sin(2πhβ(t))
, 0 ≤ t < 4T
sin πh
= xn−3 − xn−1
This document provides an overview of continuous phase modulation (CPM) techniques including:
1) CPM modulation generates a signal where the phase changes continuously according to an excess phase function.
2) Various frequency and phase shaping pulses can be used to determine the excess phase, including rectangular, raised cosine, half sinusoid, and triangular pulses.
3) Minimum shift keying (MSK) is a special case of CPM where the modulation index h = 1/2, and the excess phase changes linearly between symbol intervals
sin(2πhβ(t))
, 0 ≤ t < 4T
sin πh
= xn−3 − xn−1
This document provides an overview of continuous phase modulation (CPM) techniques including:
1) CPM modulation generates a signal where the phase changes continuously according to an excess phase function.
2) Various frequency and phase shaping pulses can be used to determine the excess phase, including rectangular, raised cosine, half sinusoid, and triangular pulses.
3) Minimum shift keying (MSK) is a special case of CPM where the modulation index h = 1/2, and the excess phase changes linearly between symbol intervals
sin(2πhβ(t))
, 0 ≤ t < 4T
sin πh
= xn−3 − xn−1
This document provides an overview of continuous phase modulation (CPM) techniques including:
1) CPM modulation generates a signal where the phase changes continuously according to an excess phase function.
2) Various frequency and phase shaping pulses can be used to determine the excess phase, including rectangular, raised cosine, half sinusoid, and triangular pulses.
3) Minimum shift keying (MSK) is a special case of CPM where the modulation index h = 1/2, and the excess phase changes linearly between symbol intervals
sin(2πhβ(t))
, 0 ≤ t < 4T
sin πh
= xn−3 − xn−1
This document provides an overview of continuous phase modulation (CPM) techniques including:
1) CPM modulation generates a signal where the phase changes continuously according to an excess phase function.
2) Various frequency and phase shaping pulses can be used to determine the excess phase, including rectangular, raised cosine, half sinusoid, and triangular pulses.
3) Minimum shift keying (MSK) is a special case of CPM where the modulation index h = 1/2, and the excess phase changes linearly between symbol intervals
xn {1, 3, . . . , (M 1)} are the M -ary data symbols
hf (t) is the frequency shaping pulse of duration LT , that is zero for
t < 0 and t > LT , and normalized to have an area equal to 1/2. Full response CPM has L = 1, while partial response CPM has L > 1. The instantaneous frequency deviation from the carrier is
1 d(t) X fdev (t) = =h xk hf (t kT ) . 2 dt k=0
Frequency Shaping Pulses
hf (t)
pulse type
L-rectangular (LREC)
1 u (t) 2LT LT
L-raised cosine (LRC)
1 2LT
L-half sinusoid (LHS)
4LT
sin(t/LT )uLT (t)
L-triangular (LTR)
1 LT
1 cos
2t LT
|tLT /2| LT /2
uLT (t)
Excess Phase and Tilted Phase
During the time interval nT t (n + 1)T , the excess phase (t) is (t) = 2h
n X
k=0
xk (t kT ).
where the phase shaping pulse is
R t 0
(t) =
hf ( )d 1/2
, t<0 , 0 t LT , t LT
For the case of full response CPM (L = 1), during the time interval nT t (n + 1)T the excess phase is (t) = h
n1 X k=0
xk + 2hxn(t nT )
During the time interval nT t (n + 1)T , the CPM tilted phase is
(t) = h
n1 X k=0
xk + 2hxn(t nT ) + h(M 1)t/T
= (t) + h(M 1)t/T
Note that s(t) can be generated by replacing (t) with (t) and fc by fc h(M 1)t/2T in (1). 4
Continuous Phase Frequency Shift Keying (CPFSK)
Continuous phase frequency shift keying (CPFSK) is a special type of CPM
that uses the shaping function hf (t) =
1 1 uT (t) = (u(t) u(t T )) 2T 2T
As a result
0 (t) = t/2T
1/2
, t<0 , 0tT , tT
Since the shaping function is rectangular, the CPFSK excess phase trajectories are linear.
+1
4 h 3 h
-1
2 h
h 0
- h
-2 h -3 h -4 h
Phase tree of binary CPFSK.
t/T
Phase-state Diagrams
-1
-1 +1
+1
-1
-1 +1
+1
+1
+1
-1
-1 +1
+1
-1
-1
Phase-state diagram of CPM with h = 1/4.
Minimum Shift Keying (MSK)
MSK is a special case of continuous phase frequency shift keying (CPFSK), where the modulation index h = 21 is used. The phase shaping pulse is
0 (t) = t/2T
1/2
, t<0 , 0tT , tT
The MSK bandpass waveform is
n1 t
nT
X xk + xn , nT t (n + 1)T s(t) = A cos 2fct + + 2 k=0 2T The excess phase on the interval nT t (n + 1)T is n1 t nT X (t) = xk + xn 2 k=0 2T The tilted phase on the interval nT t (n + 1)T is t (t) = (t) + 2T Combining the above two equations, we have
((n + 1)T ) = (nT ) + (1 + xn)
2 8
( t ) (Mod 2 ) 3 /2
/2
0 1
excess phase trellis diagram for MSK.
t/T
Example: MSK (h=1/2)
excess phase h
2 +1 +1
-1 1 +1
-1
+1 1 -1 -1 +1
-1
+1 +1
-1
+1 -1 -1 +1
-1
+1 +1
-1
+1 -1 -1 +1
-1
4 t t 4-states
t/T
tilted phase
+1 -1
0 0
-1 +1 +1 -1
-1 +1 +1 -1
-1 +1 +1 -1
-1 +1 +1 -1
-1 +1 1 +1 -1
2-states 6
t/T 0
Wireless Systems Laboratory
Excess Phase and Tilted Phase, example MSK (h = 1/2).
10
-1 +1
+1
+1
+1
-1
-1
-1
Phase state diagram for MSK signals.
11
Linearized Representation of MSK
An interesting representation for MSK waveforms can be obtained by using Laurents decomposition to express the MSK complex envelope in the quadrature form X s(t) = A b(t 2nT, xn) , n
where
b(t, xn) = x2n+1ha(t T ) + j x2nha(t)
and where xn = ( x2n+1, x2n),
x2n = x2n1 x2n
x2n+1 = x2nx2n+1 x1 = 1 and
(2) (3) (4)
t ha(t) = sin u2T (t) . 2T
The sequences, { x2n} and { x2n+1}, are independent binary symbol sequences taking on elements from the set {1, +1}. The symbols x2n and x2n+1 are transmitted on the quadrature branches with a half-sinusoid (HS) amplitude shaping pulse of duration 2T seconds and an offset of T seconds. 12
Gaussian MSK (GMSK)
x(t)
g(t)
h(t)
FM Modulator
s(t)
Gaussian Pre-modulation filtered MSK (GMSK).
With MSK the modulating signal is x(t) =
1 2T
n=
xnuT (t nT )
We filter x(t) with a low-pass filter to remove high frequency content prior to modulation, i.e., g(t) = x(t) h(t). For GMSK, the low-pass filter transfer function is 2
f ln 2
H(f ) = exp
B 2
where B is the 3 dB bandwidth.
13
A rectangular pulse rect(t/T ) = uT (t + T /2) transmitted through this filter
yields the GMSK frequency shaping pulse
(BT )2x2
1 2 2 t/T +1/2 (BT ) t/T 1/2 exp
2T ln 2 ln 2
1 t/T 1/2 t/T + 1/2
Q Q
= 2T
hf (t) =
where
v u u u t
dx Correction!
1 2 ex /2dx
2 ln 2 2 = . 4 2(BT )2
Q() =
the total pulse area is
hf (t)dt = 1/2 and, therefore, the total contribution to the excess phase for each data symbol is /2. R
14
0.4 BT=0.2 BT=0.25 BT=0.3
0.35 0.3
hf(t)
0.25 0.2 0.15 0.1 0.05 0 0
t/T
GMSK frequency shaping pulse for various normalized filter bandwidths BT .
15
The GMSK phase shaping pulse is
1 t 1 t 1 (t) = hf (t)dt = G + G 2 T 2 T 2 Z
where
x! x22 G(x) = x + e 2 ,
and () =
1 2 ex /2dx
Observe that () = 1/2 and, therefore, the total contribution to the excess phase for each data symbol remains at /2.
16
0.5
0.4
(t)
0.3
0.2 BT=0.2 BT=0.25 BT=0.3
0.1
0 0
2 t/T
GMSK phase shaping pulse for various normalized filter bandwidths BT .
17
The excess phase change over the interval from T /2 to T /2 is
(T /2) (T /2) = x00(T ) + where n(T ) = and
T /2nT
T /2nT
n= n6=0
xnn(T )
hf () d .
1 t/T 1/2 t/T + 1/2 Q Q
hf (t) = 2T
The first term, x00(T ) is the desired term, and the second term,
P n= n6=0
xnn (T ),
is the intersymbol interference (ISI) introduced by the Gaussian low-pass
filter. Conclusion: GMSK trades off power efficiency for a greatly improved bandwdith efficiency.
18
Svv(f) (dB)
-20 -40 -60 MSK BT = 0.2 BT = 0.25 BT = 0.3
-80 -100 -120
0.0
0.5
1.0 1.5 Frequency, fT
2.0
2.5
Power spectral density of GMSK with various normalized filter bandwidths BT .
19
Linearized Gaussian Minimum Shift Keying
(LGMSK) Laurent showed that any binary partial response CPM signal can be represented exactly as a linear combination of 2L1 partial-response pulse amplitude modulated (PAM) signals, viz., s(t) =
n=0
2L1 X1 p=0
ejhn,p cp(t nT ),
where cp (t) = c(t) n,p =
n X
m=0
L1 Y n=1
c (t + (n + Ln,p)T ) ,
xm
L1 X
m=1
xnm m,p,
and n,p {0, 1} are the coefficients of the binary representation of the index p, i.e., p = 0,p + 21,p + + 2L2 L2,p . The basic signal pulse c(t) is c(t) =
sin(2h(t)) sin h sin(h2h(tLT ))
sin h
, 0 t < LT , LT t < 2LT , otherwise
where (t) is the CPM phase shaping function.
20
Linearized Gaussian Minimum Shift Keying
(LGMSK) note that the GMSK frequency shaping pulse spans approximately L = 4 symbol periods for practical values of BT . Often the pulse c0 (t) contains most of the signal energy, so the p = 0 term in can provide a good approximation to the CPM signal. Numerical analysis can show that the pulse c0(t) contains 99.83% of the energy and, therefore, we can derive a linearized GMSK waveform by using only c0 (t) and neglecting the other pulses. This yields the waveform s(t) =
n=0
ejhn,o c0(t nT ),
where, with L = 4, c0(t) = n,0 =
3 Y
n=0 n X
c (t + nT ) ,
m=0
xm
Since the GMSK phase shaping pulse is non-causal, when evaluating c(t) we use the truncated and time shifted GMSK phase shaping pulse = (t 2T ) (t) with L = 4 as shown previously.
21
1 BT=0.2 BT=0.25 BT=0.3
0.9 0.8 0.7
c0(t)
0.6 0.5 0.4 0.3 0.2 0.1 0 0
t/T
LGMSK amplitude shaping pulse for various normalized premodulation filter bandwidths BT .
22
Linearized Gaussian Minimum Shift Keying
(LGMSK) For h = 1/2 used in GMSK,
an,0 = ej 2 n,0 {1, j} ,
and it follows that s(t) = A
x2n+1c0(t 2nT T ) + j x2n c0(t 2nT )
where x2n = x2n1 x2n x2n+1 = x2nx2n+1 x1 = 1 This is the same as the OQPSK representation for MSK except that the half-sinusoid amplitude pulse shaping function is replaced with the LGMSK amplitude pulse shaping function. Note that the LGMSK pulse has length of approximately 4T , while the pulses on the quadrature branches are transmitted every 2T seconds. Therefore, the LGMSK pulse introduces ISI.
