switching and Traffic 349

100 ms _
150
td = average caU delays = 2 (1 - Q.665) - ms

The average deJay on call processing during the busy call equals 150
JJ1S x 0.665 = 100 ms.
When comparing the centralized system with the decentralized ~ys-
we find that in the centralized system the average call delay time
teD1,
. · processmg
h shorter but its · much 1ess. However, the
· capac1·ty 1s
:e:~~ralized system is mo~e flexible, is easier to install, and has a
greater potential for expansion.

11 .1.5 Call drops

Call drops are caused by factors such as (1) unsuccessful completion
from set-up channel to voice channel, (2) blocking of handoffs _(switch-
. capacity), (3) unsuccessful handoffs (processor delay), (4) 1nterfer-
:O~e (foreign source), and (5) improper setting of sys~m parameters.
The percentage of call drops is expressed as -
number of call drops before completion
pcc1 = total number of accepted calls handled by set-up channels

ct
Because this percentage is based on many parameters, there is no
analytic equation. But when the number of call drops increases, we
have to find out why and take corrective action. The general rule is
that unless an abnormal situation prevails, call drops usually should
be less than 5 percent.

11.2 Cellular Analog Switching Equipment

11~2.1 Description of analog

switching equipment
Most analog switching equipment consists of processors, memory,
switching network, trunk circuitry and miscellaneous service circuitry.
The control is usually centralized, and there is always some degree of
redundancy. A common control system is shown in Fig. 1L3. The pro-
grams are stored in the memory that provides the logic for controlling
telephone call~. The processor and the memory for programming and
calls are duplicated. The switching network provides a means for in-
 ·~ C. a?

J~aves celJ site A of system A and enters cell site B of syatem B. Cell
81
tes A and B are oontl'olJed by uv0 d ""' . - · •
unit 8 - aJ b omes weak - tuetent MTSO!. When the mobile
. ign ec in cell site A MTSO A h
clidate cell site in its system and n., searc eg ,or a c~-
the handoff request to MTSO :nnot find one. Then MTSO A sends
MTso A and MTSO B, and MTso th rough a dedicated line betw~en
the call conversation. This is . B makes a complete handoff dunng
st
are many ways of implemenJ: . a one-point connection case. There
the actual circumstances ,:,0 _g mtersystem handoffs, depending on
tured by different companies th
. r, r mstan ·r
ce, 1 two MTSOs are manufac-
before implementation of int en compatibility must be determined
detailed discussion of this top~rsyste m handoff can be considered. A
Ic appears in Sec. 11.4.
-
9.10 Introduction to Dropped Call Rate _.,,,,
/

The definition of dropped call rate Th d ..

after the call is established b t ·b c e . e~rution of a dropped call is
u e1ore it is prope 1 t . t d Th
definition of "the call is establ · h d" r Y ermma e . e
pletely by the setup channel. ;;t~ere~::npso!hs~bt_lt~te cafll is lsledtup codm-
·1 bl •. h
to no avai a e voice c annels this is count d
1 I 1 Y o a ca rop ue
bl
dropped call. ' e as a ocked call not a
If there is a possibility that a call will drop d t th ·
· d · h . . ue o e poor s1gna1
of the ass1gne v01ce c annel, ~his Is considered a dropped call. This
case can ha?pen ~he~ the mobile or portable units are at a standstill
and the radio earner 1s changed from a strong setup channel to a weak
voice channel _due to the selective frequency fading phenomenon.
The perception of dropped call rate by the subscribers can be higher
due to:

1. The subscriber unit not functioning properly (needs repair).

2. The user operating the portable unit in a vehicle (misused).
3. The user not knowing _how to get the best reception from a portable
unit (needs education).

Consideration of dropped calls. In principle, dropped call rate can be

set very low if we do not need to maintain the voice quality. The
dropped call rate and the specified voice quality level are inversely
proportional. In designing a commercial system, the specified voice
quality level is given relating to how much C/1 (or C/N) the speech
coder can tolerate. By maintaining a certain voice quality level, the
dropped call rate can be calculated by taking the following factors into
consideration:

l
298 Ol'tllJ'f~ ,tf,r,e h ercentage (saY 90%) that all
v1· de ~igna1 _coverAge basebd on at !~en signal level.
• t,he received
_
~ s1gn. J will be a ove 1c,- d adjacent channel 1nterfer-
8 ·
•
l. Pro 1
2 Maintain the spcc1·t, ed co-chflfl11e
. busan hour, i.e., the wors t mterfer-
·
. l vels in each cell dunng a y
ence case.
e ll d pped rate is calculated as pos-
3 Since the perfoflll"-nce of the tcil f~~rn the radio link to the PSTN
· sible call dropping · in cvefY. s age
f the handoff 1n• the network will· be
connection, the response tune osrnall the response time for a hand
a fact-or when tl,e cell becomes • to reduce the call dropped·
rt r in order
1
off request has to be s 10 e
rate. . . ,. A uo a 1gon'th m w1·11 also
f th handoff and the 1vu-u•
O
4_ The s1gnahng e
impact the call dropped rate. ..
The relationship among the voice quality, system capacity and call
· dropped rate can be elCpressed through a common parameter C/I.
5
Relationship among capacity, voice quality, dropped call rate. Radio Ca-

pacity m is e1'pressed as follows:

BrlBc (9.10-1)

m= Jf(cll:.
where BrfB, is the total number of voice channels. Br!B,
is a given
number, and (C/1) is a required Cll for designing a system. The above
equation is obtained8
based on six co-channel interferers which occur
in busy traffic, i.e., a worst case. In an interference limited system
the adjacent channel interference has only a secondary effect. The der'.
ivation of Eq. (9.10:1) will be e"pressed in Chap. 13. Eq. (9.10-1) can
be changed to the following form:
(C/Ils =~(Br;:} = (!:Y ·;. (9.10-2)

Sine? t~e (C/Ils is a required Gil for designing a system, the voice
the r d" •t . . s 1s reduce d,
quahty 1s based on the (C/1)8 • When the specified (C /I) ·
a io capacr y is mcreased. When the measured (CI I) is less th
!1:U::"cified (C/I)s, both poor voice quality and dropped calls c:::

always
Coverageuses the equal-stren
of 90% coverag : th ;ntour. The coverage in cellular cells
90 t
tion tool (Lee Model) des:;bed::. ~qual-S ~ength contour: The predic-
strength contour at level C with 50 %
ha~- 4 isand
time used to predict
50% area in the equal-
a cell. For
 Handoffs and Dropped Calls 299

example,
-120 dBm. let If
C C= ::: 102 dBm, which is 18 dB above the ambient noise
;ncrease the lev - - 102 dBm is 50% equal-strength contour, then
the following eqe 1 C + 10 dB contour which can be calculated from
uation:

P (x' < A) = JA
-oo
1
~CT exp
[(y' - m)2] dy
2CT2
I

(9.10-3)

(9.10-3)
£q.sired signalislevel
the and
cumulat· . .
- . ,ve distribution function where A is the
e mis the mean I I · 1 t £'. d' d
dto terrain contour. If A - C + 10 = -92 eve." ,s ong- erm
dB and CT= 8 dB: ,a ing ue

S =p (x < -92 - 8(-102)) = p ( X < B

10) = 0.9082 (9.10-4)

Eq. (9.10-4) can also be interpreted as being at a -92 dBm contour,

the s,gnal above the level of -102 dBm is 90.8%. Of course, the level
of - 102 dBm is determined to be 18 dB above -120 dBm which is the
ambient noise level. The (C/N)s of 18 dB is the required level for
getting a voice quality. ',,.

~ rmula of Dropped Call Rate

The dropped call rate can be calculated either using general formula
or by a commonly used formula.

General formula of dropped call rate.The general formula of dropped

call rate P in a whole system can be expres~~d as:

p =1 - [1 l=I "• .
a)C' P.
(9.11-1)

where (9.11-2)

P. is the probability of a dropped call when the call has gone through
n handoffs and 2 (9.11-3)
X = (1 - 8)(1 - µ)(1 - 6v)(l - ~) •
 300 Chapter Nine

& = Probability that the signal is below the specified receive threshold (in a
noise-limited system).
µ. = Probability that the signal is below the specified cochannel interference
level (in an interference-limited syS t em).
T = Probability that no traffic channel is available upon handoff attempt

when moving into a new cell.

6 = Probability that the call will return to the original cell.
[3 = Probability of blocking circuits between BSC and MSC during handoff.
ex" = The weighted value for those calls having n handoffs, and ~~=o ex" = 1
N = N is the highest number of handoffs for those calls.

Eq. (9.11-3) needs to be explained clearly as follows:

(1) 2 and 22 are two events, z 1 is the case of no traffic channel in the
c~ll, z2 is the case of no-safe return to original cell. Assuming that
Z1 and Z2 are independent events, then
-. P(z 2lz 1) • P(z 1) = P(z 1) • P(z 1) = 0 • T
I
' (2) (1 - J3) is the probability of a call successfully connecting from the
old BSC to the MSC. Also, (1 - J3) is the probability of a call
successfully connecting from the MSC to the new BSC. Then the
total probability of having a successful call connection is:

BSC (old)- MSC (1 - J3))(1 - J3)2

MSC - BSC (new) (1 - J3)

(3) The call dropped rate P expressed in Eq. (9.11-1) can be specified
in two cases:
1. In a noise limited system (startup system): there is no fre-
quency reuse, the call dropped rate PA is based on the signal
coverage. It can also be calculated under busy hour conditions.
In a noise-limited environment (for worst case)
6 -- 01

0
T=T}
= e: the conditions for the noise limited case
J3 = J31
2· In an inte_rference-limited system (mature system): fre-
quency reuse is applied, and the dropped rate p is based on
8
 Handoffs and Dropped Calls . 301

•nterference level. It can be calculated under busy hour

the 1
conditions.
In an interference-limited environment (for worst case)

0,-='T}
= 0: the conditions for the interference limited case
J3 = J32
Eq. (9.11-1) has to make a distinguished difference between PA
and pB· The cases of PA and PB do not occur at the same time.
When capacity is based on frequency reuse, the interference
level is high, the size of the cells is small, and coverage is not
an issue. The call dropped rate totally depends on interference.

commonly used formula of dropped call rate. In a commonly used for-

mula of dropped call rate, the values of T, 0, and~ are assumed to be
very small and can be neglected. Then Eq. (9.11-3) becomes:
X = (1 - 8)(1 - µ) (9.11-4)

Furthermore, in a noise-limited case, µ - 0, Eq. (9.11-1) becomes:

N
PA = L cxnPn = L <Xn[l - (1 - 8)n] (9.11-5)
n=O

and in an interference-limited system, 8 - 0, Eq. (9.11-1) becomes:

N
PB = I cxnPn = I an[l - (1 - µ)n] (9.11-6)
n=O

Handoff distribution of calls, The an is the weight value for those

«n•
calls having n handoffs. Then the handoff distribution of all an's is
needed for calculating Eq. (9.11-1), or Eq. (9.11-5), or Eq. (9.11-6). The
relationship of all an's is:
N

n=O
I an= 1

TThhe handoff distribution of calls a can be assumed as follows:

,
e ct n in macrocells is used for calculating
n
the dropped call rate PA:
 302 Chapter Ntne

K .:::
ds::..:o:.:_f::._
_ in
_' U,: ni::
ts:___ __:_:
" _::H::an::d=ofli
:,:e:..:P..:.er:...C
:..&
_IJ_ _ _Pe_rc_e_nt:--:-o-:-:-
f _U_ni_ts_ _ _ a.
100% ---
Handset Units n "' 0 20% °o
0
-= 1
Mobile Units n = 0 60% o = 0.2
n == 1 20% a, :: 0.6
n= 2 °2 :: 0.2
The a" in microcells is used for calculating the dropped call rate p8 :

Kinds of Units
Handset Units
n Handoff's Per Call
n=O
n= 1
Percent of Units
80%
20%
CXo
<l1
-
a.
= 0.8
= 0.2
20% <lo = 0.2
n=O
Mobile Units 60% <l1 = 0.6
n= 1 01 2 = 0.2
20%
n= 2

The values of a" are used for calculating the dropped call rate. For
instance, calculating the general formular of dropp~d cal~ rate (Eq.
/ (9.11-1)) in macrocells (noise-limited system) for mobile umts.
2
PA = 1 - [0.2X0 + 0.6X 1 + 0.2X ]

= 0.2P0 + 0.6P1 + 0.2P2 (9.11-7)

where Xis expressed in Eq. (9.11-3). In Eq. (9.11-3), the values of 1 ,

r
e, and 13 are usually small. Therefore, the value of X is heavily de-
pendent on 6 and µ.

1 9.12 Finding the Values of 6 and µ.

The values of 6 and µ can be derived for a single cell case and in the
case of a handoff. The single cell cas~ solution is used for estimating
the blocked calls. The reason behind this is that the probability of 8
and µ in a single case is used for the blocked call rate of setting up
calls. Assuming that after a call is set up, the call will not be dropped
in a cell until the mobile unit travels into the handoff region.

Fonnula for 6 and µ.. We first find the value of 6 in a single cell by
integrating Eq. (9.10-3) over a whole cell to find the area Qin which
the measured x will be greater than A(r) - m!<J'. The mean value m
is a specified receive level. A is the signal level which is a function of
A(r) that exceeds m at the distance r which is less or equal to the cell
radius R. .

Q = Jo
(R (
P x >
A(r) -
<J'
m) • 2-rrrdr (9.12-1)
 H•tdofl'9 and Dropped Calla 303

b bility 8 that the signal is below a specified receive threshold

!}1~ paron; se-Iimit.ed environment system is
,n JD
-rrR2 - Q
S = 1rR2

= 1- ,,!, f (1 - P (x < A(r)"- m)) ·2,rrdr (9.12-2)

robability µ. that the signal is below the specified signal level C

fbe ptbe interference level/ in an interference-limited system can also
0.,er
be eXPressed as:
'TTR2 - Q
µ, = 'TTR2

= 1 - ,,~• f( 1 - P (x < A(r)"- C)) ·2-rrr • dr (9.12-3)

;.1
e xnaY use the numerical calculation to solve Eq. (9.12-2) and Eq.
2_3) for drop~d calls due to handoffs.

ulation of 6 and µ. in a single cell. 6 is calculated numerically in a

1
ca. ce-1:-ited case. The cell can be divided into five rings as shown in
no1s .u.uu
Fig, 9,ll. Eq. (9.12-2), then can be expressed as:

~
5

P·
(
X > Ai(ri) - m)
__::........;..___ • a-
6 =1 - i=l '
_____1T_R_2_ _ __
<1 '
(9.12-4)

where
A -(r.) - m)
1 - p . ( x < ·' '
' u = P·' ( x > A;(r;)(] - m)

ai = 1r[2i - lJrf
5
(9.12-5)
La;= 1rR2
i=l

in asingle cell. A5(r5 = R) is the desired signal level at the cell radius
R::: 5r1, Let .

l
 _/_ !

304 Chapter N\ne

. X
A(r.)
I I - m) -p
PI ( >
- - i
(J

for simplicity. Eq. (9.12-4) can also be expressed as:

5
li (1 - p) · ai (9.12-6)
8 :::: _!_:;i=.!-1- - - -
-rrR2

Eq. (9.12-6) is also the equation for obtaining the value of I'- in the
interference case.
&, and v-. are improved due to the natural two-site diversity in the handofl
region. Due to natural situations providing equivalent two-site diver-
sity in the handoff region, in region a,, the probability of dropping a
call is reduced by 1 - (1 - p 5) 2 as compared with p,. In region a,, the
probability of dropping a call is 1 - (1 - p 4)(1 - p 6) as compared with
p,. p is tho probability of a dropped call due to the fact that the
handoff
6 takes place in a 4 by the new cell coverage. Therefore, &h and
f.Lh are expressed as:

(1 - p5)2a5 + (1 - p4)(1 - Ps)a4 + (1 - p3)a3 + (1 - P2)a2 + (1 P1)a1

1TR2

(9.12-7)

~e aware tha~ _Pi is the probability of having a successful call and p 1

1s the probab1hty of a dropped call.
Example 9.1 Given cr = 6 m = _ 104 dB
value of & during h d 'ff? (S . m, A 5 -_ -96 dBm, find the
h a an o . ee Fig. 9 11 )
Based
dBm, A,on =
the-80 dB/de:ru=le'...we can obtam
40 dBm ·. ·A, = -92 dBm, A, = -87
'
1 68 dBm, As = -99 dBm and also

p (X < -96 - 6(-104)) = 0.9082, p4 = 0.948, p 3 = 0.9977,

5

P2 = l, P1 = 1, Ps = 0.7967
Then applied to Eq. (9.12-7), we obtain
 Handoffs and Dropped Calls 305

High Problblllty
Hlndoff Region

--------- ,,,
.............
''
''
''
''
''
'I
' I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

' I
' I

-- _______ _
' I
'
......... ... ,,,.,"' I

__ ,,

Figure 9.11 The diagram for calculating the dropped calls due to handoffs.

6h ::
(1- p )2a5 + (1 - p4)(l - P6)a4 + (1 - Pa)aa + (1 - P2)a2 + (1 - P1)a1
5
-rrR2

= 0.64%

example 9.2 Given a = 6, I = -104 dBm, CII = 12 dB, and the signal
received is requested to be 8 dB above the average C/1, find the value
of µh during a handoff? Based on the 40 dB/dee rule, C = -92 dBm
and A = -84 dBm, we obtain A 1 = -50 dBm, A 2 = -62 dBm, A3 =
5
-75 dBm, and A = -80 dBm, A 6 = -87 dBm. Then applying Eq.
4
(9.12-7), we find:

JJ-h =
(1 - Ps)2a5 + (1 - p 4)(l - Pa)a• + (1 - Pa)aa + (1 - P2)a2 + (1 - P1)a1
-rrR2

= 1.45%

References
1. "Advanced Mobile Phone Services," Special Issue, Bell System Technical Journal, Vol.
68, January 1979.