Content uploaded by Sirin Tekinay
Author content
All content in this area was uploaded by Sirin Tekinay on Sep 11, 2019
Content may be subject to copyright.
Handover
and
Channel Assjanment
J
in
Mobile
Cellular Networks
Quick and
timely
handover has a crucial effect on how users
perceive quality
of
service, however, handover strategies should
not
be
too
complicated.
~
Sirin Tekinay and Bijan Jabbari
Bijan
Jabbari
IS
~SSOCIB~P
pro
iesJor
of Electrical
ond
Com-
puter
EnRineering
at
George
Mason
Linwerqy'J
School
of
Information Tcrhnologv
and
Engineering
in
Fa~rfm,
Virginia
.Pnn
Zekinay
/Studen(
Memhpr,
IEEE)
LY
c~wendy
completing requiremmrs
for
rhe
Ph.
D.
degree
at
Georg~
Mason
Unharsirys School
qf
lnfornlation Technology and
Engineering
In
Farfar,
Virgmia
hc rapid growth in the demand for
mobile communications has led the
industry into intense research and
development efforts toward\
a
new
generation of cellular systems. One
of
theimportant objectives in the devel-
opmentofthenewgenerationisimprovingthequal-
ityofcellularservice,with
handoversnearlyinvisiblc
to the Mobile Subscriber
(MS).
In
general, the
handover function isamost frequentlyencountered
network function and has direct impact
on
the
perceived quality of servicc. It provides continua-
tion of calls as the MS travels across cell bound-
aries, where new channels are assigned by thc
new Base Station
(BS)
and the Mobile Switching
Center
(MSC).
The system performancc characteristics include
probability
of
blocking of new traffic, probability
of forced termination
of
ongoing calls, delay in chan-
nel assignment, and total carried traffic. There
is
a
tradeoff between the quality of service and
implementation complcxity of the channel alloca-
tion algorithms, number of database lookups and
spectrum utilization.
In
selecting
a
channcl
assignment strategy, the objective is
to
achieve a high
degree of spectrum utilization for a given quality
of
service with thc least possible number
of
database lookups and simplest possible algo-
rithms employed at the
RS
and/or the
MSC.
Handover prioritizationschemesarechannel assign-
ment stratcgics that allocate channels to handover
requests more readily than originating calls. Prior-
itization schemes provide improved pcrformance
at the
expense
of reduction
in
the
total
admitted traffic.
Inthisarticle,weprovideataxonomyofthechan-
ne1 assignmcnt strategies along with the complex-
ity
in each cellular component. Next, we consider
various handover scenarios and the roles of the
BS
and MSC. We then discuss the prioritization
schemes and define the required intelligence dis-
tribution among the network components.
Strategies and Functionality
Efficient utilization of
the
scarce spectrum
allocated
€or
cellular communications is certainly
one of the
majorchallengesincellularsystem
design.
Alloftheproposedstrategiessuggest
thereusage
of
the same radio frequencies
in
noninterfering
cells. Channcl assignment strategies can be clas-
sified into fixed
[l],
flexible
[2]
and dynamic
[3]
(see Fig.
1).
Table
I
provides
a
summary
of
these
strategies, along with the role assumed by the
MSC with each of them. The MSC function corn-
mon to all channcl assignment strategies is the
storage and update of information on which MS
is being served
on
which channel. This informa-
tion
is
essential for network-directed criteria (involved
in other network functions as well) such as loca-
tion information
of
MSs, control traffic loads and
overall traffic loads. In the descriptions of vari-
OU~
channel assignment strategies that follow, we
focusonthecasewhereallcellsunderconsideration
belong to the same MSC.
Fixed
Channel Assignment Strategies
The common underlying theme in all
ficd
assign-
mentstrategiesisthepermanentassignmentofaset
of channels
to
each cell. The same set
or
radio
frequencies
is
reused by another cell at some dis-
tance away. The minimum distance at which
radiofrequenciescanbereusedwithnointerference
is
called the "cochannel reuse distance," which is
accepted to be three cell units in the seven-cell
cluster model.
The basic fixed assignment strategy (see Fig.
2)
implies that a call attempt at a cell site can
only be served by the unoccupied channels
of
the
predetermined set
of
channels at that cell site;
otherwise, the call is blocked. Here. the only role
of the MSC
is
to inform the new
BS,
and receive
a
confirmation or rejection mcssage from the new
BS,
about the handover. The
MSC
keeps track of
serving channels for the purpose of updating
stored information regarding the location of
the
MS.
Other fixed assignment methods are variations
of
the basic strategy describcd above, with vari-
ous channel-borrowing methods (see Fig.
3).
We
will demonstrate the role of the
MSC
with the
simple borrowing, hybrid assignment, and bor-
rowing-with-channcl-ordering
strategies.
42
0163-6804191R01.00
1991°
IEEE
IEEE
Communications
Magazine
November 1991
Channel Assignment Strategies
1
*
Flxed
Slmple
Borrowing
Scheduled Call-by~Call
Optimized
Predictive
Ordering
Borrowing with
Figure
I,
Clnssification
of
channel allocation
strategies
In the simple b~~rrc~\\,ingstrategy,
if
all
permanent
channels
0L
a cell are busy, a channel can be bor-
rowed
from
a neighboring cell, provided that this
channel does not interfere with the existing calls.
When a channel is borrowed, additional cells are
prohibired from using it. The MSC supervises
the borrowing procedure. following an algorithm
that favors channels of cells with the most unoc-
cupied channels to be borrowcd. The algorithm
“locks” the borrowed channel toward the cells
that are one
or
two
cell
units away from the borrower
cells. The MSC keeps record
of
free, serving and
borrowed (therefore, locked) channels and
informs all involved
BSs
about locked channels. The
rewardofincrcascdstoragerequirementattheMSC
and the need for database
lookups
is
a
lower call
blocking prohahility up
to
a
certain traffic level.
In heavy traffic,smce
borrowedchannelsarelocked
for
at least five additional cells, channel utiliza-
tion efficiency is degraded.
This trend is improved by the hybrid channel
asqignment strategy proposed in
[4].
In
this strat-
egy. permanent channels of a cell
are
divided into
two groups: one group can be uscd only locally,
i.e.. within the ccll; the other can be borrowed.
Tncratioofthenumbersofchannelsinthet\~ogroup
is determined
a
priori.
depending
on
an estima-
tion ofthe traffic conditions.
In
addition to its duties
in
the simple borrowingstrategy, in the hybridchan-
nelassignmentstrate~,theMSChastolabelnllchnn-
nels with rcspcct to the group
to
which they belong.
The borrowin@-with-channel-ordering strategy
suggested in
[j]
introduces
a
further improve-
ment on the channel-borrowing concept. It elab-
orates
on
the idea of hybrid assignment by
dynamically varying the local-to-borrowable
channel ratio according tu the changing traffic
conditions. Each channel has adifferent adjustable
Figure
2.
Fixed channel assignment strategy.
.A
-
G
denote diflerent
sorts
of
channels permanently
assigned to cells
Assignment
I
MSC
Fundionalities
st
Fixed
Mignmeni
I
inform new
BS
Keep
track
of serving &ann@&
_.
If
free/serving/locked channels
5-1
Keep track
c
traffic load
to
each channel
Borrowing with call
or
borrowed
Channel Ordering Assign
a
probability
of
being either
used
for a local
Keep track of free/servingllocked channels
(predictive)
I
traffk
probability
of
being borrowed and
i’r
ranked with
respect to this probability,
so
that channels
toward the bottom
of
the list are more likely to
be borrowed. and vice vel-sa. Each time a call is
atlempted. an algorithm at either the
MSC
or
BS
is
run
to choose the most “appropriate” channel
among
all
frcc channels. looking at their associat-
cd probabilities.
If
this is part
of
the
BS
function-
ality, the
MS(:
must be informed
of
the resulting
assignment. The
MSC
determines and updates each
channel’s probability
of
being borrowed, based
011
the trafficconditions, by usingan adaplive algorithm.
The channel assignment strategy can be made
more complex by allowing intracellular handovcr,
i.e., immediate reallocation of a releascd higher-
rank channel to a call existing on a lower-rank
channel. The aim of such reallocation is to minimize
the number of calls on the relatively more “bor-
rowable” channels in order
to
rzducc
the locking
effect of borrowtd channels
in
additional cells. Rcal-
location is achieved by
a
comparison algorirhm
accommodated at either thc
BS
or
MSC,
which
Is
invoked each timc
a
channel
is
freed.
-
Figure
3.
Borrowing ytrategies. Channel a4
is
borrowed and
now
lucked
to
cells naarked
bitedfrorn using
a4
i‘
r
II
1%.
Cells marked
“X”
were alreadyproho-
DynamicChannelAssignment
Strategies
In
contrast
to
fixed assignment, in dynamicassign-
ment strategiescells hare
nochannels
to
thcmsclvcs
but
refer
all
call at[-empta tc1
the
MSC,
which
manages
all
channel
assignment in its region.
Each
time
a
call attempt arrives, the
BS
asks
the
MSC
forthe channel with the minimumcost
to
be
assigned.
The
cost
function dcpcnds
on
the future blocking
probability, usage
frequency
of
the
candidate
channel,
the
reuse
dlstancc
of
the channel,
and
so
on. The
MSC
decides, on
a
call-by-call basis,
which channel
to
assign
to
which call attempt by
searching for
thcavailablcchannelforwhich
thecost
function is minimum. It needs
to
have informa-
tlo~~regardir~gcl~ar~neloceupancydistrihrltionsunder
currenttrafficconditlvnsalidothernetwork-direct-
cd
criteria. as well as radio channel measure-
ments of individual
MSs.
Flexible Channel Assignment Strategies
Flexible channel assignment strategies com-
bine
aspects
of
both the fixed
and
dynamic
strate-
gics in
the
sense that each cell is assigned a set
of
permanent channels that typically will suffice
under
hght
trafficloads. The
MSC
holdsasetofflex-
ible
channels
and
assigns
thesc
to
cells
whoic
permanent channels
have
become inadequate
under
increasing traffic
loads.
The distribution of these
emergencychannelsamongthcccllsinncedofthem
is
carried out
hy
the
MSC
in either
a
scheduled or
predictive manner
[2].
Iftheflexiblechannelsarereasi~gnecior~asched-
uled
basis.
it
is assumed that future changes in
tr;ifficdisrributionarcpinpointedintimeandspace.
I'l~echar~geina~signmentofflcxiblcchannclsisthen
made at the predetermined
peaks
nf
traffic
change.
In the predicrivc assignment strategy, the traf-
ficintensityor,equi~aalcntlq.
thcblockingprobability,
is
conitantly
measured
at
cvery cell
site
so
that
the reallocat~on
ol
the
Ilexihle
channels can
bc
carried out by
the
MSC
at
any
point in time.
Flexible assignment strategies, like
call-hy-call
dynamic stratcgics. rsquire the
MSC
to
have
up-
to-dalc
information about the traffic pattern
in
its
area
and other network-dircctcd critcria in order
tn
manase
its set
of
tlcxihlc channels efficiently.
Possible Handover Scenarios
The channcl assignment 5tratcgics described
above
arc used whcncccr
a
ncw call
or
handovcr
reque\t
i$
rcceivcd
hy
rhe
RS
or
MSC.
Some
assgnment strateglcs prlorilix handover rcquests
in ol-dcr
to
protect ongoing calls from forced ter-
mination. Before describing
the
handover
priori-
tization schemes. wc rcvicw thc
handovcr
proccss
(see
Table
11).
'Ihedecisiun
lhat
a
handovcrahall
takc
placecm
be made by both
the
MS
and the
BS
by monitol-
ing
the
channel quality.
If
the decision is made
by
thc
MS
alonc,
a
handovcr
rcqucst
is procided
to the BS.
The
new
BS
is
determined by
the
MS
orLIS<:. Ifiti~determinetihytheR.1S,thecandid~rtc
BS
is provided
to
the
MSC.
We
note
that thc
decisions made
hy
the
MS
are based
on
radio
channel measurements only. whereas the
MSC
is
in a position
to
judge according
to
a
collection of
critcria,
includingner\l.ork-dircctedont.s~uch
aqthc
traffic dlstrlhution
111
the area.
Radio Channel Measurements
From theviewpoint ofthc network,
the
dctcction
ofthc
need
for
handover
and its timelycxccution
are
challenging
taika.
Momentary
tadinga in
the
communication channels between the
MS
and
BS
may occur due to geographical and cnvironmen-
tal
factors wcll within
thc
cell.
This
means
that
the
decrease
in
the power level
of
these channels
should
be observed for
a
certain
amount
of time
before it can be
concluded
that the
MS
is
actually
movingawayfromthcBS. Ontheotherhand,ifthe~-e
actually
IS
a
nerd for handover.
it
must
be
rcipond-
edtunssoonaspossibleinorderton~inimizcthcrisk
of
forced
termination
of
rhc
call. In order to
detect
thc
need
fur
hando\er.
the
MS
need\
to
take measurements on thc channcl
it
is currently
using
as
well
as
the
broadcast
channels
of
the
neighboring
cells.
Diffcrcnt standards for cellu-
laroperations kpecifi
different
procedures for these
Task
I
MS
I
BS
I
MSC
Radio
I
Make periodic measurements
04
Channel broadcast channels
Measurements current and neighboring
Send results
to
BS
Start measurements
Issue
Handover
Request
I
Monitor backwards channels
Give measurment order
to
MS
I
Send
results
to
BS
Send measurement results
to
MSC
Request handover
I
Evaluate handover request
Inform new
BS
Evaluate handover requests
Permit/drop/
delay (queue) handover request
Accept/block/
Request handover Inform new
BS
Task
MSG
New
BS
MS
Confirm/
Disconfirm
I
Handover delay handover
Table
2
Intelligence
distribution
among
IMS,
BS
and
.MSC
in
handover
procedures
measurements (see bibliography).
According to thc Telecommunications
Indus-
try
Awxiation (TIA) standards. thc
BS
monitors
the backward channels of
all
MSs
with which it is
communicating. When it detects a significant
drop in the power level, it sends the
MS
a
mea-
surement order. Upon receiving the measure-
ment order, the MSstarts takingmeasurements. The
measurementresultsarereportedtotheBSwiththe
frequeneyprescribedin themeasurement order. The
Pan-European
GSM
standards suggest that the
MSshould take measurcmentsall the time and report
the results periodically to
the
BS.
This eliminates
the need for the
RS
to constantly monitor
all
backward channels.
A
promising method
of
radio channel measuremcnts would be interac-
tivclyvarying the intervals behveen the taking and/or
reporting
of
measurements.
Roles
of
the
BS
and
MSC
in Handover
Procedures
The
BS
receiveseithermeaiurementresultsonly,
which
it
has
to
evaluate
to
decide whcther
a
han-
dover isnecessary,
ortheIlleasurementresult~
togeth-
er with thc next
BS
selected
by
the
MS.
In
the
first ease, the
BS
issues the handover request,
if
necessary,
and
sends
it
to the
MSC.
Then the
MSC
picks the best BS to serve the continuation
of
the call.
In
the second case, the
BS
merely
sends the MSC the request for handover to the
candidate
BS
specified by thc
MS.
In
both
cascs,
thc
MSC
informs the new
BS
of the handover request.
The new
BS,
depending
on
the channcl assign-
men1 strategy (and possibly the handover prioriti-
Lation scheme). may accept, block,
or
queue the
handovcr request. It informsthehfSCregarding the
status
of
the
handover rcquesr. Depending
on
the
responseofthenewBS,theMSCmaypermit.delay,
or
drop the handover request.
Handover
Policies
In
some channel assignment strategies, the
BS
handles handover rcquevtsin exactlythc same
man-
ner
as
it
handlcs originating calls. Obviously, such
schemes suggest that the probability
of
forced
termination
of
an ongoing call due to unsuccess-
ful handover equals the probability
of
blocking an
originating call. From the
MS’s
point
of
view,
however. forced termination of an ongoing call
is
sig-
nificantly less desirable than blocking
a
new call
attempt. Therefore, methods for decreasing the
probabilityofforced termination by prioritizing han-
dovers at the expense of a tolerable increase in
Guard
Channels
-
d
l
Figure
4.
Decrease in total trafic as afunction
of
the number of guard channels
[
61
1
BS
1
Figure
5.
Handover and receiver thresholds. Line;
motion
from
BS
1
to
BS
2
is
assumed; handover
must occur in
[tO,tl]
call blocking prohability have been devised
in
order to increase the quality
of
cellular service.
We now present
two
prioritization schemes.
The Guard Channel Concept
The
“guard channel” concept
wa\
introduced
in the mid-1980s
[h,
71.
It offers a generic means
of improving
the
probability
of
successful han-
dovcrbysimplyreservinganurnberofchannelsexclu-
sively for handovers. The remaining channels can
beshared equallybetweenhandovcrs andoriginating
calls.
The pcnalty is the reduction
of
total carried
traffic (see Fig.
4)
due
to
the fact that fewer chan-
nelsaregrantedtooriginatjngcalls,anditistheorig-
inatingcallsandnottheor~goingcallsthatreallyadd
to the total traffic. This disadvantage can be bypassed
by allowing the queuing of originating calls.
Intu-
itively, we can say that thc latter method is feaci-
blc because originating calls are considerably less
sensitive to delay than handover requests.
Another shortcoming
of
the employment
of
guard
channels, cspecially with fixed channel assign-
ment strategies, is the risk of inefficient spcctrum
utilization. Carefulestimationofchanneloccupancy
time distributions’ is eswntial in order
to
mini-
mize this risk by determining the optimum
num-
ber of guard channels.
With flexible
or
dynamic channel assignment
strategies, the guard channel concept is revisited
inamodifiedmanner.
Cellsdonotkeepguardchan-
nels
in their posscssion. The
MSC
can keep a
collection of channels
only
for handover requests,
oritcanhavcanumberofflexiblcchannelswithasso-
ciated probabilities
of
being allocated for han-
dovcr requests.
Queuing
of
Handover Requests
The queuing
of
handover rcque\ts, with or
without thcemployment ofguard channels, isanoth-
er gencric prioritization scheme offering rcduced
probability of forced termination. Thcre is again
a
tradeoffbehveen the increase in service quality and
the corresponding decrease in
total
carried traf-
fic. Before we discuss its conscquences,
we
briefly
describe this
5cheme.
Handover can occur in the hme interval
dur-
ingwhich the ratio of the power levels reccived from
the current and next BSs is betwccn the
‘.ban-
dover threshold” and the “receiver threshold”
(see Fig.
5).
The handover thrmhold is set at the
point where the power received
from
the
BS
of
a
ncighhoringcellsite hasstarted toexceed thepower
received from thc current
BS
by a ccrtain amount
-
With
flexible
or
dynamic
channel
assignment
strategies,
the
guard
channel
concept
is
revisited
in
a
mociified
mannm
IEEE
Cornmumcations
Magazine +November
1901
45
-
One
of
the
aims
of
our
current
research is
to
improve the
qualily
of
cellular
service
by
modifiing
the queue
discipline
in
queuing
handovers.
and/or for a certain time. The receiver threshold
is the point at which thc rcccived power from the
BS is at the minimum acceptable level. At this point,
sincecommunicatingwiththecurrentBSisnolonger
possible, the call will be terminated unless a suc-
cessful handover to an eligible cell has already
occurrcd. Queuing handover requests is made
possible hy theexistenccofthe timeinterval that the
MSspends between these
two
thresholds. The max-
imum possible waiting time in thc queue is given
by this interval. The allowable queue size needs
to be detcrmined. Computation of the queue
size requires knowledgc
of
the traffic pattern of
the area, the major factor
of
which is
thc
cxpect-
ed number of handover requests.
In
the case
of
high demand for handovers, the assumption of
infinite queue size introduces an undesirably
large decreasc in total carried traffic
[6].
Fur-
thermore, the probability
of
forccd termination
is
still strictly greater than zero, because thc han-
dover request canonly wait until the receiver thresh-
old is reached. This is why handover requests are
much more sensitive to delay in service than orig-
inating calls. Indeed, qucuing handovers has
been widely discussed; some are in favor of it bccause
of the decrease in the probability
of
forced termi-
nation
it
offers, while others argue that the delay
insensitivity
of
originating calls makes it more
feasible to queue new call attempts rather than
handover requests.
One of the aim of
our
current research is to
improve the quality of cellular service by moditj-
ing the queue discipline in queuing handovers
[9].
The queuing system is not viewed as "first come
first serve." A handovcr request is ranked accord-
ing to how close the
MS
stands to (and, possibly, how
fast it
is
approaching) the receiver level. Thc
necessary radio channel measurenlents are already
made; thcrcfore. the
only
additional complexity
in
implementing thc modification is a fairly sim-
ple comparison algorithm tobe run continuously
on
the stored handover requests.
Summary
In
this article, we have reviewed various handover
scenarios and suggested several ways of distribut-
ing intelligence betwecn the MS,
BS,
and
MSC,
focus-
ing
on
their respective roles in thcse scenarios.
We
have described the effect of different channel
assignment strategies and handover prioritization
schemes
on
BS
and
MSC
functions.
The main criteria used to compare the perfor-
mance
of
a
cellular systcm model under different
assumptions are probability of call blocking,
probability of forced termination, total carricd
traffic, delay in channel assignment, and number
of databasc
lookups.
These criteria together
define the cost function, the minimization
of
which, along with quality of servicc improvement,
is the objective. We have proposed a method of
prioritizing handover requests by queuing them
in such a way that the
one
with the maximum
probability
of
forccd termination is served first.
References
[1]M ZhangandT
P
Yum,"Compdr,sonsofChannel-AssignmentStrate-
gies
In
Cellular Moblle Telephone Systems."
IEEE
Trans.
on
Veh<cular
Tech, YOI
38, No". 1989
[2]
1
Talima and
K.
Imamura.
"A Strategy for Flexlble Channel
Assign-
ment
In
Moblle Communlcatlon Systems."lEEE Trans.
on
131
D.
C.
Cox and
D.
0
Reudnlk. "lncreaslng Channel Occupancy
In
Vehicular Tech,
vol
37,
May 1988.
Large
ScaleMob~leRad~oSystems:Dynam~cChannelAsslgnments,"
IEEE
Trans.
on
Vehicular
Tech.,
"01.
22.
NO".
1973.
[41T.J.KahwaandN.D.Georganar,"AHybr~dChannelAssignmentScheme
terns."
IEEE
Jrans.
on
Commvn..
vol.
26,
Apr.
1978.
~n
Large-Scale. Cellular-Structured MobileCommunlcatlon Sys-
151
5.
M. Elnoubl,
R.
Smgh. and
5.
C. Gupta. "A New Frequency Chan-
nel
Assgnment Algorithm in High Capacity Mobile Communl-
rat~onSystems,"lEEETrans.onVehicularJech.,vol.3l,Aug.1982.
[61
R.
A. Guerln, Ph D thesls. Dept.
of
Electrical
Englneerlng. CA
lnst
of
Tech..
Pasadena. CA. 1986
171
D.
Hong and
5.
5
Rappaport. "Trafflc Model and Performance
Analysts for Cellular Moblle Rad10 Telephone Systems with
Prl-
orftued and Nonprtarltized Handoff Procedures,"
IEEE
Trans.
on Vehicular Tech.. vol.
35.
Aug. 1986.
[SI
R.
A. Guerln. "Channel Occupancy Tlme Dlstrlbutlon
In
a
Cellular
Radlo System,"
IEEE
Trans.
on
Vehicular Tech
,
vol
34. Aug.
[9]
5
Teklnay
and
B.
Jabbar,.
"
A Measurement-Bared Queuelng
1987.
Scheme for
Protection
of Handovers in Moblle Cellular Sys~
terns." Tech, Rep, George Mason Unlv, 1991
Bibliography
lnternat~analConsultat~veComm~tteeforTelephaneandTelegraph(CCITT)
BlueBookRecs. Q.1001. Q.1002.Q.1003. 0.1004.and0.1005.
vol.
VI, fasclcle VI.12.
Electromcs Industry Asroclatlan (EIA)flIAPro].
No.
2078,
"Cellular
Radlotelecommunicatlons Intersystem Operatmr' Intersystem
Handoff." 15-41
2,
Feb 1991
European Telephone Standards lnstltute (ETSI)
TC
GSM. Rec. GSM
03.09, "Handover Procedures." Feb 1990
ETSlTCGSM.Rec
GSM08.06,"SignallingTrdnsportMechanlsmsbetween
the
BSS
and MSC Procedures," June 1990.
Biography
Bljan Jabbar, (Senlor Member, IEEE) recelved the
B
S
degree from
Arya-Mehr Unlverrlty. Tehran,
Iran,
In
1974, the
M.5.
and Ph.D
twely. all
~n
electrical
engmeerlng.
and
the
M.S.
degree
~n
engineering-
degrees from Stanford Unlverslty, Stanford.
CA.
In
1977 and 1981, rerpec-
economic systems from Stanlord Unwersity in 1979. From
1979
to
1981 he was wlth Hewlett Packard. After
graduation.
from
1981 to
1983, he
war
an
Assistant Professor
ai
Southern Illinois Unlverslty.
Carbondale,
IL.
From 198310 1985 hewas
withSatelllteBuslnessSystems
(now
MCI Telecommun~catlanr). McLean. VA. where he managed pro-
tecture of the
SBS
next
generation
communlcatlons system
In
1985.
gramsonsyrtemsrequirementrdef~nition,rystemspecif~cat~on,andarch~-
he became Dorector
at
MIA-COM Telecommunlcatlons
far
develop-
ment
of
Advanced Data Communlcatlons Networks.
In
1988. he
jolned the School of Informatton Technology and Englneering at
George Mason Univerrlty, Fairfax, VA, where he is currently associate
professor of Electrical and Computer Englneering
HIS
current research
actlvltles Include
architecture
and protocols and performance model-
ing of broadband
telecommunications.
Intelligent networks, control
and slgnalllng
for
flxed and moblle telecommunlcatlons networks. Hr
IS
a member of Eta Kappa Nu
and
the Assaclation for Computlng
Machinery.
51r1n
Teklnay (Student Member.
IEEE)
received the 8.S. and
M.S.
degrees
from
Bogazicl Unlversity, Istanbul. Turkey, in 1989 and 1991
respedively,~nelectrlcalenglneenng.
She~rcurrentlycompletingrequ~re-
andEnglneenng,GeorgeMasonUnlvers~ty.Fairfax,VA.
Herareaofresearch
ments for the
Ph.D
degree
at
the School
of
lnformatlon Technology
control. trafflc analyr~r, and performance evaluation.
Mr
Tekinay
IS
a
is
moblle
cellular
communlcatlon
networks wlth roncentratlon
on
call
member of Eta Kappa Nu and the
IEEE
Communications
Soclety
46
IEEE
Cornmumcations
h4agazine
Novrmber
1991