IEEE TRANSACTIONS ON COMMUNICATIONS, VOL 36. NO.
5, MAY 1988
Digital Filter and Square Timing Recovery
MARTIN OERDER, STUDENT MEMBER, IEEE, AND
A bstract-Digital realizations of timing recovery circuits for digital
data transmission are of growing interest. In this paper, we present a new
digital algorithm which can be implemented very efficiently also at high
data rates. The resulting timing jiller has been computed and verified by
simulations. In contrast to otber known algorithms, tbe one presented
here allows free running sampling oscillators and a new planar filtering
metbod whicb prevents synchronization hangups.
1. INTRODUCTION
D
IGI TAL realizations of reccivcrs for synchronous data
signals-baseband as well as QPSK or QAM signal s - ar e
of growing interest as the capabilities of signal processors (for
low data rates) and application spccific intcgratcd circuits (for
high data rates) increase. These receivers have to include
al go r ithms for timing recovery. Several such discrete-time
algorithms have been proposed during the last few years [1]
[3]. The majority of these solutions, howcvcr, include only the
integration of one part of the ti ming synchronization, namely,
the generation of some kind of timing error signal, into the
digital part of the receiver. This error signal is then typically
used to control an analog veo which generates thc sampling
strobes.
Due to the advantages of an integrated realization, however,
as much of the receiver as possible should be digital. This
means that the input signal shoul d b e sampled at a fixed rate by
a free running oscillator and all further processing should then
be done digitally using these samples. For symbol detection,
this means that thc optimum decision metrics must be
generated from the given samples by some sort of interpolation
which is controlled by an estimate of the current timing offset
[4]. Therefore, wc need an algorithm which determines this
absolute timing offs et (not only a timing error signal) from thc
gi ven samples of the signal.
Such an algorithm is proposed in this paper. It is the digital
counterpart of the well-known continuous-time filter and
square timing recovery [5], [6], but it extr acts the t iming
information from the squared signal in a new way. The
analysis of the timing jitter presented in this paper leads to
results that are similar to the continuous-time case, although
the method of analysis is different.
Another main contribution of thc paper is a new method of
hangup-free fil t erin g of the timing signal . With all other
known timing recovery methods a major problem is that the
synchronization loop can get stuck at an unstable equilibrium
poin t . In this paper, we show how this can be a voided thr ough
planar filtering of two-dimen s ion al timing esti mates.
The final section of the paper presents a digital realization of
the timing detector which is suitable for VLSI integration also
at high data rates .
Paper approved by the Editor f(lf Signal Design, Modulation, and Detection
of the IEEE Communications Society. Manuscript received April 18. 1987;
revised September 23. 1987. This paper was presented at GLOBECOM '87,
Tokyo, Japan. November 1987.
The authors are with Lehrstuhl fur Elektrische Regelungstechnik, Aachen
University of Technology, West Germany.
IEEE Log Number 8820399.
0090-6778/88/0500-0605$01.00 1988
605
HEINRICH MEYR, FELLOW, IEEE
II. TIMING ESTIMATION
Here we consider the timing recovery for digital data
transmission by linear modulation schemes (PAM, QAM,
PSK). The rcccived signal (PAM) or the equivalent low-pa ss
signal (QAM, PSK) can be written as
00
r(l)= angr(t-nT-E(t)T)+n(t)
u(t) + n(t). (I)
=
Where an are the complex valucd transmitted symbols with
mean power I (e.g., I, i with QPSK), gr(t) is the
transmission signal pulse, Tis the symbol duration, n(t) is the
channel noise which is assumed to be white and Gaussian with
power den s ity No, and (t) is an unknown, slowly var yi ng
time dela y.
Now timing recovery means the estimation of the delay (t)
to enable the optimal dctcction of the data. Because varies
very slowly, in a digital realization, we can process the
rcccived signal section by section . For each section Llm' we
can assume E to be constant and obtain an estimate Em. This
estimate must then be combined with the previous estimates
(i.e., it must be filtered) such that the optimal estimate Em is
obtained. The latter can be used to control an analog or digital
sa mpler for the detection.
Below we consider a special type of timing estimator which
is particularly suited for digital realization. It is similar to the
continuous-timc filter and square synchronizer in that the input
signal is squared and the resulting spectral component at the
symbol rate is extracted by a filtering operation. In Fig. I the
algorithm is shown. After a receiving filter [impulse response
gR(t)l the signal f(t) = r(t) * gR(t) is sampled at rate NIT
("*" deno tes a convolution). We thus have samples
h=f(kTIN). (2)
The sequence
with
represents the samples of the filtered and squared input signal
and cont ai ns a spectral component at liT. This spectral
component, which in a conventional synchronizer is cxtracted
by a PLL or a narrow-band filter is here deter mined for every
section o f l ength LT (i.e., from LN samples) by computing
the complex Fourier co effic ient at the symbol rate
(m+l)U\'-1
Xm= xke-j2dIN. (4)
k-mLN
Asis shown in the next section, the normalized phase Em =
-1/271' a rg (Xm) of this coeftlcient is an unbiased estimate f o r
E.
IEEE
606 IEEE TRANSACTIONS ON COMMUNICATIONS. VOL. 36, NO.5. MAY 19

TIN Ilm.11 LN 1 Using the identity (A6) from Appendix A, we obtain for tl

xke
-jlnk/N expectation of X
x, k: mLN LN-l

Fig. I. Discrete-time filter and square estimator.

E[X]= E[xk]e-iM1N (1
kO

The sampling rate must be such that the spectral component

at IIT can still be represented, i.e., NIT> 21T. We use N =
4 for practical reasons. In the case of bandwidth efficient
modulation with a singlesided bandwidth of less than liT, the
with
receiving filter gR(t) also has a singlesided bandwidth of less
than liT and thus the squared signal has a singlesided
bandwidth of less than 21T. Therefore, with N = 4. the
5'lx(t)]= [00 x(t)e-jh/t dt.
sequence Xk completely describes the underlying continuous
time signal. At this point, we introduce the following functions to simpli
the notation:
III. STATISTICS OF THE ESTIMATE

In this section. we compute the statistics of the estimate m pAt) = g(t)g*(t -nT) (1
as a function of the pulse form g(t) and the noise power
p"(f) = 5' [Pn(t)]. (1
density No of the additive noise. We assume m = and omit
the index m for the sake of simpler notation. We then have

A. Mean LN
The mean of the estimate is E[Xl =""T 5'[Po(t - ET)J/ liT

E[]=E [;: arg (X) ] (5)

LN
= Po(l/T)e-i2n
T
(1

For small variance of the estimates wc can linearize the arg and thus

(
operation.

E[fl= arg
-I
(E[X]) E [] =-=--.l. arg
27r
LN
T
p0(1IT)e-J2,, )
(Ll
27r

)
1
=E-- arg Po(lIT). (I
=-=--.l. arg E[XkJe-J2dIN (6) 211'
27r kO
Therefore, under the assumption
The linearization is valid, of course, only for I arg(X)1 7r.
However, due to the subsequent filter operations, which are arg Po(1IT)=O, (I
discussed in Secti on IV, this is the only case of interest.
is an unbiased estimate of the clock phase E. But even if (I
We first have to compute the expectation of the squared
is not valid, the mean of exactly equals the required sampli
signal
offset as we show below. We assume

E[xd=E
[Inoo a"g(kTIN-nT-ET)+fi(kTIN) n gR(t)= gj.( - t +0' T)
We then have
(generalized matched filter). (J

(7)
g(t)=go(/-o:T) with gO(t)=gT(t) * gj.(-t) (J
The expectation must be taken with respect to the Jomt
distribution of the symbols a n and the noise n(t). Noise and 1(t)= ango(/-nT-ET-O'T)+fi(t). C
symbols are independent of each other. Therefore, and with
E[n(t)] = 0, the cross term of the binomial in (7) vanishes.
Since go(t) is symmetrical, the optimal sampling instant is
The remaining terms are
goU 0), i.e., for the symbol an at

E[xd=E [Inoo n
=

ang(kTIN-nT-ET) topt, n =nT +ET +0: T,

i.e., the required sampling offset is ( + O')T. Evaluating (
then yields
+E[I fi(kTIN)i2]
00 00

= E[ana!lg(kTIN-nT-ET) Since go(t) is symmetrical, thc Fourier transform in (21)

rcal.
Therefore, we have
. g*(kTIN-mT-ET)+E[lfi(kTIN)i21. (8)
arg Po (1 IT)=- 2Jl'0:
With noise power u2 and independently distributed symbols of
mean power 1, we have and thus

(9) E[E] =E+ 0: (:

which is exactly what is required for symbol detection.
OERDER AND MEYR: DIGITAL FILTER AND SQCARE TIME RECOVERY 607

B. Variance The three terms [30a)-c)] represent the parts of the timing
jitter that are generated by (signal x signal), (signal x noise),
Here we determine the variance of the random variable E,
and (noise x noise) interaction.
i.e., the mean square error of the estimation. We assume
C. Conditions for Asymptotically Jitter-Free Timing
Recovery
arg Po(lIT)=O (24) In this section, we study the conditions to be fulfilled by the
transmit and receive filters necessary for the timing estimate to
to simplify the notation. (It can easily be shown that the results
have zero variance in the noiseless case (No = 0), i.e., the
are valid for arbitrary and arg Po.) We then have conditions for the s x s-portion of the variance

var l;;J = E[ ;; 2]

1
= E [(arg (X2] (34)
(21r)2

1 E [(1m X)2] to be zero.

- (25) We have
(211-)2 (E [Re X])2 .
The latter approximation is valid since the imaginary part of X
Pm(t)= g(t)g*(t -mT) (35)
has zero mean and the variances of both imaginary and real
part are small compared to the squared real mean.
Pm (f)=G(f) * (G*( -f)e-j2'ffmT)
From (14) and (24) follows

LN
= r", G(f- v)G*( -v)CJ2wnT dv (36)
E [Re Xl =E[Xl =-Po(lIT). (26)
T and use of the abbreviation

The variance of the imaginary part is H(/)= G(l/T-f)G*( -f) (37)

yields

For real valued g(t). i.e., symmetrical joint transfer function

of the transmit and the receive filter, we have
LN-\ LN-\
= E[XkXk'] sin (hkIN) sin (hk'IN) (27) G*( -/) =G(f) (39)
k'"O k"O
H(f)=G(l/T-f)G(f). (40)
with
That means that H(f) is symmetrical around 1I2T and
(28)
1m Pm(l/T)= r", 1m H(v) eos (211"vmT) dv. (41)
By using some approximations which are valid for large L, the
expectation can be computed (Appendix B). If the results are Therefore, a sufficient condition for zero jitter is
used in (25), we obtain
1m H(f)=O (42)
(29) which can be obtained, for example, with
with
gel)= g( t) (symmetrical pulse shape)
- (43)
and also of course with all linear-phase pulses
(30a)
g(7+t)=g(7-t) (44)
1 1 2h as they act like the corresponding symmetrical pulse g(t - T)
(30b) with an additional timing delay fO = T. The conditions (43)
u;xn =(211")2 L No (Po( l I T2
and (44), however, show that the optimal receive filter in the
synchronization path is a matched filter
T
1 1 '2 Re cI>(l/T) (45)
u2 = -- _ N20 (30e)
nxn
(211")2 L (Po(1/T2 These results are valid, of course, only with the approxima
tions made in Section III-B, in particular, only for large
h= roo roo g(t)g*{t')rp(t-t') estimation intervals LT. In the case of short intervals, the
estimate exhibits jitter, but the spectrum of the jitter has a zero
at the origin and can thus be suppressed by low-pass filtering.
sin (211"tlT) sin (211"t' IT) dt dt' (31 )
True absence of jitter can be obtained in general only by using
(32) nonoverlapping pulses.
'l'(f)=5'[",2(t)l
This is in contrast to the conventional continuous-time filter

"'(7)= r", gR(t)g(t+7) dt. (33)

and square timing recovery. In the continuous-time case, the
timing is determined by detecting the zeros of the timing wave,
608 tEEE TRANSACTIONS ON COMMUNICATIO'lS. VOL :In. NO. '1, MAY 198

L=l
Therefore. true jitter-free timing recovery is possible if the x

timing wave exhibits only amplitude jitter. but no phase jitter.

The latter can be achieved. for example. with locally
symmetric pulses [7 J. In our case. however. the estimation is 1t:-02
based on samples which have an arbitrary offset from the zeros E

ami thus exhibit random amplitude fluctuations. Therefore.

only asymptotically jitter-free recovery can be obtained.
:; lE-94
E

D. Simulation Results
Fig. 2 shows the variance of the estimates f (29) for several IF-B5
estimation intervals L where both transmit and receive filter c
.

are fourth-order Butterwortb filters with corner frequency 0.71

iE-ltb 5xn
T and the modulation format is 8PSK (solid lines). The
markers show the results of Monte Carlo simulations (5000
IE B7 --------r--'---'---r--'
estimates for each point). In addition, for L = 64, the three
15 2B 25 30 35 B
parts of (29) are shown by the dotted lines. The simulations are 4
E/No
very close to the theoretical results. Only for L < 4 are there
errors due to the approximations in the computation of the Fig. 2. Computed and simulated variance of the timing estimate for sever<
intervals L.
variance which are valid for large L only. For L = I and EI
estimation Transmit and receive filter are fourth-orde
Butterworth, I, = 0.7/ T. Dotted lines: 5 x 5, 5 X n, and n X n-part fa
No = 0 dB the simulation result is smaller than the theoretical
L 64.
result. This is due to the finite range of E. Thc variance tcnds to
=

1112 when f is uniformly distributed in the estimation range.

IE-01
Fig. 3 shows the corresponding curves for linear phase x L=I I
filters with a transfer function amplitude similar to thc above
Butterworth filter. For L 2: 16 and with moderate EINo the 1E-02 L=4
L=1b I
I
E
simulation results match the theory very well. In particular,
the predicted missing of an s x s-portion of the variance can IE-B] L=b4 I x

be seen. Because the absolute variances are much smaller than C

in Fig. 2, the effects of the finite observation intervals LT are ;: 1E-04

E

much more visible here, especially for large EI No

IE 05-
E. Frequency Offset c

Since we usc a free running sampling oscillator, a frequency

offset between transmit and receive timing may be present
lHb
rcsulting in a continuously rising or falling ,. In contrast to
carrier recovery, however, the frequency offset m tlmmg i-----'-----'----"'----;----- T
IE -0 7
o Ii 15 20 25
I
3B 35
I
40
recovery is vcry small (10-5 1 0 - 2 of thc symbol rate). We
E/No
can thcrefore always find an observation length L = L' for
Fig 3,
which we can consider E to be approximately constant. Thcn
As fig, 2. but hnear-phase filters,

all considerations of the previous sections apply. For L > L'

inspection of the estimation algorithm reveals that the estimate
X is nothing but the average over estimates from shorter
intervals. Therefore, also the mean of the estimate f is just the
average of the timing delay E over the observation interval, as
long as the variation of E is smailer than T12. The latter
condition, of course, limits the possihle observation length L.
Similarly, for small frequency offsets, the variance of the
estimates can be expected to be nearly independent of the
frequency offset. For larger frequency offsets, one would have
to examine whether the algorithm behaves like its continuous
time counterpart that exhibits a significant increase in timing
jitter in the presence of frequency offset.
IV. PLANAR FILTERING OF THE ESTIMATES
Due to frequency offset and random variations of the delay
E , the observation length L is limited. The estimation,
however. can be significantly improved if the knowledge of
the statistical properties of E is used to postfilter the estimates.
For example, a simple "random walk" model for the time
delay E leads to a first-order Kalman filter. The variance of the
filtered estimates can be computed from the variance of the
unfiltered estimates and the random walk parameters [8J.
Since the range of the estimates fm is finite, the filter
innovation must be reduced to the range [-0.5,0.5] as shown
in Fig. 4.
With this kind of filtering, however, the following situation
can occur. If the true value E is at a value 0.5 distant from the
filtered value E, the estimates E also vary around this value and
thus the innovation is at 0.5 and vanishes in the mean. Then
Fig. 4. Filtering of the estimates.
the filter is in an unstable equilihrium amI it can remain ther
some time in spite of the large error.
This is a problcm which arises also for other models an
filters and with almost all known synchronization methods
even if they-like the one presented above-could be calle,
. 'open loop" at first sight. Because of the periodic behavior c
the filtered delay "hangups" can occur in the filter loop.
Below we present a realization which avoids these prot
lems. The central idea is to filter a complex phasor instead c
the corresponding (periodic) angle. The term Xm from Fig.
is such a phasor. Instead of first determining the angle of thi
phasor and filtering the angle, we can apply a Kalman filter t
the phasor itself and use the angle of the filtered phasor t
control the sampling (Fig. 5).
In Figs. 6 and 7, a situation for filtering of E and filtering c
X is shown. In the first case a hangup can occur if the error i
E is approximately 0.5 (corresponding to an angle of 7r). Wit
planar filtering, however. the filtered value X movcs eorreetl
(with a step width which depends on the filter coefficient
present) towards its place. Thus, hangup problems cann<
occur any more.
OER])ER AN]) MEYR DIGITAL FILTER AND SQUARE TIME RECOVERY 609

FIg. 5. Planar filtering at the e'timale,.

[IJ: Lat,h

Ts

_
T" J"lc--__-JnL..____
Fig. 6. Example for the trajectory of estimate and filtered value with
T. ----.ll nL.. __
filtering of the delay values
T,
T, ___ nL____

Fig. 8. Fast digital realization of the detector.

VI. CONCLUSIONS
Re The proposed timing recovery enahles a VLSI realization of
digital receivers which can operate on a sampled input signal
without any feedback to the sampling device. The latter can
operate at a fixed rate with a free running oscillator. The
planar filtering algorithm results in very fast and hangup-free
timing recovery.

Fig. 7. fig. 6, but planar filtering.
The filtered estimate E-m of Fig. 5 has of course the finite
range L 0.5, 0.5J again and with only small variation if' X",
ApPENDIX A
Equivalence of discrete-time and continuous-time computa
tion of the Fourier coefficients of periodic band-limited
functions. Assuming x(t) to be a Tperiodic and NI2T-band
limited signal, we show that

_ IT
jumps can occur bctwccn 0.5 in Em. The interpolation unit.
kO LN 1
however, which is controlled by (m can easily discriminate LN
x(kTIN)e )hkIN= x(t)e-J2ffIIT dt,
these "wrap-around" jumps from true variations of the time
T 0
delay and therefore determine the underlying infinite range
estimate and correctly compute the decision metric [4]. (AI)
As a final remark. let us note that with the digital filter and
To do so, we stali with the integral form. Due to the low-pass
square timing estimation, the planar filtering is nothing but a
limitation, we can rewrite the signal by using sine-interpola
(weighted) summation of succcssivc values X",; and this is
tion (sine x = sin x/x)
merely an extension of the estimation interval LT in the
algorithm for computing Xm (Fig. 1) with an additional
weighting of the terms. \ T x(t)e-
"0
j2,(/T dt
V. REALlZATlON OF THE DETECTOR

(71" )
\,T i:
Fig. 8 show, a possible realization of the computation ofXm t _nT
_ _ _IN
= x(nTIN) sine e-)2.1IT dt
'0
which allows high data rates through the use of parallel
,, - TIN
processing and pipelining. oo

With a double set of latches. the quadruples of samples

belonging to an estimation interval of lenglh L = I are
(A2)
collected. Since the sin and cos functions take on only values 0 kO I /'1,'-1
and I, no multiplications arc necessary. The samples can
then be processed at the symbol rate II T rather than at the 41T
x(kTIN)
k=kO
sampling rate. Squaring and addition can be divided by latches
into further pipeline stages. Thus, the fact that the estimation
algorithm needs 4 samples per symbol (instead of one or two
(A3)
as other algorithms do) is relevant virtually only 10 the AID
converter and therefore the estimator can be used even at high
data rates. We are currently incorporating the detector into a
CMOS standard cell chip which will run at about \0 Mbits/s. (A4)
In cases where a low number of samples per symbol is
important (e . g. . when adaptive ceho cancellers are used), the
T kO+I."-1
actual sampling rate can be reduced to 2 samples per symbol
by using a simple all-pass filter to generate the missing x(kTIN)e-j2dIN. (AS)
=-
LN *-'0
samples [2].
610 IEEE TRANSACTIONS ON COMMUNICATIONS. VOL. 36. NO.5. MAY II

In particul ar , with x(t) = 2::_ -00 y(t - nT), we havc

k +L.'V-I
O LN
x(kTIN) e-j2.kIN=- Y(IIT) , (A6) g*(kTIN-n T) E[n(k'TIN) n*(k' TIN) ]
T 2
'"

=No Po(kTIN-iT) <p(O) (BII

ApPENDIX B i= -00

We first compute the expectation

E3= E[anja:2] g(kTIN-njT)
ElxkXk' ] n, n2

=E [( anjg(kTIN-nT) +ii(kTIN ) g*(k' TIN-n T) E[n(k'TIN) n*(kTIN) ]

00
2

g(kTIN-iT) g*(k'TIN-iT)

(
=No

"2
an2g(kTIN-nT) +ii(kTIN ) *

i=-00

cpk-k') TIN) (81

(
"3
an3g(k'TIN-nT) +ii(k'TIN ) E 4 =E[n(kTIN) n*(kTIN) n(k'TIN) n*(k'TIN) ]

=N&<p2(0) +N&<p2k-k') TIN) . (B1

( an4 g(k'TIN-nT) +ii(k'TIN )] * The corresponding terms in (27) are now computed. TI
approximations are valid for large estimation intervals LT.
LN-I LN-I 00 00

(BI) Sll= Po(kTIN-iT)

k=O k'=O i_-oo j=-OO
With
PoCk'TIN-jT) sin (21rkIN) sin (27rk'IN)
E [a;ii(t) ]=0 (B2)
E[a;] =0

E[a;an = O'j
(B3)
(B4)
= [ TIm LN
Po(lIT) ] 2
=0 (B1:

LN-\ LN-l 00 00
(B5) S12=

(i
E[a,aj]=0 pJ(kTIN-iT)
k=O k'=O i=-oo j=-oo

for i =j*k / =

for i=/*j=k
p iCk'TIN-iT) sin (27rkIN) sin (27rk'IN)
E[a,oj",aI1 = (B6)
for i=j=k=/
00 00 00

""L pj(kTIN) pj(k'TIN)

otherwise
k=-oo k'=--OOj=-OO

sin (21fkIN) sin (27r k' IN)

00

(B7) =L(NIT) 2 (1m PiC1IT2 (81

J= -00
10 of thc 16 terms which result from (B 1) vanish. The L'V-l UV-l
following terms rema in: SI3=(y-2) Po(kTIN) Po(k'TIN)
=0
(B8) k=O k'

with sin (27r kIN) sin (21fk'IN)

00 '" 00

E, = E[anja: an3a:4] ""(y-2)L Po(kTIN-iT)

2
TI l n2 '13 n4

. g(kTIN-n1 T) g*(kT/N-n2 T)
Po(k'TIN-iT) sin (27rkIN) sin (21rk'IN)

. g(k'TIN-n3T) g*(k'TIN-n4T) (89) =L(NITj2(y-2) (1m Po(1IT)) 2=0. (B1

Therefore,
00

= Po(kTiN-iT) Po(k'TIN-jT) '"

1= -00 J=-oo
SI ""L(NIT) 2 (1m Pj(lIT2 (B1
j= -00

00 00
+ pj(kTIN-iT) p/k'TIN-iT)
and
i= - 00)= 00 LN-1 LN-J 00

S2 =No Po(kTIN-iT) <p(O)

00
k=O k' ",,0 i_-oo
+(-y-2) Po(kTIN-iT) Po(k'TIN-iT)
/= - 00 . sin (27rkIN) sin (21fk'IN) =0 (Bl
OERDER AD MEYR: DIGITAL FILTER AND SQUARE TlME RECOVERY 611

LN-! LN-I with

S3=No g(kTIN-iT)g*(k'TIN -iT)
k=O k'-O i--=--o:>

(; Y
l{!k-k')TIN) sin (27rkIN) sin (27rk' IN)
L
00 Po(1IT)
g(kTIN)g*(k'TIN)
k-=--c;ok'=-oo (1m Pm{lIT2
1 m
I{!k - k')TIN) sin (27rkIN) sin (27rk' IN) (B2 2)

""'L{NIT)2No roc roo g(t)g*(t')I{!(t-t')

. sin (27rtIT) sin (27rt' IT) dt dt'

: = L(NIT)2NoIJ (BIg)
LN-l LN-l I I 213
S4= (1{!2(O)+1{!2k-k')TIN = No (B23)
L
( )
(2 7r)2 Po(1IT2
kO k' O

N ' 1
sin (27rkIN) sin (27rk'IN) - LTN 6 - Re \fr(1/T)

( )
1 T 2
LN-l LN-I
(27r)2 LN 2
= N6 1{!2k-k')TIN) - Po(IIT)
kO k' O T
T
sin (27rkIN) sin (27rk'IN) - Re 'Jr{lIT)
1 1
2LI\o'-2 = ___
N2 _2 ___
(B24)
(27r)2 L 0 (Po(1/T2
""'N 1{!2{nTIN)
m=O n= 00
(m + n) even ACKOWLEDGMF.NT

We would like to thank the reviewers for their valuable

comments and suggestions,
REFERENCES

[I] K. H. Mueller and M, Muller, "Timing recovery in digital synchro

00
nous data receivers," IEEE Trans. Cornman., vol. COM-14, pp,
rp2(nTIN) 516-530, May 1976.
m=O n=-oo [2] O. Agazzi, C. -P. J. T7eng, D. G. Messerschmitt. and D. A. Hodges,
(m + n) even "Timing recovery in digital subscriber loops ," IEEt.' Trans. Com
mun., vol. COM-33, pp. 558-569, June 1985.
13] F. M. Gardner, "A BPSKiQPSK tlilling-error detector for sampled
I

e
receivers, " IEEE Trans. Commun., vol. COM-34, pp. 423-429.
- (cos (27rmIN)-cos (27rnIN
2 May 1986.
[4] M. Oerder. G, Asch id, R. Hiib, and H. Meyr. "An all digital
00 implementation of a recei ver for bandwidth efficient communication,"
=N6 (LNI 2)rp2(nTIN) cos (27rnIN) in Signal Processing 1I1: Theories and Applications. I. T. Young el
al.. Eds. New York: Elsevier, 1986, pp. 1091 -1094.
[5J L E. Franks and J. P. Bubrouski, "Statistical properties of timing jitter
in a PAM timing recovery scheme." Ifo:EE Trans. Cornman. . vol.
COM-22, pp. 913-920, July 1974.
r6] M. Moeneclaey, " Pr e iliter optimization for the filter and 'quare
synchronizer," Arch. Elektr. Ubertragung, pp, 257-261, 1984
[7] A. N, D'Andrea, U. Mcnga li , and M. Moro, "Nearly optimum
LN2 preljltering in clock recovery," IEEE Trans, Cornman. . vol. COM-
=- N6 Re \fr(1IT) 34, pp. 1081-1088. Nov, 1986.
2T [X] R. D. O. Anderson and J. B, Moore, Optimal Filterin". Englewood
Cliffs. NJ: Prentice-Hall, 1979.
(BI9)
*
Finally, we can write
Marlin Oerder (S '88) was born in Aachen, West
Germany, on August 26, 1957. He received his
(B20)
Dipl.-[ng. degree in electrical engineering from

,
Aachen University of Technology in 1984, Cur
The three terms represent the parts that are generated by rently, he is pursuing the Ph,D. degree at Aachen
(signal x signal), (signal x noi se) and (noise x noise). University of Technology.
Correspondingly, the variance of is His interests are in the field of digital communi
cations with emphasis on digital synchronizatIOn
techniques.
(B2 1)
612
Heinrich Meyr (M'75-SM'83-F'86) received the
Dipl.-Ing. and Ph.D. degrees from the Swiss
Federal Institute of Technology (ETH) , Zurich, in
1967 and 1973, respectively.
From 1968 to 1970 he held research positions at
Brown Boveri C orporation, Zurich, and the Swiss
Federal Institute for Re actor Research, From 1970
to the summe r of 1977 he was with Hasler Research
Laboratory, Bern, Switzerland, His last position at
Has ler was M anage r of the Research Department.
During 1974 he was a Vis i ting Ass is tant Professor
with the Department of Electrical Engineering, University of Southern
California, Los Angeles. Sinc e the summer of 1977 he has been Professor of
IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 36, NO.5, MAY 198:
Electrical Engineering at the Aachen Technical University (RWTH), Aachen
West Germany, His research focuses on synchroniz ation , digital signa
proces sing , aDd in particular, on algo rithms and a rchi tectures suitable fc
VLSI impl ementation, In this area he is frequentl y in
demand as a consultar
to industri al concerns, He has published work in various fields and journal
and holds over a dozen pa tents .
Dr. Meyr served as a Vice Chairman for the 1978 IEEE Zurich Semim
and as an International Chairman for the 1980 National Telecommunication
Conference, Houston, TX, He served as As soc iate Editor for the lEE.
TRANSACTIONS ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING fror
1982 to 1985, and as Vice President for International Affairs of the lEE
Communications Society,