IEEE TRANSACTIONS ON COMMUNIC.VrION T E C H X O L O G Y VOL. COM-1.5, NO.

4 AUGUST 1967

E. T. Harkless was born in Cleve- H. F. Lenzing (S’56-M’62) was

land, Ohio, on June 25, 1922. He born in Jersey City, N. J., on June
received the B.S.E.E. degree 1947 5, 1934. He attendedRCA Institute,
and N1.S in physics in 1949 from N. Y., graduating in 1958. In 1960,
Case School of Applied Science, he received the B.S. degree in physics
Cleveland, Ohio. from FairleighIlickinson Univer-
He served inthe U. S. Army sity, Teaneck, N. J., and in 1962,
Signal Corps during World War 11. received the 1vZ.S. degree in physics
He joined the Apparatus Develop- from Stevens Institute of Tech-
ment Department of Bell Tele- nology, Hoboken, N. J.
phone Laboratories, Murray Hill, He joined Bell Telephone Lab-
N. J., in 1949, and worked on oratories, Whippany, N. J., in
development; of wavegnide and 1959. Since then he has worked on
coaxial apparatus.He has sub- various microwave projects, par-
sequently transferred to the Trans- ticldarly in component develop-
mission SystemsJlevelopment Di- ment. From 1964 to 1965hewas apart-timeinstructor in the
vision where he has worked on several different. transmission systems Department of Electrical Engineering of the Newark College of
suchas microwave radio relay systems, Telstar, and waveguide Engineering, Newark, N. J. At present, heis a member ofthe
transmission of millimeter waves. Transmission Ilivision a t Bell Telephone Laboratories, Holrndel,
Mr. Harkless is a member of Tau BetaPi and Sigma Xi. N. J.! where he is working on X-band radio relay.

Design of Space Diversity Receiving andTransmitting

Systems for Line-of-Sight Microwave Links
HIDE0 MAKINO AND ICAZUO MORITA

REFERENCE : Makino, H., and Morita, K. : DESIGN OF SPACE unfavorable propagation paths, it is necessary to reduce the
DIVERSITY RECEIVING AND TRANSMITTING SYSTEMS noise bursts or interruptions caused by fading so that the
FOR LINE-OF-SIGHT MICROWAVE LINKS, Nippon Telegraph
and Telephone Public Corporation,Tokyo,Japan. Rec’d 6/13/66; overall circuit performance may meet the CCIR recom-
revised 3/27/67. Paper19TP67-1076, presented at the 1967IEEE mendations.
InternationalConvention, New York, N Y. IEEE TRANS. ON Fortunately, itis rather easy inJapan toselect propaga-
COMMUNICATION TECHNOLOGY, 15-4, August 1967, pp. 603- tion paths through the mountainous terrain. Therefore,
614.
fading causedbyout-of-phaseconditionsbetween the
direct and the earth (or sea) reflected wave is rather un-
ABSTRACT: Severe fading is a substantial factor that impairs the
important, except for a few examples. However,multi-
quality or reliability of transmission over line-of-sight microwave
relay systems, producing noise bursts and signal interruptions. path fading due toradio ducting is quite important.
This paper describes space diversity receiving andtransmitting Channelswitchingsystems,i.e.,frequencydiversity
systems developed by the Electrical Communication Laboratory of reception, are generally used for microwave relay links.
Nippon Telegraph and Telephone that reduce the effects of such I n fact, they are an effective protection against equipment
fading. Theoretical research studies and resultant design criteria,
failure. Theycannot alwayseliminatefading, however,
an explanation of fundamental circuits, several example of diversity
effects, and the application of the criteriato the design of line-of- when the latteris due to radio ducting, because the correla-
sight microwave links using space diversity are described. tion of signal strengths between adjacent radio channels
is inadequate. On the contrary,further improvements
KEYWORDS: Design, Diversity Reception, Fading,Microwave are necessary, by making use of space diversity principles
Relay Systems, Multipath Pi0 agation, Propagation, Reliability, involving receiving ortransmittingwithdualantennas
Signal-to-Noise Ratio, Transmission.
vertically separated by a proper spacing.
Although various alternatives can be considered in the
I. INTRODUCTION space diversity technique, the method combining two re-

T HE PERFORRIBNCE of line-of-sight microwave ceived signals controlled in phase at the radio frequency
relay systems may sometimes fail t o meet the trans- level has been adopted as the best, considering both per-
mission objectives recommended byCCIR because of formance and economy.
unexpectedsevere multipath fadingdue to radiowave The diversity gain should be defined in order to measure
ducting or because of unsatisfactory path geometrywhich the diversity effects. Since the diversitygainis closely
it was necessary to select for economic reasons. On such related to thespace or frequency correlation coefficient, the
 604 -4UGUSTTECHNOLOGY IEEE TRANSACTIONS
COMMUNIC.4TION ON 1967

relationship between them should be explored. Frequency A.lternately, in an individual phase control system,since
correlation has a particularly important meaning for the phase control can be individually accomplished with each
performance of the common phasec.ontlro1system described radio channel, good performance can be easily obtained.
later. Proper antenna spacingshould b,e studied, consider- A rapid response can also be expected because a smallsize
ing both diversity gain and economy. The automartic phase phase shifter can be used, although the system is somewhat
controlsystem, beingaform of servo-mechanism,has expensive.
several problemsto be solvedif a sttable and rapid response Furthermore,the spacediversity principle may be
is to be obtained. applied either at the receiving station or the transmitting
The purpose of this paper is to describe the theoretical station of a link.
research results on two typesof diversity systems used by In the latter, as will bedescribed in Section V,two
N T T : common phase controland i ndividua.1phase control. transmitting antennas are used and the phase angle be-
Design criteria are suppliedfor the spa'ce diversity systems tween the two transmittedsignals is controlled so as to be
in order to get goodperformanceandeconomy.Basic always in-phase at the receiving station with a single re-
equipment needed for space diversity systems, and some ceiving antenna.
improved parts developed in recent years are described. If space diversityreceiving systems only are used and are
Several examples of diversity effects obtained 011 various applied in opposite directions to each other on a circuit,it
propagation paths are analyzed, and finally, design criteria is necessary to install space diversity equipment a t each
forspacediversitysystemsare :~pplied to line-of-sight two-way repeating station. Alternately, if space diversity
microwave relay systems. transmittingsystemsare applied togetherwithspace
diversity receiving systemsonacircuit, the diversity
11. VARIOUS ALTERNATIVES IN SPACE .DIVERSITY SYSTEMS equipment can be installedat alternate repeating stations.
Two methods for producing a received signal from the This affords considerable convenience of constructing the
two receiving antennas are: overall space diversity system.
1) selecting and using only, the larger of the two re- 111. A.NALYSIS OF SPACE EFFECTS
DIVERSITY
ceived signals Definition of Diversity Gain
2) combining the two received signals which are con-
trolled in phase. It is necessary to establish the definition of diversity
gaininordertomake clear thediversity effect. Two
It may be unsatisfactory, howl?ver, t o apply the first methods can be employed to define the diversity gain.
method to a high-quality microwwe link, because of the I n Fig. 1, curve A shows thecumulativeprobability
interruptions due to selector switching transfer time. distribution of the relativesignal level (i.e., when the level
The second method has three possibilities: (a) combin- received in free space propagation is taken as zero dB)
ing the received signals in the radio frequency band; (b) received by asingle antenna undersevere fading conditions.
combining them in the intermediate frequency band; (c) Thisdistributioncan be approximated,ingeneral, by
combining them in the base band. Rayleigh's formula.[']
It may beimpossible to apply (c) to a heterodyne repeat- Curve B in Fig.1 shows the cumulative probabilitydis-
ingsystem because thereareneithermodulatorsnor tribution of diversity received signals.
demodulators at the intermediate repeating stations.
Method (b) may be very effective when the changes in 4 /o
phase-angle between two received signals are very rapid, eM
d
such as with an over-the-horizon link, but this is not the v)

case with aline-of-sight microwave link.I t also may notbe ,u

economicalbecause two receivem arerequiredforeach g o
d

radio channel.
W Q
L
Method (a), which combines the R.F received signals in YI 3 -10
phaseinthewave-guidecircuit,isthebest becausea 4

mechanical endless rotary-phase-shifter without thermal g

noise can beused,and simple structure andsuperior 3 -a
economy can be obtained. s
d I I I I I 1 1 1 1 1 1 I I I \ I
Y
Two systems available in (a) arc: space diversity system4
with conmlon phase control, and space diversity system f -a 1 1 I I I I I I I I I I I \1
B?pP e 9 9/ 95 PO 80 76&~v49sv.?d /O S
. / 6.1 0.6/
with individual phase control. (%>
In the common phase control s,ystem, phase control for Cumulative Probability
several radio channels included in the link can be com-
monly accomplished by a single phase control circuit and X , : Diversity Gain defined by Definition 1)
the system can be economically composed, though require-
DG, : Diversity Gain defined by Definition 2)
ments for good performance are ratlher strict, as will be Fig. 1. Typical example of cumulativeprobability distribution
curves of received signal by single reception (A) and diversity
described in Sections 111 and IV. reception (B).
 MAKINO AND
MORITA: SYSTEMS FOR
SPACE DIVERSITYMICROWAVE LINKS 605

Diversity gain can then be defined as follows: Along with (l), empirical formula ( 2 ) was obtained by
analyzing a number of data on the correlation of the re-
1) the difference in dB (or received signals power ratio)
ceived power.
between single and diversity received signals corresponci-
That is
ing to the cumulative probability of one percent
2 ) the ratio of cumulative probabilities between single kS2 = exp[-O0.00021{Ah.f.d0.4d + - (k2Z)2}}]
and diversity received signals at therelative receivedsignal
level of - 20 dB. (2)
The first definition is preferable for measuring the link where
quality for the signal-to-noise ratio, and thesecond for the ks2 meanvalue of space correlation coefficient under
probability of interruptions. severe fading conditionsin theworst season
Assuming that curve A. is a Rayleigh distribution and Ah vertical antenna spacing, meters
the variationbetween main and sub-received signal is per- f frequency,GHz
fectly random(Le., the space correlation coefficient is zero), d length of propagation path, km
diversity gain at a probability of one percent becomes11.5
dB.
k = d r 2 / ( 1 r2) +
r amplituderatiobetweendirectandearth reflected
Curve B in Fig. 1 shows that a diversity gain of about waves, assuming it to be considerably smaller than
9 dB is obtainedat a cumulative probabilityof one percent. unity
On the other hand, the cumulative probabilities of single 1 path difference betweendirect andearth reflected
and diversity received signals at a relative received signal waves, cm.
level of -20 dB arealso shown in Fig.1, being one percent A comparison of the values predicted by ( 2 ) and the
and 0.01 percent, respectively, and the ratio, i.e., diversitymeasured values is shown in Fig.2 . There is a fairlygood
gain indefinition 2 , is 100. This means that theprobability agreement between them.
of interruptions in diversity reception becomes 1/100 of In the case of the propagation path with the earth-
that in single reception when the interruption level is set reflected wave, kS2 is reduced as the earth reflection in-
a t a relativereceived signal level of - 20 dB. creases. The relationship between the length of propaga-
CorrelationCoeficient tion path d and vertical antenna spacing Ah is shown in
Fig. 3(a) and (b) in thecase off = 4 and 6 GHz and = 0,
Thecorrelation coefficient betweenmain andsub- taking ks2as a parameter.
received signals is a very important factor which affects
the diversity gain. There are space and frequencycorrela-
tion coefficients to be considered.
Frequency Correlation Coeficient
The frequency correlation coefficient k,2 of the received
Space Correlation Coeficient power under severe fading conditions is theoretically given
Propagation tests on various paths have revealed that by thefollowing
space correlation is large in the horizontal plane and is
small in the vertical plane undersevere fading conditions.
On a propagation path with an earth (or sea) reflected
wave, whose reflection coefficient is nearly unity, the field
intensity a t a receiving station reveals a simple height-
Therefore, the space correlation can be deter-
mined by the relation between height-pattern pitch and 1 .o
antenna spacing.
On a path where the reflected wave is almostobstructed
by a shielding ridge, the nlultipath propagation due to
radio ducting becomes dominant and the path differences
-
-J0
between the components of amultiplewavechange at >
-0
random. A. heightpattern,therefore, is not generally P 0.5
Y)

evident and the space correlation can beconsidered to 0

5
-
exponentially decrease as the antenna spacing increases. NJ
-x
The vertical correlation coefficient kS2is related to fre-
quency f, antenna spacing Ah, and standard deviation AP
of path-length difference which varies due to meteoro-
0
logical conditions on the path, and mag be given in the
exponential form as 0 0.5 1 .o
k,,2'lTheoreticol Voluel
kS2 = exp[-K{Ah.f.APf ] (1)
Fig. 2. Comparison between theoretical and measured values 01
where K is a constant. the space correlation coefficient.
606 TRANSACTIONS lEEE ON COMMUNICATION
TECHNOLOGY AUGUST 1967

where under severe fading conditions, is theoretically given by

the following fornIula:[51
AP deviation of length of muhiplf: propagation paths
around mean value AP, = d p 2+ { ( k l ) 2 - (k21)2] (4)
X wave
length.
where AP is a standard deviation of path-length difference
The frequencycorrelation coefhcient of the received due to radiowave ducting.
power against a frequency difference ,Af is shown in Fig. A.s the standard deviation of the path-length difference
4,taking AP as a parameter. due to radio ducting is usually less than severalcenti-
Meanwhile, the standard deviation ,Apeof the effective meters on a path lengthof 50 to 60 km, frequency correla-
path-length of the resultant wave which is given by the tion coefficient K12 of the received power between the
vector sum of the direct and the earth reflected waves adjacent radio channels can be evaluated by (3) and be-
comes nearly unity as shown in Fig. 4. The existence of
an appreciable earth reflected wavewould reduce the value
of k f 2 because the effective path-length difference be-
tween the direct and earthreflect,ed waves is usually large.
Theoretical Analysesof the Relationships between Correlation
Coeficients and DiversityGain
I n regards to diversity gain, only the space correlation
coefficient may be considered in the individual phase con-
trol system. Both space and frequency correlation coeffi-
cients should be considered in the common phase control
system.
Generally, a joint probability of fading signals can be
obtained by taking the correlation coefficient of received
power between signals as a parameter. Diversity effects of
the controlled channels are given by taking kS2and k12 as
d (km) d (kml parameters.I61 Without describing theminute processes
only the theoreticallyobtainedprobabilitydistribution
Fig. 3. Relationships between path length and vertical antenna
spacmg. curves will be shown here:

Af (MHZ)
Frequency Difference

Fig. 4. R~dationsllipsbetween frequency difference and correlation coefficient.

Y
PACEMORITA: AND MAKINO FOR MICROWAVE LINKS 607

In the case of k,2 = 1, 1cs2 # 0: The cumulative probabil- ability density region is shown in Fig. 7(a) for the case of
ity p ( R 2 ) in a small probability density region is shown in ks2 = 0.6. Therelationshipsbetween k,” anddiversity
Fig 5(a). gain are also shown in Fig. 7(b)
taking the relativereceived
The relationships between kf2 and diversity gain given signal levels as parameters.
by the second definition are also shown in Fig.5(b), taking For the common phase control system, Fig. 6(b) and Fig.
the relative received signal levels as parameters. 7(b) show that the diversity gain a t a controlled channel
I n the case kV2= 0, k,” # 0: The p(R2)in a small prob- rapidly decreases as k f 2 decreases from unity, so that the
ability density region is shown in Fig. 6(a). The relation- frequency correlation coefficient shouldbenearlyunity
shipsbetween k f 2 and diversity gain are also shown in to get goodperformance. On theotherhand,forthe
Fig.6(b)takingtherelative received signal levels as individual phase control system,the frequency correlation
parameters. may not be substantial, so that only the space correlation
In the case of k,2 # 0, lc,” # 0: The p ( R 2 )in a small prob- decides the diversity gain.

( X 1
Cumulative Probability

0 0.2 0.4 06 0.8

‘0 0.2 0.4 0.6 0.8 /
/

K’, K;
Space Correlation Coefficient Frequency Correlation Coefficient

(b) (b)
Fig. 5 . (a) Theoreticalcumulativeprobabilitydistribution of Fig. 6. (a) Theoretical cumulative
probability
distribution of
diversity received signal. (b) Relationships between space diversity received signal. (b) Relationships between frequency
correlation coefficient and diversity gain wheu k,z = 1. correlation coefficient and diversity gain of controlled channel
when kS2 = 0.
 606 IEEE TRANSACTIONS O N COMMUNICATION TECHNOLOGY AUGUST 1967

The relationshipsbetween antenna spacing and propaga-

tion path length calculated from ( 2 ) are shown in Figs.
r(
3(a) and 3(b) at the frequency of 4 and 6 GHz, respec-
J tively, taking ks2 as a parameter. With regard to feeder
systems, more attention should be paid in the con~mon
phase control system than in the individual phase control
system. This is because the phase angles of every radio
signal included in the link are commonly controlled by
only a common phase controller in the common pha,se con-
trol system.In order to meet this requirement the following
conditionsarerequired for the commonphase control
system :
1) On the standard condition, i.e., free space propaga-
tioncondition,themainandsub-antennasshouldbe
set so that the phase differences between the main and
sub-received signals in the two radio channels are nearly
equal at the apertureplanes of the antennas.
2 ) Feeder lengths from the main and sub-antennas to
the combining point should beprecisely equal.
During a fadingsituation,condition 1 would be dis-
turbed and the diversity effects wouldbemore or less
decreased according to the amount of frequency correla-
tion.
To meet the second of these conditions, the lengt,hs of
the main and subfeeder should be electrically as equal as
possible. A U-shaped adjustable waveguideisinserted
into the shorter feeder to make the electricalfeeder lengths
equal.
For the individual phase control system, the restrictions
described previously may not be needed when considering
the diversity effect. However, considering that the space
diversity system may be applied t o heavy loaded micro-
wave links, the equivalentelectrical lengths of both feeders
should be coincident so that the FM distortion of multi-
plexed telephone signals can be minimized.

V. SPACEDIVERSITY
EQUIPMENT
0
. 0.2 O& P A 0.8 / Phase Control System
The phase control system is a servo-nlechanism which
Kf
consists of a phase detector to detect the phase difference
Frequency Correlation Coefficient between the tworeceived signals, a mechanically rotatable
(b) microwave phase shifter to shift the phase angleof one of
the two received signalsso that thetwo received signalsare
Fig. 7. (a) Theoretical cumulative probability distribution of diver- automatically kept in phase, and a combiner t o combine
sity received signal. (b) Relationships betwteen frequency correla-
tion coefficient, and diversity gain of controlled channel when the two signals which are controlled in phase.
k.2 = 0.6. A phase discriminating method is used in the common
phasecontrolsystem,andaphasesensingmethodis
1:v. DESIGNCRITERIA OF THE SPACE I h V E R S I T Y SYSTEM usedin the individualphasecontrolsysteminorder
to detect the phase differences.
A large antenna spacing necessaryis to get agood diver-
sity effect. However, as shown in Fig. 5(b), diversity gain Common Phase Control System
changes considerably inthe range where the space correla- Figure 8 shows a schematic block diagram of the common
tion coefficient ICs2 is between1.0 and.0.6 but, changes rather phase control system, in which the solid line shows the
little in the range where k s 2 is between 0.6 and zero. More- space diversityreceiving system and the dotted line shows
over, toolarge an antenna spacing would not beeconomical the space diversity transmitting system.
for anantennatower. It would be reasonable, there- In the space diversity receiving system with common
fore, to take an antenna spacing such that ICs2 does not phase control, oneof the radio signals received by the two
txceed 0.6. receiving antennas ( 8 and~ Ar2)
~ is taken out througha di-
MAKIXO A N D 310RITA:
DIVERSITY
SPACE SYSTEMS FOR MICROWAVE LINKS 009

” antennas), the received signals at the counterpart (dis-

tant) receiving station should be in phase. The require-
ments for propagation and feeder length are just thesame
as in thespace diversityreceiving system.

Individual Phase Control System

Figure 9 shows a schematic block diagram of a space
diversity receiving system with individual phase control..
A wide-band microwave phasemodulator(PM) using-
ferrite material is inserted into main (or sub) feeders a n d
sinusoidually modulates the phase angles of all received
signals by about f15 degrees. The modulating frequency
u u u f, is about 35 Hz and isfed fromthe oscillator panel(OSC)
Fig. S. Schemat,ic block diagram of a space diversity system with installed at thecontrol bay (AH).
common phase control. The microwave signals received by the two antennas
(Arl and Arz) are separated by the channelbranching
rectional coupler (DCL) anda branching filter(A.WD), and filters (BF) on both the A and B sides. Each branched
fed to a control bay(AH). In thiscontrol bay, bothmicro- signal of the B side iscombined with thatof the respective
wave signals are converted to the intermediate frequency A side by a Magic Tee (MT) through an automatically
via a common local oscillator. The intermediate frequency rotatable phase shifter(A.PS).
signals are separately amplified and the phase difference An amplitude modulated (AM) component is generated
between them is detected by a phase discriminator. The on each combined signal as shown in Fig. 10 and the fre-
output of this phase discriminator becomes zero when the quency of the A.M component becomes 2fp when the phase
phase difference is 90 degrees, and a positive or negative angle between both signals is zero. An fp component is
dcvoltagecanbeobtaineddependinguponaleading generated when the phase angle is not zero. The phase of
or lagging phase angle from the 90 degrees. the f, component is polarized depending uponthe direction
The dc output voltage is convertedto ac (usually com- of the relative phase shift.The AMcomponent is detected
nlerical frequencyis used) by a current chopper, and fed to by an AGC circuit of the main IF amplifier of the receiver
a %phase motor which drives the microwave rotary phase (R), then amplified and phase detected. It then drives a
shifter (RAPS) inserted into one side of the feeders. two-phase motoradapted to the APS. The two-phase
With this structure, the phase difference between the motor does not rotate from the 2fp component. It does so
microwavesignals is controlledso as always to be 90 only from the f p component. The direction or rotation
degrees. A 90-degree phaseshiftershouldthereforebe depends uponthe phase of the f p component.
inserted into one side of the arm which feeds the local Thus, both signals are always automatically combined
oscillator power. Thus,thephase angle between the in phase. As these actions are takingplace individually in
two signals can be controlledso as always to be in phase. each radio channel,the several restrictions on propagation
In this system, only one of several radio channels in- and settingof antennas andfeeder lengths maybe reduced
cluded in the link is directly controlled. Therefore, it is compared withthe common phase control system.
called a controllingchannel and aradiochannel at the The effects of the transmitted information (Le., multi-
centralfrequency of the link is usually chosen for this plexed telephone and television signals, etc.) due to the
purpose. The phase angles of the otherchannels are application of phaeemodulation tothe received radio
controlled just the same as with the controlling channel. signals can be entirely neglected because the amount of
For this reason, this method is called the space diversity frequency modulation corrcsponding to the applied phase
system with conamon phase control. modulation is very small.
In the space diversity transmittingsystem withcommon Figure 11 shows a schematic block diagram of a space
phase control, the transmitting power of each transmitter diversity transmitting systemwith
individual
phase
is divided into two parts by a Magic Tee (TMT). One control. At the right side station, the transmittingpower
part is fed to the transmitting antenna A t l through a U- of each radio channel is divided into two parts by an MT.
shaped phase equalizer (PE), and the other is fed to the One part is fed to antenna (Atl)through a branching filter
transmitting antenna A t , through a microwave wide-band (WD). The other part is fed to the other transmitting
rotary phaseshifter (TAPS). TAPSis connected with antenna (At2)through APS, a branching filter, and a com-
RAPS by a self-synchronous motor (SM)and always ro- mon PM which is the same as is used ina receiving
tates simultaneously with RAPS. The transmitted signals system. When the transmitted signals radiated from both
are radiated from the two transmitting antennas with a transmitting antennas are received by a single receiving
phase difference equal to that of the received signals. If antenna ( A T ) at the left side receiving station, an ampli-
the propagation conditions in both directions are nearly the tude modulation component is generated on the received
same(thisassumption is reasonable for the horizontal signal which is the same as was found in the receiving
andnot largelyseparatedtransmittingand receiving system described before. This Ah4 component is detected
610 IEEE TItSNShCTIONS
COYBIUNICATION ON TECHNOLOGY AUGUST 1967

Fig. 11. Schematic block diagram of a space diversity transmission

system with individnal phase control.
L J

Fig. 9. Schematic block diagram of a space di- Phase Control Characteristics as a Servo Loop
versity receiving system with individualphase
control. The preferable operating conditions for a servo-mecha-
nism are, in general, aftercompromising between rise-time
21P
and over-shoot,[71
a a 0
gain margin 10 to 20 dB

Bb
phase margin 40 to 60 degrees.
Therefore, the phase control system used in the space
diversitysystem could also be designed so as to meet
these requirements.
0 From the viewpoint of rapid response, the moment-of-
inertia of the rotating part of the phase shifter should be
as small as possible. I n general, however, some minimum
limitexistsfor the moment-of-inertia, and, unless t.he
requirements stated previously aremet, some com-
pensatingnetworkmustbeinsertedinto the feedback
loop in order to improve the stability of phase control.
A rotary phase shifter for individual phase control can
generally be made smaller than for common phase control
since a narrow frequency band characteristic is allowable.
Therefore, more rapidresponse is expected in an individual
phasecontrolsystem.Figure 12 is a photograph of a
rotary phase shifter for a 6-GHz individual phase control
system.
I n Fig. 13, curvesB and Aillustrate the gain-phase
characteristics,respectively, withandwithout aphasc
compensating network for individual phase control. The
curve B adequately meets the requirements stated pre-
viously.
Some Improvements in Recent Years
Some improvements havebeen carried out on individual
Fig. 10. Amplitude
modulation
generated by
combining two signals. phase control systems in recent years.
On a propagation path over the sea or along a coast
line, the changing of the amplitude or phase angle of
by the AGC circuit of the main IF amplifier of a receiver received signals is occasionally more rapid than inland.
(R) ineachradiochannel.The'output f, signals are Therefore, there are possibilities of losing adequate
transmitted back to the right side ;station through a con- phase control. I n this case, the response of the phase shifter
trol link (TR) and control the APS ofleach radio channel should be as rapid as possible. An improvement of about
at the right side station. 10 times has been obtained by using a phase sensing fre-
Thus, the phase difference betw1:en the divided trans- quency of 405 Hz, along with aphaseshifterwitha
lnitted signals canbe controlled so a s t o be always in phase lightened rotor which is driven by a motor operated a t
with the radiochannel at thereceiving station. that frequency.
M41ClNO AND MORITA: SPACE1)IVERSlTY SYSTEMS FOR MICROWAVELINKS 611

-
Fig. 12. Rot,ary Phase shifter for a 6-GHz individual phase contxol
system.

-10
Distance [ km I

(a)
70.6

220 ZOO 490 /dO /do /20 ra,

-a
Phase (degrees) flW fly b9 f
C r PO 84 74 ry).rO4030 1D 10 J / Q, o,~/

Fig. 13. Example of gain phase charac- (%I

teristics of an individrlal phase control
system. Cumulative P r o b a b i l i t y
(b)
On the other hand, the loop gain of the phase control Fig. 14. (a) Profile of the propagation path between Nakatsu-
system changes together with
the changing gainof the main ham and Horigoshi. (b) Cumulative probability distribution
TF amplifier with an AGC circuit in the microwave re- of the received signal level.
peater, since the IF amplifier is used as a part of the loop-
circuit. A.s this condition is rather undesirable for stable Figure16(b) shows a diversitygain of 12.5 dB at a
operation,aparticularmethodhas been developed to probability of one percent, which is almost the maximum
stabilize the loop-gain. Each received signal is given an gain expected on this path.
amplitudemodulationbya low frequency oscillator a t
a, frequency of 230 Hz. The loop gain is controlled so as VII. A.PPLYING CRITERIA OF THE SPACE DIVERSITY
tomakeconstant theoutputamplitude of the 230-Hz SYSTEM TO THE DESIGN OF LINE-OF-SIGHT MICROWAVE
signal by controlling the gain of the low frequency ampli- RELAY SYSTEMS
fier with the AGCcircuit. Thus, the loopgaincanbe Line-of-sight microwave linksshouldbe designed so
stabilized in spiteof the changing gain of the IF amplifier. toas meet CCIR transmission It has beem
determined, however, that among the four objectives, if
VI. EXAMPLES OF DIVERSITY EFFECT
the hourly mean noise power (7500 pW) and interruption
To date, much data on propagation characteristics and (0.01 percent, lo6 pW) objectives aremet,theother
diversity effects have been obtained. The data illustrated objectives can also be adequately met. Since the hourly
here with theirprofiles of the propagation paths are typicalmean noise power and interruptions depend closely upon
examples. severe fading in which the probability distribution of re-
Figures 14(b) and 15(b)show the diversity effects of the ceived power is given by Rayleigh’s formula, it is very
common phase control system and Fig. 16(b) shows that helpful in link design to be able t o predict the occurrence
of the individual phase control system. probability (or time-rate) of severefading for appro-
In Fig. 14(b), a diversity gain of 10 to 13dB is obtained- priately long time periods (including the worsk month) on
a t a probability of one percent. The effects of frequency various propagation paths, anda t various frequencies.
correlation are also shown. A s shown in Fig.17(a), an appropriately long time period
Figure15(b) is for a number of periods undersevcre T, including the worst month (usually two orthree months
fading conditions on a propagation path over the sea. in the summer season) may be considered t o find out Zrn,
 IEEE TRANSACTIONS O N COMMUNICATION
TECHNOLOGY AUGUST 1967

Chika awa

Distance f k m I

-/0

-a

- 30

Fig. 15. (a) Profile of the prc,pagattion path be- (!%I

tween Chikagawa and Shiokubi. (ID) Cumulative
probabilit,y dist,ribution of the received signal

.-
level. Cumulative Probability

(b)
Fig. 16. (a) Profile of the propagation path between Niyama and
Noboribetsu. (b) Cum1dative probability distribution of the re-
ceived signal level.

---
m--- -Free Space Level
Thermal Noise
Wo
No

Irite'rruption l e v e l Ws
I
Rayleigh Rayleigh Rayleigh
Thermal Noise NS

(a)

Free Space Level Wo

Thermal Noise fi0

-------- .-. - ...-.-..-..._I-____ I .-.___, I

- Ilnherruption Level W E
Thermal Noise Ns-

Rayleigh
(b)
Fig. 17. (a) and (b) Occurrence time-rate of severe fading.
SYSTEMS
AND DIVERSITY
MAKINOSPACE MORITA: FOR
LINKS
MICROWAVE 613

the sum of fading time periodswhich would frequently ap- Therefore pl becomes
pear in the season. I n aiming a t a reduction of interrup-
N o
tions, shallow fadings can be excluded. Then, Z r f l / T can p1 = P o -. (9)
be defined as anoccurrence probability (or time-rate) po of N s
severefadingfor T . The probability of severefading I n aiming at the interruption objective, the value of N s
shouldbeincreased as the propagation path-length be- is lo6 p W according to the CCIR transnlission objectives.
comes longer. Then pl in percent is given by
The following empirical formula gives the occurrence
probability po of severe fading as a result of propagation No
pl = - .p o . 100 (percent) (10)
tests in theworst month for many years inJapan: 106
where N o is the thermalnoise power in a telephone channel
in free-space propagation.
where On the other hand, assuminga homogeneous modem
section (i.e., from modulator t o demodulator)with dis-

k Q . t)'2
. .
tance 8, the probability of allowable interruptions for this
section is given as follows
Q 2.1 x (over the mountains) d
p = - X 0.01 (percent) = 0.0011 (percent) (11)
5.1 X lop9 (over a flat plain) D
1.9 x (over the sea)
where D, a length of the hypothetical reference circuit,
f frequencyin GHz
is 2500 km, and 8 is about 280 km, since the hypothetical
d path lengthinkm
reference circuit is divided into nine honlogeneous modem
a 3.5.
sections.
If the sum of pls of several propagation paths included
Of course, since the empiricallyobtainedformula (5)
in a modem section is larger than p , space diversity should
was originally based on a number of propagation data,
be applied over the unfavorable path or paths in the sec-
actual valuesof po vary around the mean value calculated
tion.
from formula ( 5 ) .
For example, an occurrence probability po of severe
For each fading time period T,, the probability distribu-
fading which will occur on a path whose length d is 86
tion of received signal levels is given by Rayleigh's
km over a flat plain,at a frequency of 4 GHz, in the worst
formula, in which an attenuation from the mean level
month, can be estimated by ( 5 ) and pl by (10) as follows:
coincides with its probability asfollows :
po = 3.02 percent
1 percent -20 dB pl = 0.0018 percent
0.1 percent - 30 dB
0.01 percent -40 dB, etc. where 60 p W is taken asa valueof N o .
The value of pl is larger than p only by itself, therefore
Then, noting the free space level (nearly mean level) space diversityshould be applied to this path.
W oand the interruption level W s , the interruption prob-
ability p , for a total fading time periodis given by VIII. CONCLUSION
A set of guideposts to design space diversity receiving
ws
Pn = -
and transmitting systemsfor line-of-sight microwave links
wo has been presented. The relationships between diversity
gain, that is, a measure of the diversity effect, and space
because the Rayleigh probability distribution is still kept or frequency correlation coefficients have been quantita-
for a total fading time periodZ T , as shownin Fig. 17(b). tively explored. It has also been shown that the criteria
Accordingly, the interruption time-ratepl for T is given forapplyinga common phasecontrolsystemare con-
by siderably morestrict withrespect to frequency correlation.
Outlines of space diversity equipment used a t present
Pl = pop, = p , * -w, . and of some improvementsinrecentyearshave been
wo introduced.
Finally, necessary criteria for the application of space
On the other hand, thermal noise power in a telephone diversity systems to line-of-sight microwave relay links
channel is inverselyproportional to the received signal have been presented, employing the concept of occurrence
power, that is probability of Rayleigh fading. The space diversity tech-
niques have now become usable in the design of micro-
wavelinks, owing to the aforementioned successful re-
search.
614 IEEE T R A N S A C T I O N S O N COMMUNICATION T E C H N O L O G Y AUGUST 1867

The microwave relay systemsdeveloped by N T T satisfy

the CCIRtransmission objectives through the application
of space diversity techniques.
While the use of space diversity lmhniques as protection
against interruptions caused by set’ere fadingis the subject
of this paper, this technique also substantially assures the
improvement not only of an hourly nnean signal-to-noise
ratio of a link, butalso FM distortion which may occur on
a propagation path of a heavy loadedmicrowave link. For
quantitative analysis, however, considerations different
from those of interruptions areneelied, and these arebeing
carried out a t present (1966) in the Electrical Conmmnica- peaters,spacediversity receiving
tion Laboratoryof NTT. andtransmitting systems for 4- a:ld 6-GHz microwave links, an
11-GHz short haul microwave system, aud a 6-GHz atlxiliary radio
repeater for a 6-GHz microwave system. At prescllt, he is engtged i n
ACRNOWLEDGivIENT research in microwave propagatioll :tnd 011 radio t,ro.lmnission
I n carryingoutthisstudy,theauthorshave been systems. He is now the Director of the Radio Propngtt-ion Section of
the ECL.
indebted to many other staff memberswith the Radio Dr. Makino is a member of theInslitute o f E1ectric:al Com-
TransmissionSection of the Ele1:trical Comnlunication munication Engineers of Japan.
Laboratory(ECL), especially t o T. Morozumi and Y.
Numano, formermemberswith the ECL, for theirin-
valuablecooperation, and t o M. Hara for hiswork on
improvements.Theyfurther paJ‘theirrespects to .!I
Ikeganli with the ECL for his advice, and to .Prof. R f .
Nakagami of Kobe University, Japan! for his theoretical
analyses.
REFERENCES
Kazuo Morita was born in Tokyo,
[I] K.Kakitaand AI. Morita,“Fading in microwave relays,” Japan, on March 6, 1927. He
Rev. ECL, Japan, vol. 6, pp. 352-357, 1958. graduated from the Dependent
121 L. Lewm, “Diversity reception andautomatic phase cor- Technical College of Waseda Uni-
rection,” Proc. ZEE?vol. 109, pp. 295-304, 19F2.
H. Enomoto, “Fading in line-of-sight microwave propagation,” versity, Japan, in 1950. He re-
J . Znst. Elect. Commun. Engrgs, Japan, vol. 41, 1958. ceived the Ph. D. degree from Kyoto
f4I H. Makino, T. Morozumi, K. Morita, and Y. Numano, “The University, Kyoto, Japan, in 1961.
space diversitysystem for line-of-sight microwave link,” Rev. Since 1950, he has worked on the
E C L , N T T , Japan, V O ~13,
. pp. 1137-1138, 1964. problem of microwave propagation
is] H. Enomoto, “Freqrlency modulrLted signal distortion by in the Radio Transmission Section,
fading,” J . Inst. Elect. Comrnun. Engrgs, Japan, vol. 39, 1956. Electrical
CommunicationLab-
[el M. Nakagami and S. Kaneku, “Certain characteristics of com-
oratory,
Nippon
Telegraph and
mon phasecontrolspace diversity,” Rev. ECL, A‘TT, vol. 13, pp. Telephone Public Corp., where he
75-54, January-February 1965.
E. B. Ferrel, “The servo problem as a transmission problem,” is a Senior Engineer.
Proc. ZEE, vol. 33, pp. 763-767, November 1945: Dr.Morita is a member of the
[*I C C I R Doc. 10th Plenary Assembljy, Geneva, vol. 4, recom- Institute of Electrical Communica-
mendation 393, pp. 60, 1963. tion Engineers of Japan.