantaneous Power Response
The magnitude squared of any sample in the interval τ4 and „ If one or both of the terminals moves, the path
τ0 + Tp will equal 2 phases change because the path lengths change
γ2 N −1

∑α (t )e
2 jθ i ( t )
r (t ) = i „ The path amplitudes do not change much
4 i =0 These changes yield large changes in the magnitude
r(t) „
of the received waveform

t |r(t)| in dBm [not real data]

hb ( t , τ )
We say, “the multipath is not resolved”

τ 0 τ1 τ2 τ3 τ 4 delay t

2
 Average Power for
Narrowband Fading Narrowband Signals
„ This same type of fading happens to a „ Assuming the channel is ergodic, the
digital waveform if the symbol period is ensemble average may be approximated by a
much larger than (>10 times) the time average:
T
t+

{ }
2
channel “length” 1 2
γ2 N

∫ ∑ α ( s )e
2 jθ i ( s )
Eαθ r (t ) ≈ ds
„ Such long symbol periods correspond to T T 4 i =0
i
t−
“narrowband” signals 2
where the interval [t-T/2,t+T/2] corresponds
to a local area

Uncorrelated Scattering Pulse Width << τmax

„ Assume that the phases of different paths are „ Now consider a small pulse width
uncorrelated p(t)
„ Then the time average simplifies to γ

{ }≈ ∑α
N
2 2
Eαθ r (t ) i Tp
t
i =0
hb (t ,τ )
where
T
t+
2
1
α ≈ ∫α
2 2
i i ( s )ds
T T delay
t− τ0 τ1 τ2 τ3 τ4
2

Multipath Resolved Wideband Signal

„ Pulses do not overlap „ The Fourier Transform of such a narrow
pulse has a wide spectrum
r (t ,τ ) P(f)
p(t)
γα 0 (t )e jθ ( t )
0

γα1 (t )e jθ (γα
t)
1

2 (t )e
jθ 2 (t ) γ
γα 3 (t )e jθ (γα
3t)
jθ 4 (t )
4 (t )e F.T.
τ Tp
t
τ0 τ1 τ2 τ3 τ4

3
 Average Power for Wideband
Power Delay Profile (PDP) Signals
„ The PDP is a time-average of |r(t,τ)|2 over a „ The average power is the integral of the
small interval (assuming the terminal is PDP
moving) t+
T
+∞
2
1
P (τ ) ≈ ∫ r (s,τ ) ds
2 PAVG = ∫ P(τ )dτ
T T 0
P(τ ) t−
N −1
γ 2α 02 2
γ 2α12
γ 2α 22
= ∑ α i2
γ 2α 32 γ α 2 2
i =0
4

τ0 τ1 τ τ3 τ 4
τ
2

Local Average Powers Are The

Same Moments of the PDP
„ Narrowband and Wideband averaged „ Channels are often described by their
powers are equal rms delay spread
„ To compute rms delay spread,
normalize the PDP to make it like a PDF
for a random variable (unit area) and
then find its standard deviation
„ Must you use excess delay to compute
rms delay spread?

Mean Delay Second Moment

„ Must first compute the mean delay „ Next need the second moment of this
+∞
“PDF”
∫ τ P(τ )dτ +∞

∫0 τ P(τ )dτ
2
τ= 0
+∞
τ = +∞
2

∫ P(τ )dτ
0
0
∫ P(τ )dτ
For this to be mean excess delay, the origin
of the τ axis needs to be the time of the first
arriving path

4
RMS Delay Spread Example Data [Rappaport, ’02]

„ Recall that standard deviation is the

square root of variance and variance is
the second moment minus the first
moment squared

Variance ()
σ τ2 = τ 2 − τ
2

− (τ )
2
rms delay spread στ = τ 2

How RMS Delay Spread Can

Be Used The Frequency Domain View
„ If στ<<symbol period, assume „ στ<<symbol period implies that the
“narrowband” fading effects frequency response of the channel,
„ If στ>>symbol period, assume +∞
H (t , f ) = ∫ h(t ,τ ) exp(− j 2πfτ )dτ ,
“wideband” fading effects (will need an −∞
equalizer, CDMA or OFDM) doesn’t vary much with frequency over
the bandwidth of the transmitted signal

Narrowband Case Flat Fading

„ The channel “appears flat to the signal” „ When στ<<symbol period, we say the
H (t , f ) Transmitted signal undergoes flat fading
Signal Spectrum
„ The channel frequency response is
approximately flat over the signal
bandwidth

f
900ΜΗz

5
Wideband Case Frequency Selective Fading
„ στ>> symbol period implies that the „ When στ>> symbol period, we say the
frequency response of the channel varies signal undergoes frequency selective
significantly with frequency over the fading
bandwidth of the transmitted signal
„ The channel frequency response is
H (t , f ) Transmitted
Signal Spectrum strong for some frequencies and not for
others within the signal bandwidth

f
900ΜΗz

Summary References
The multipath channel model has a discrete number
„
of propagation paths „ [Rapp, ’02] T.S. Rappaport, Wireless
„ Each path has amplitude, phase and delay Communications, Prentice Hall, 2002
The PDP is the local average of the magnitude
[Molisch, ’01] Andreas F. Molisch (ed),
„
squared of the impulse response of the channel „
„ Average power of the channel is the integral of the Wideband Wireless Digital
PDP
„ Average power is same for narrowband and Communications, Prentice Hall PTR,
wideband channels 2001.
„ The fading is “flat” or “frequency selective”
depending on the comparison between rms delay
spread and the symbol period