ML Detector
ML Detector
ML Detector
Detection
Channel
ML Detection
Error Probability
SPSC
Maximum Likelihood Sequence Detection 1
Channel
Delay Spread
time dispersion
intersymbol interference (ISI).
frequency selective fading
Channel Model
passband PAM
baseband PAM
SPSC
Maximum Likelihood Sequence Detection 2
Channel Model
SPSC
Maximum Likelihood Sequence Detection 3
Discrete-Time Equivalent
Channel Model for PAM
∞
⎡ ⎛ 2π ⎞ ⎤ ⎡ ⎛ 2π ⎞ ⎤ ⎡ ⎛ 2π ⎞ ⎤
P (e jωT
)= ∑
m =−∞
G ⎢ ⎜
j
⎣ ⎝
ω − m ⎟⎥ E ⎢ ⎜
B
T ⎠⎦ ⎣ ⎝
j ω − m ⎟⎥ ⎢ ⎜
F
T ⎠⎦ ⎣ ⎝
j ω − m ⎟⎥
T ⎠⎦
∑ F ( j (ω + m ) )
∞ 2
jωT 2 N0 2π
S Z (e )= T
T m =−∞
SPSC
Maximum Likelihood Sequence Detection 4
Matched Filter as
Receiver Front End (1)
matched filter as receive filter
discrete-time equivalent channel model
SPSC
Maximum Likelihood Sequence Detection 5
Matched Filter as
Receiver Front End (2)
autocorrelation of baseband receive
pulse shape h(t)
∞
ρ h (k ) = ∫
−∞
h(t )h* (t − kT )dt
folded spectrum
(( ))
∞
1 ∞ 2
S h (e jωT ) = ∑
k =−∞
ρ h (k )e jω kT = ∑ H j ω+m T
T m =−∞
2π
SPSC
Maximum Likelihood Sequence Detection 6
Whitening of the
Matched Filter (1)
matched filter Î colored noise
jωT jωT
S Z (e ) = 2 N 0 S h (e )
spectral factorization
S h ( z ) = γ M ( z ) M (1/ z )
2 ∗ ∗
SPSC
Maximum Likelihood Sequence Detection 7
Whitening of the
Matched Filter (2)
equalization with inverse minimal phase filter
SPSC
Maximum Likelihood Sequence Detection 8
Detection
ML Detection of a Single Symbol
ML Detection of a Signal Vector
ML Detection with Intersymbol
Interference
Sequence Detection
Markov Chains
Markov Chain Signal Generator
The Viterbi Algorithm
SPSC
Maximum Likelihood Sequence Detection 9
Detection
Estimation Î transmitted signal is contiuous-
valued
Detection Î transmitted signal is discrete-
valued
Model for detection:
SPSC
Maximum Likelihood Sequence Detection 10
ML Detection of a Single Symbol
Special case of MAP detector if p A (â ) = const
pY | A ( y | â ) p A (â )
p A|Y (â | y ) =
pY ( y )
ML chooses â ε Ω A
To maximize likelihood pY | A ( y | â )
Measure of the quality Pr[error ] = Pr[ â ≠ a ]
SPSC
Maximum Likelihood Sequence Detection 11
ML Detection of a Signal Vector
Vector of Symbols
Maximize f Y|S ( y | sˆ ) = f N|S ( y − sˆ | sˆ ) = f N ( y − sˆ )
Equivalent to maximize
1 ⎛ 1 2⎞
f N (y − sˆ ) = exp ⎜ − 2 y − sˆ ⎟
(2π ) M / 2 σ M ⎝ 2σ ⎠
Equivalent to minimizing y − sˆ
SPSC
Maximum Likelihood Sequence Detection 12
ML Detection With Intersymbol
Interference (1)
Y = hA + N
SPSC
Maximum Likelihood Sequence Detection 13
ML Detection With Intersymbol
Interference (2)
ML minimizes distance âh and
observation y
y − hâ = y − 2 y , h â + h â
2 2 2 2
Equivalent to maximizing
2 y, h â − h â
2 2
y, h = ∑ ym hm = [ yk * h− k ]k =0
m
SPSC
Maximum Likelihood Sequence Detection 14
ML Detection With Intersymbol
Interference (3)
SPSC
Maximum Likelihood Sequence Detection 15
ML Detection With Intersymbol
Interference (4)
Exponential complexity
Message of K M-ary symbols
Î MK matched filters
Î MK comparisons
SPSC
Maximum Likelihood Sequence Detection 16
Sequence Detection
Markov Chains
Markov Chain Signal Generator
The Viterbi Algorithm
SPSC
Maximum Likelihood Sequence Detection 17
Markov Chains (1)
SPSC
Maximum Likelihood Sequence Detection 18
Markov Chains (2)
Trellis diagram
Node
Branch
Path
SPSC
Maximum Likelihood Sequence Detection 19
Markov Chain
Signal Generator (1)
Sequence of homogenous Markov
chain states Ψ k
State transitions S k = g (ψ k ,ψ k +1 )
M
Observation function g (ψ k ,ψ k +1 ) = ∑ hi Ak −1
i =0
State of shift-register
Ψ k = [ X k −1 , X k − 2 ,..., X k − M ]
SPSC
Maximum Likelihood Sequence Detection 20
Markov Chain
Signal Generator (2)
ISI Model
Shift-register process
SPSC
Maximum Likelihood Sequence Detection 21
Markov Chain
Signal Generator Example
hk = δ k + 0.5δ k −1
SPSC
Maximum Likelihood Sequence Detection 22
The Viterbi Algorithm (1)
SPSC
Maximum Likelihood Sequence Detection 23
The Viterbi Algorithm (2)
SPSC
Maximum Likelihood Sequence Detection 24
The Viterbi Algorithm (3)
SPSC
Maximum Likelihood Sequence Detection 26
The Viterbi Algorithm
Example (2)
SPSC
Maximum Likelihood Sequence Detection 27
The Viterbi Algorithm
Example (3)
SPSC
Maximum Likelihood Sequence Detection 28
Error Probability Calculation
Error Event
Detection Error
Upper Bound of Detection Error
Lower Bound of Detection Error
Symbol Error Probability
SPSC
Maximum Likelihood Sequence Detection 29
Error Event
(a) length 1,
metric from real sequence 1.25
(b) length 2,
metric from real sequence 3.5
SPSC
Maximum Likelihood Sequence Detection 30
Detection Error
Pr[e] = Pr [ Ψ ] Pr ⎡⎣ Ψ
ˆ Ψ⎤
⎦
w(e) … total number of detection errors
in error event e
Pr[e] depends on real path and chosen
path Î estimate Pr ⎡ Ψˆ Ψ⎤
⎣ ⎦
SPSC
Maximum Likelihood Sequence Detection 31
Upper Bound of
Detection Error (1)
Pr ⎡⎣ Ψ (
ˆ | Ψ ⎤ ≤ Q d (Ψ
⎦
ˆ , Ψ ) / 2σ )
Q … cumulative probability distribution
d …Euclidian distance of real an chosen
path
SPSC
Maximum Likelihood Sequence Detection 32
Upper Bound of
Detection Error (2)
Pr[detection error] ≤ ∑ w(e) Pr[ Ψ ]Q ( d min / 2σ ) + other terms
eε E
R = ∑ w(e) Pr[ Ψ ]
eε B
SPSC
Maximum Likelihood Sequence Detection 33
Lower Bound of
Detection Error (1)
SPSC
Maximum Likelihood Sequence Detection 34
Lower Bound of
Detection Error (2)
Using total probability
Pr[detection error] ≥ ∑ Pr[Ψ ]Q (d min (Ψ ) / 2σ )
Ψ
P= ∑ε Pr[Ψ ]
Ψ A
SPSC
Maximum Likelihood Sequence Detection 35
Symbol Error Probability (1)
SPSC
Maximum Likelihood Sequence Detection 36
Symbol Error Probability (2)
SPSC
Maximum Likelihood Sequence Detection 37