Upload
efosa-aigbe
View
220
Download
0
Embed Size (px)
Citation preview
7/23/2019 Exploiting Tal
1/46
m Si gy Isiui i uvsLiig Uivsiy Liigs uivsi
giN4706S,giN4706-ES
LiU-ITN-TEK-A--13/054--SE
Exploiting Tracking Area List
Concept in LTE NetworksMohsin Nawaz
2013-10-04
7/23/2019 Exploiting Tal
2/46
LiU-ITN-TEK-A--13/054--SE
Exploiting Tracking Area List
Concept in LTE NetworksExamensarbete utfrt i Elektroteknik
vid Tekniska hgskolan vid
Linkpings universitet
Mohsin Nawaz
Examinator Di Yuan
Norrkping 2013-10-04
7/23/2019 Exploiting Tal
3/46
Upphovsrtt
Detta dokument hlls tillgngligt p Internet eller dess framtida ersttare
under en lngre tid frn publiceringsdatum under frutsttning att inga extra-ordinra omstndigheter uppstr.
Tillgng till dokumentet innebr tillstnd fr var och en att lsa, ladda ner,skriva ut enstaka kopior fr enskilt bruk och att anvnda det ofrndrat frickekommersiell forskning och fr undervisning. verfring av upphovsrttenvid en senare tidpunkt kan inte upphva detta tillstnd. All annan anvndning avdokumentet krver upphovsmannens medgivande. Fr att garantera ktheten,skerheten och tillgngligheten finns det lsningar av teknisk och administrativart.
Upphovsmannens ideella rtt innefattar rtt att bli nmnd som upphovsman iden omfattning som god sed krver vid anvndning av dokumentet p ovanbeskrivna stt samt skydd mot att dokumentet ndras eller presenteras i sdanform eller i sdant sammanhang som r krnkande fr upphovsmannens litterraeller konstnrliga anseende eller egenart.
Fr ytterligare information om Linkping University Electronic Press sefrlagets hemsida http://www.ep.liu.se/
Copyright
The publishers will keep this document online on the Internet - or its possiblereplacement - for a considerable time from the date of publication barringexceptional circumstances.
The online availability of the document implies a permanent permission foranyone to read, to download, to print out single copies for your own use and touse it unchanged for any non-commercial research and educational purpose.Subsequent transfers of copyright cannot revoke this permission. All other usesof the document are conditional on the consent of the copyright owner. The
publisher has taken technical and administrative measures to assure authenticity,security and accessibility.
According to intellectual property law the author has the right to bementioned when his/her work is accessed as described above and to be protectedagainst infringement.
For additional information about the Linkping University Electronic Pressand its procedures for publication and for assurance of document integrity,please refer to its WWW home page: http://www.ep.liu.se/
Mohsin Nawaz
7/23/2019 Exploiting Tal
4/46
Exploiting Tracking Area List ConceptIn Long Term Evolution Networks
Mohsin Nawaz
Wireless Networks and ElectronicsDepartment of Science and Technology(ITN)Linkping ni!ersity" Sweden
7/23/2019 Exploiting Tal
5/46
Ab tract
Signaling o!erhead d#e to tracing and paging ser E$#ipments (Es) has always %een a
concern for the network operators& Tracking 'rea (T') is a logical gro#ping of cells in Long
Term E!ol#tion (LTE) networks& T's manage and locate Es in a LTE network& ' well known
performance consideration is that of Tracking 'rea pdate (T') o!erhead !ers#s paging
o!erhead&
LTE gi!es more flei%ility in terms of T' config#ration compared to pre!io#s generations& This
flei%ility is offered %y means of Tracking 'rea List (T'L)& T'L is epected to o!ercome certain
limitations of the con!entional T' scheme& ow to eploit this flei%ility to ha!e a T' design
that red#ces signaling o!erhead remains an open iss#e& This thesis eploits the T'L concept in
LTE networks in order to e!al#ate its performance& 'n approach to design T'L #sing E traces
is presented& The e!al#ation of its performance is done %y comparing it with the con!entional
T' and the r#le of th#m% T'L designs& The cost of collecting E traces #sing T'L is taken into
consideration when e!al#ating performance of E*trace %ased T'L design&
7/23/2019 Exploiting Tal
6/46
Acknowled ent
+irst of all" I wo#ld like to thank 'lmighty 'llah for gi!ing me strength to accomplish this work&
I am also !ery gratef#l to my family who s#pported me d#ring this phase&
I am especially !ery pleased to show my gratit#de to my thesis s#per!isor Sara ,odarres
-a.a!i& I am !ery gratef#l for the effort she p#t thro#gho#t and witho#t her pro!ision it wo#ld
not ha!e %een accomplished&
I wo#ld also like to thank professors at ITN department at Linkping ni!ersity for the
technical s#pport they pro!ided thro#gho#t my ,asters/ degree which helped me d#ring my
thesis work&
7/23/2019 Exploiting Tal
7/46
Techncal Defnton
The following terms are #sed thro#gho#t this thesis report&
1stGeneration (1G)
In early 0123/s S la#nched the 0 stgeneration of cell#lar network system called 'd!ance ,o%ile 4honeSystem (',4S)& It was an analog#e system %ased on fre$#ency mod#lation technology& It wasdesigned for only !oice applications&
2nGeneration (2G)
In 0110" the 5ndgeneration of cell#lar network systems %ased on circ#it switching was la#nched and
called 6lo%al System for ,o%ile 7omm#nication (6S,)& It was the first digital technology %asedsystem which also pro!ided data ser!ice along with !oice& It also laid fo#ndation to f#rther coming5&86 (64-S)" 5&986 (ED6E) and :6 networks&
!rGeneration (!G)
In 5330" the :rdgeneration of cell#lar network systems %ased on packet switching was la#nched& Thereare two standards that are termed as :6& These incl#de ni!ersal ,o%ile Telecomm#nications System(,TS) and 7D,'5333& :6 networks impro!e on packet switching %ased ser!ices& :6 networkspro!ide higher data rates compared to pre!io#s generations&
Long Term Evolution (LTE)
LTE is a candidate for ;6& LTE like :6 is %ased on /packet switching/& The difference is that LTE network
is an all I4 %ased network& LTE offers greater #p*link and down*link data rates& LTE also offers flei%ilityin network design incl#ding location management&
LTE Ile "oe
When a E is switched on %#t the network doesn/t know where eactly the E is" it is said to %e in idlemode& owe!er" the network in this case knows in which T' the E is located %#t not eactly whichcell&
LTE Active "oe
When E is connected !ia any channel like d#ring a call" the network knows in which cell the E islocated and E can send or recei!e data at any time& In this sit#ation the E is said to %e in acti!emode&
LTE #etac$ "oe
When E is powered off" it is in the detached mode& owe!er" e!en the detached mode" it is importantthat the network knows that the E is powered off&
%anover
When the recei!ed signal strength is lower than a certain le!el from the ser!ing cell with the help ofnetwork" E is shifted to another neigh%oring cell with highest signal le!el witho#t terminating the call&ando!er occ#rs when a E is shifted from one comm#nicating channel to another channel d#ring acall or data transfer thro#gh the core network& This happens to a!oid call termination when the cellgets o#t from the ser!ing range of one cell and enters the ser!ing range of another cell&
7/23/2019 Exploiting Tal
8/46
%ar %anoverard ando!er means that %efore the connection is made with other cell" the first connection is lost&
7/23/2019 Exploiting Tal
9/46
of F r
Figure 3.1: Symmetry limitation..............................................................................................................18
Figure 3.2: Example of TAL allocation (TA sie ! 1 cell".......................................................................1#
Figure $.1: %E trace collection................................................................................................................22
Figure $.2: Example of a ro& of %E'trace scenario matrix.....................................................................2$
Figure $.3: Start cell selection..................................................................................................................2$
Figure $.$: ext cell selection.................................................................................................................2)
Figure ).1: *ayleig+ ,istri-ution function ( ! 1"...................................................................................28
Figure ).2: /oisson ,istri-ution function ( 0 ! )"....................................................................................2#
Figure ).3: omparison of cell loa, ,istri-ution for %E'trace scenario &it+ original ,ata..................31
Figure ).$: omparison of cell +an,oer ,istri-ution for %E'trace scenario &it+ original ,ata..........32
Figure ).): Simulation setup (-loc4 ,iagram".........................................................................................3$
7/23/2019 Exploiting Tal
10/46
nd of Tabl
Ta-le 1: Signaling oer+ea, for t+e conentional TA ,esign..................................................................35
Ta-le 2: Signaling oer+ea, for %E'trace -ase, TAL ,esign using 12)6 ()7" %E traces......................3
Ta-le 3: TA% oer+ea, of t+e %E'trace -ase, TAL ,esign.....................................................................3
Ta-le $: /aging oer+ea, for t+e %E'trace -ase, TAL ,esign................................................................38
Ta-le ): ost of ,ata collection (%E'trace scenario ".............................................................................3#
Ta-le 5: Total oer+ea, comparison for t+e t+ree TAL ,esigns...............................................................3#
7/23/2019 Exploiting Tal
11/46
Tabl of Content
+apter 1 ntro,uction............................................................................................................................16
1.1 Scope............................................................................................................................................16
1.2 9utline..........................................................................................................................................11
+apter 2 Trac4ing Area.........................................................................................................................12
2.1 Trac4ing Area...............................................................................................................................12
2.2 Location anagement..................................................................................................................12
2.3 Location Area %p,ate Sc+emes...................................................................................................12
2.$ ommon Static Location Area %p,ate Sc+emes.........................................................................13
2.$.1 Al&ays %p,ate.....................................................................................................................13
2.$.2 eer %p,ate.......................................................................................................................13
2.$.3 *eporting ells....................................................................................................................13
2.$.$ Forming LA.........................................................................................................................13
2.) ommon ;ynamic Location Area %p,ate Sc+emes....................................................................1$
2.).1 Time'-ase, %p,ate..............................................................................................................1$
2.).2 oement'-ase, %p,ate.....................................................................................................1$
2.).3 ;istance'-ase, %p,ate........................................................................................................1$
2.5 /aging Sc+emes............................................................................................................................1$
2.5.1 Simultaneous /aging...........................................................................................................1$
2.5.2 Seloc4 ;iagram of Simulation Setup............................................................................................3$
).5 umerical *esults........................................................................................................................3)
). onentional TA ..........................................................................................................................35
).8 %E'Trace -ase, TAL ,esign.........................................................................................................3
7/23/2019 Exploiting Tal
12/46
).# ost of ;ata ollection................................................................................................................38
).16 *ule of T+um- TAL...................................................................................................................3#
).11 onclusion..................................................................................................................................$6
+apter 5 onclusion..............................................................................................................................$1
7/23/2019 Exploiting Tal
13/46
Chapter 1 Introduction
Chapter 1 ntrod ct on
There has %een a tremendo#s growth in cell#lar comm#nications especially mo%ile %road%and
networks& Long Term E!ol#tion (LTE) is a candidate for ;6 that aims to impro!e mo%ile
comm#nications with an emphasis on mo%ile %road%and& LTE offers new feat#res that impro!e
network efficiency and performance&
,o%ility ,anagement (,,) is one of the key f#nctions in mo%ile networks& It pro!ides means
to locate ser E$#ipments (Es) and deli!er ser!ices (s#ch as calls" S,S etc&) to Es in mo%ile
networks& There are two aspects of mo%ility management" location management (also called
reacha%ility) and hando!er management (also called session contin#ity)& Location
management keeps track of Es in mo%ile networks so that when needed mo%ile phone
ser!ices can %e deli!ered to Es& ando!er management ena%les session contin#ation when
Es change cells (access points)& This thesis foc#ses on location management aspect of ,, in
LTE networks&
>ne of the key tasks of location management is to keep track of Es in the network& Tracking
'rea (T') is a logical gro#ping of cells in a network and is #sed to track and locate (page) Es&
The T' concept is similar to Location 'rea (L') concept in the circ#it*switched (7S) domain
and -o#ting 'rea (-') concept in the packet*switched (4S) domain&
1.1 Scope
This thesis aims to address T' planning and optimi.ation concepts in LTE networks& >ne of the
main considerations in T' design is to red#ce the signaling o!erhead& Signaling o!erhead
consists of #pdate o!erhead and paging o!erhead& In the standard T' #pdate (T') and
paging scheme" a E registers in a T'& The T' is recorded in ,o%ility ,anagement Entity
(,,E)& Whene!er the E mo!es to a new T'" it needs to #pdate ,,E a%o#t the new T'& The
E performs a T' to #pdate ,,E& This #pdate contri%#tes to signaling o!erhead and is
termed as T' signaling o!erhead& When the E is called" the ,,E pages all the cells
%elonging to the last registered T' for the E& This paging also contri%#tes to signaling
o!erhead and is termed as paging signaling o!erhead&
0
7/23/2019 Exploiting Tal
14/46
1.1 Scope
T' design of a network needs to %e changed according to changing E mo%ility and
distri%#tion otherwise the T' design may ca#se more signaling o!erhead& T' design
optimi.ation often in!ol!es changing T's for cells& This change ca#ses ser!ice interr#ption
since the cells need to %e restarted for the change to take effect& To red#ce ser!ice
interr#ption the new T' design cannot differ a lot from the pre!io#s T' design&
Tracking 'rea List (T'L) is a scheme introd#ced in :644 -elease 2 ?0@& In this scheme" a cell can
%e part of m#ltiple T's& Each cell has a list of T's& The cell gi!es this list to Es& The list is kept
%y Es #ntil they enter a cell that does not %elong to any T' in the list& LTE standard allowscells to gi!e different T' lists to different Es& In ,,E" a E location is known to the
gran#larity of the T'L allocated to it&
Signaling o!erhead can %e !irt#ally eliminated if T'Ls are designed #sing E mo!ement and
call patterns (E traces)& owe!er o%taining this data is diffic#lt and costly& This thesis presents
an approach to design T'L #sing E traces& The approach is aimed to ser!e as a greenfield
design and not a re*optimi.ation&
1.2 Outline
C$apter 2 descri%es !ario#s location #pdate and paging schemes& The con!entional T'
scheme and the signaling o!erhead that is #sed thro#gho#t the thesis is also disc#ssed&
C$apter !descri%es the limitations of the con!entional T' scheme& T'L concept is introd#ced
and how it can mitigate some of the performance limitations of the con!entional T' scheme
are eplained
C$apter - descri%es how T'L can %e applied to large*scale networks and eplains the E*
trace %ased T'L design and r#le of th#m% T'L
C$apter .descri%es the design aspects of T'L schemes and present n#merical res#lts from
sim#lations
C$apter / draws concl#sion and prospects the f#t#re work
5
7/23/2019 Exploiting Tal
15/46
Chapter 2 Tracking Area
Chapter 2 Track n Area
This chapter pro!ides an introd#ction to %asic materials and %ackgro#nd for tracking area
planning (T'4)& Signaling o!erhead form#lation for the con!entional T' scheme" which is
considered thro#gho#t the thesis" is pro!ided&
2.1 Tracking Area
T' is a gro#p of cells (similar to L') in which a E can mo!e freely witho#t performing
Tracking 'rea pdate (T') to #pdate ,o%ility ,anagement Entity (,,E) a%o#t its location
in the network&
2.2 Location Management
Location ,anagement (L,) means keeping track of idle Es within a network& Location
management ena%les network to locate Es for deli!ering calls and maintaining connections
as the Es mo!e in the network& L, consists of two stagesA finding the location of E and call
deli!ery& In the first stage" network is di!ided into L's& Es send L' when they change L's?5@& 'n L' is a gro#p of cells which are paged in case of an incoming call& 7all deli!ery to a E
is achie!ed %y finding the last registered L' and paging all the cells in that L'&
2.3 Location Area Update Schemes
There are !ario#s L' schemes that aim at optimi.ing L' o!erhead& The schemes can %e
generally categori.ed into static and dynamic& In static L' schemes" the L's are fied and
Es perform L' when they change L's while in dynamic L' schemes" L's are %ased on
Es call and mo%ility patterns ?5@& Static L' scheme are easy to implement and are not
comp#tationally epensi!e since they do not rely on characteristics of Es& Dynamic L'
schemes are comp#tationally epensi!e and re$#ire storage of E data for comp#tation&
Dynamic L' schemes pro!ide more o!erhead red#ctions compared to static L' schemes
%#t re$#ire caref#l network design for efficient comp#tation ?;@&
:
7/23/2019 Exploiting Tal
16/46
2.4 Common Static Location Area pdate Schemes
2.4 Common Static Location Area Update Schemes
In static L, schemes Es either perform L' when they change cell or after specific time
inter!al& Some of the common static L, schemes areA
20-01 Alwas *pate
In always #pdate scheme" each cell is a reporting cell& ' reporting cell is a cell in which a E
#pon entering sends a L' to #pdate its location in the network& L' occ#rs whene!er a E
mo!es from one cell to another cell& E sends a L' on entering a new cell& This allows the
network to know which cell a E is c#rrently located& This red#ces the paging o!erhead to
.ero %eca#se network always knows the location of each E& owe!er" L' o!erhead is
maim#m ?B@&
20-02 Never *pate
In the ne!er*#pdate scheme" L' ne!er takes place since all the cells %elong to a single L'&
This means a E mo!es %etween cells witho#t e!er performing L'& When a E is called"
network pages all the cells in the network& This res#lts in maim#m paging o!erhead&owe!er" L' is .ero ?9@&
20-0! eporting Cells
In the reporting cells scheme" there are no L's& ' E #pdates network a%o#t its location when
it enters one of the reporting cells& To locate a E" a search is cond#cted in the !icinity of the
last reporting cell !isited %y the E& The reporting cells are predefined and efficient
arrangement re$#ires E characteristics&
20-0- orming LA
This L' scheme is a com%ination of always*#pdate and ne!er*#pdate schemes& 7ells are
gro#ped into L's& Es mo!ing within a L' do not perform L's& 'n L' occ#rs whene!er a
E mo!es from one L' to another& Whene!er there is a need to locate a E" all cells in the last
known L' of the E are paged& This com%ination of always*#pdate and ne!er*#pdate schemes
pro!ide a %alance %etween L' and paging o!erheads& This L' scheme is referred to as the
standard scheme and is the ass#med scheme in this thesis&
;
7/23/2019 Exploiting Tal
17/46
2.! Common "#namic Location Area pdate Schemes
2. Common !"namic Location Area Update Schemes
Dynamic L, offer m#ch more flei%ility The parameter of dynamic L, schemes can %e
modified to fit indi!id#al Es and scenarios& Some of the common dynamic L, schemes areA
20.01 Time34ase *pate
In time*%ased #pdate scheme" each E sends an L' after a certain period of time& This is the
simplest of the schemes since Es need to follow only their local clocks ?5@&
20.02 "ovement34ase *pate
In mo!ement*%ased #pdate scheme" a E sends an L' after it crosses %o#ndaries of a
certain n#m%er of cells (threshold)& This scheme is more comple than time*%ased %eca#se
Es need to %e aware when they cross cell %o#ndaries ?5@&
20.0! #istance34ase *pate
In distance*%ased #pdate scheme" a E sends an L' after it mo!es a certain distance since
the last L'& This scheme is the most comple since Es need to know the topology of the
network in order to calc#late distance mo!ed ?5@&
2.# $aging Schemes
4aging schemes determine how a called E will %e paged& Some of the common paging
schemes areA
20501 &imultaneous +aging
In sim#ltaneo#s paging" when a E is called all the cells in the last reported L' for the E are
sim#ltaneo#sly paged&
7/23/2019 Exploiting Tal
18/46
2.$ %aging Schemes
of the E are paged according to their n#m%er& This pro!ides a way to red#ce paging
o!erhead since all 4's may not %e paged& owe!er" since 4's are paged se$#entially" a delay
in connection occ#rs if the E is not located in the first 4'&
2.% Tracking Area !esign Optimi&ation
nder the con!entional T' scheme" the main o%=ecti!e is to ha!e a T' design that minimi.es
total signaling o!erhead& T' of !ery small si.e will res#lt in lower paging o!erhead %#t there
will %e ecessi!e T' o!erhead& >n the other hand" T's of !ery large si.e will red#ce or
!irt#ally eliminate T' o!erhead %#t will res#lt in ecessi!e paging o!erhead&
,#ch research has %een done on this trade*off& The research done on location registration
strategies ?0B@"?09@"?02@ and paging strategies ?09@"?01@"?53@"?50@ which s#ggest special
modifications in the network and additional E information that is costly to o%tain ?:@& This is
the reason that con!entional T' concept is commonly #sed ?:@&
The con!entional T' scheme for T' and paging is followed in the thesis& This means that a
E crossing a T' %o#ndary will lead to T' and all cells in the registered T' of the called E
are paged sim#ltaneo#sly&
2.' User ()uipment States in Mo*ilit" Management
' #ser e$#ipment (E) in LTE is any de!ice s#ch as smartphone" laptop etc& that an end*#ser
#ses to comm#nicate& E is almost the same concept as ,o%ile Station (,S) or ,o%ile
Terminal (,T) #sed in pre!io#s generation of cell#lar networks& +rom a mo%ility perspecti!e" a
E can %e in three statesA LTEC'7TIE" LTECIDLE and LTECDET'7ED& In theLTEC'7TIE state" a
E is connected to the network (i&e& has an esta%lished radio connection)& In this state" thenetworks know the cell the E is in& Therefore acti!e Es do not contri%#te to T' and paging
o!erhead since network is always aware of E location in this state& In LTECIDLE state" the
network is aware a%o#t the E location to the gran#larity of a T' (a gro#p of cells) since E is
in power*conser!ation mode and does not #pdate network a%o#t its location on each cell
change& Instead it informs the network when it changes T's& In LTECDET'7ED state" the E is
in a transitory state and is in the process of searching and registering with the network&
,ost Es are in LTECIDLE mode which means that the network is aware a%o#t their location at
B
7/23/2019 Exploiting Tal
19/46
2.& ser '(uipment States in Mo)i*it# Management
the gran#larity of a T'& 7ell load and cell hando!er data is #s#ally the only realistic data
a!aila%le from a cell#lar network& owe!er" this data is for Es in LTEC'7TIE state& If it is
ass#med that Es in LTECIDLE state ha!e the same mo%ility and call pattern as the Es in
LTEC'7TIE state" a good estimation of E mo!ement and call acti!ity can %e made& This is the
ass#mption #sed in the thesis&
' E trace is de ned as the cell*to*cell mo!ement and the call arri!al pattern of a E in afi
speci c time period& E traces can %e #sed to red#ce signaling o!erhead of the network sincefi
the network can %e optimi.ed to mo!ement and call patterns of Es&
2.+ Signaling O,erhead Calculation
Let the cells in network %e denoted %y Nce**s + ? 1 @ . . . @N and the T's in the network %e
denoted %y TT A ! ? 1 @ . . . @ T . The T' each cell %elongs to can %e represented %y
t + B t1, . . . , tNC where ti is the T' of cell i& T' design t can %e represented %y an N - N
symmetric and %inary matri S(t)& Each matri elements i (t"represents whether the two cells
are in the same T' or not&
si . (t)={1 i/ ti=t .6 otherwise (2.1"Let ui%e the n#m%er of Es in cell i for a gi!en time period and h i %e the n#m%er of Es
mo!ing from cell i to cell& The !al#es of ui and h i can %e determined from cell load and
hando!er data& The amo#nt of o!erhead of one T' is denoted %y cuand o!erhead of one
paging is denoted %y cp & 7all acti!ity factor i&e& pro%a%ility that a E needs to %e paged is
denoted %y 0The total o!erhead i&e& s#m of T' and paging o!erhead is calc#lated asA
cS1( t)=iN
. N
(cu hi . (1s i . (t))+ cp
uis i .( t)) (2.2"
In (5&5)" the first term within the parenthesis is the T' o!erhead for Es mo!ing from cell ito
cellif the cells are not in the same T' and the second is the paging o!erhead for paging Es
in cell iif the two cells are in the same T' ?58@&
9
7/23/2019 Exploiting Tal
20/46
Chapter Tracking Area List
Chapter Tracking Area List
This chapter pro!ides an introd#ction to Tracking 'rea List (T'L)" a T' scheme introd#ced in
:644 -elease 2 ?:@& This scheme allows for more flei%le T' design& It is epected that this
scheme will remo!e certain limitations of the con!entional T' scheme& 7on!entional T' which
is the most #sed T' scheme in networks today has performance limitations& This chapter
considers limitations of the con!entional T'& The idea of T'L and how it allows for a more
flei%le T' design are eplained&
3.1 Limitations o- Con,entional TA
In a con!entional T'" cells are di!ided into m#t#ally ecl#si!e sets and each set is a T'& ' E
performs a T' whene!er it changes T'& There are some limitations of con!entional T'
schemeA
!0101 +ing +ong E''ect
In con!entional T' scheme" a E performs T' whene!er it mo!es to a cell %elonging to anew T'& Whene!er a E mo!es %ack and forth %etween two cells %elonging to different T's"
there will %e ecessi!e T's& This phenomenon is referred to as the ping*pong effect&
!0102 "ass "o4ilit &ignaling Congestion
Signaling congestion may occ#r when a large of n#m%er of Es %eha!e in a similar manner"
e&g& massi!e and sim#ltaneo#s E mo!ement in train scenario& Signaling congestion is ca#sed
%y ecessi!e T's from Es in a short time inter!al&
!010! &mmetr Limitation
nder the con!entional T' scheme" if cell ' and cell < are in the same T' then neither of them
can %e in any other T'& If this limitation can %e remo!ed" signaling o!erhead may %e red#ced&
2
7/23/2019 Exploiting Tal
21/46
.1 Limitations o/ Con3entiona* TA
Example6 7onsider the two cells ' and < in fig#re :&0& nder the con!entional T' scheme"
either the cells are in the same T' or not& +rom signaling o!erhead form#la in (5&5)" total
signaling o!erhead c0when %oth cells ' and < are in the same T' is gi!en in (:&0)
c1= cp (uA+u4) (3.1"
'nd when cell ' and < are in different T's" the signaling o!errhead c2is gi!en in (:&5)
c2=cu
(hA+hA) (3.2"
Now ass#me that if cell ' and cell < can ha!e different perspecti!es to one another e&g& cell '
considers < in a different T' %#t cell < considers cell ' in the same T' then the signaling
o!erhead c is gi!en in (:&:)
c3=cu
hA4+ cp
u4 (3.3"
+or c1>c3 and c2>c3 to %e tr#e" cp uA>cu hA4 and cu h4A>cp u4 m#st %e tr#e i&e&
when cell ' has high load and there are few Es mo!ing from ' to < and cell < has low load
and more Es mo!e from < to '& In s#ch a scenario" remo!ing symmetry limitation will lead to
a lower signaling o!erhead&
3.2 Tracking Area List
T'L concept allows for more flei%le T' design and management& In the T'L scheme #sed in
the thesis" each T' consists of one cell& This implies that the T's are non*o!erlapping& Each cell
1
5igure .16 S#mmetr# *imitation
7/23/2019 Exploiting Tal
22/46
.2 Tracking Area List
has a list of T's which is assigned to E& The E does not perform T' as long as it is in a cell
%elonging to one of the T's from the list& T' is performed when a E mo!es to a cell that is
not incl#ded in any of the T's in the c#rrent list& The E then recei!es a new T'L& The location
of E is known at the gran#larity of the T'L allocated to the E&
In fig#re :&5 a T' has only one cell& T'0 has cell 0" T'0B; has cell 0B; and so on& S#ppose cell
0 is the starting cell of E& E gets T'L of cell 0& E now mo!es from cell 0 to 0B;& Since T'
0B; is in T'L of cell 0" no T' is performed& Same holds when E mo!es to cell 0: and 030&
When E mo!es from cell 032 to cell 9" T' takes place since T'9 is not in T'L of cell 0& E
now holds T'L of cell 9& 7ell 59 and cell B3 are in T'L of cell 9 so no T' takes place& When E
enters cell 08" T' occ#rs and E gets T'L of cell 08& Thro#gho#t the mo!ement network
always knows the location of E to the gran#larity of a T'L&
T'L scheme is epected to remo!e some limitations of the con!entional T' scheme& >ne of
03
5igure .26 '7amp*e o/ TAL a**ocation 8TA size + 1 ce**9
7/23/2019 Exploiting Tal
23/46
.2 Tracking Area List
the known limitations of the con!entional T' scheme is the ping*pong effect& T'L can a!oid
ping*pong T' o!erhead %y adding two ad=acent T's in the same T'L hence eliminating T'&
'nother limitation of the con!entional T' scheme is the mass mo%ility scenario e&g& Es in a
train crossing o!er to a new T'& T'L has the a%ility to assign E specific T'Ls& ence the cells
along the train track can assign different T'Ls to Es& The E in the train will ha!e different
T'Ls and hence will not perform T' at the same time mitigating mass T's&
00
7/23/2019 Exploiting Tal
24/46
Chapter 4 Imp*ementing TAL Scheme in Ce**u*ar Networks
Chapter ! ple ent n TA "che e n Cell lar #etwork
's disc#ssed T'L is epected to remo!e some limitations of the con!entional T' scheme& The
main challenge is how to apply T'L in large*scale networks& Tailoring T'L according to E
mo!ement and all pattern is the key to ha!ing an optimal T'L config#ration that will red#ce
signaling o!erhead& '!aila%ility of indi!id#al E mo!ement and call patterns wo#ld facilitate
generating an optim#m T'L that co#ld essentially eliminate signaling o!erheadA Each E
wo#ld get a specific T'L containing all cells it is epected to tra!erse thro#gh #ntil it is paged&Indi!id#al E mo!ement and call patterns is costly to o%tain& 'lso the !alidity of the
information is limited %eca#se E traces is the history of Es mo!ement and the f#t#re
mo!ement and call patterns for Es might change&
4.1 U(trace *ased TAL
6etting traces for idle Es was a costly and infeasi%le eercise in pre!io#s cell#lar
technologies& LTE howe!er offers some possi%ilities for collecting a small part of E traces&
>ne possi%ility is the #se of applications s#pporting 64S to collect E mo!ement and call
data& 'nother possi%ility is to ha!e a T'L config#ration that will facilitate collecting E trace
data&
T'L scheme in LTE offers the flei%ility of gi!ing E specific T'L and this can %e #sed to collect
a percentage of E traces& If the network gi!es a list to a E that only contains the #pdating
cell" then it is possi%le to trace that specific E in the network& 7onsider the scenario in +ig#re
;&0&
7/23/2019 Exploiting Tal
25/46
4.1 ':trace )ased TAL
The amo#nt of T' performed %y the traced Es is the cost of collecting part of the E traces&
There is no paging o!erhead associated with traced Es since their location is always known&
4.2 U(trace scenario
The E mo!ement directly affects the signaling o!erhead ?05@& In other words" signalingo!erhead is dependent on the E traces& In order to compare the T'L scheme and the
con!entional T' scheme" %oth m#st %e applied to the same E*trace scenario& ' E*trace
scenario is a set of E mo!ement and call pattern o!er a gi!en time that matches a gi!en set
of cell load and hando!er data& There can %e n#mero#s E*trace scenarios for one set of cell
load and hando!er data& E*trace scenario is generated #sing a mo%ility model& E*trace
scenario is dependent on the mo%ility model" the amo#nt and speed of each E/s mo!ement&
'ss#me that cell hando!er" cell load and call intensity factor 0data are a!aila%le for a network&
0:
5igure 4.16 ' trace co**ection
7/23/2019 Exploiting Tal
26/46
4.2 ':trace scenario
The generation of E*trace scenario start with determining the starting cell for each E& 7ells
with high load are more pro%a%le to %e the starting cells for Es& The pro%a%ility that a cell iis
the starting cell is calc#lated as in (;&0)&
%start( i)=ui
k=1
N
uk (;&0)
where"%start pro%a%ility that cell iis the starting cell
ui cell load of cell iN total n#m%er of cells
The net cell for a E is dependent on the last !isited cell of a E& The E will mo!e to
neigh%ors of last cell with high n#m%er of hando!ers& The pro%a%ility that cell is the net cell
for the E" gi!en that the last cell was i" is calc#lated as in (;&5)&
%ne7t(i , .)= hi .
k=1
N
hik (;&5)
where"%ne7t8(i , ) pro%a%ility that the net cell is when the c#rrent cell is i
hi cell hando!ers from cell ito cell
N total n#m%er of cells
the E*trace scenario is defined as a matri& The n#m%er of rows is e$#al to the n#m%er of
Es & The E*trace scenario matri is for a gi!en time d#ration T& Time d#ration T is di!ided
into e$#al time inter!als t& The ser!ing cell of E u in the time inter!al t is represented %yelement at (u,t"of the scenario matri& The d#ration of each inter!al is Dt. The dimension of
the scenario matri is T
t. The first col#mn of the matri is calc#lated #sing (;&0) and the
remaining col#mns are calc#lated #sing (;&5)&
7/23/2019 Exploiting Tal
27/46
4.2 ':trace scenario
+ig#re ;&5 ill#strates how a row of a E*trace scenario matri looks like& The time inter!alTis
di!ided into 03 inter!als 8t1,...,t1;99 with Dt d#ration each& The E remain in cell i for one Dt"
then mo!es to celland stays there for :Dt & +inally the E mo!es to cell kand stay there for
the rest of the time&
The E*trace scenario matrices are pop#lated #sing %startand%ne7tas in (;&0) and (;&5)& Start cell
for each E is selected #sing%startwhile the s#%se$#ent cell are selected #sing%ne7t& +ig#re ;&:
and +ig#re ;&; eplains start cell and net cell for each E& ' random n#m%er is generated
%etween B6@1C& +rom the +ig#re ;&:"%startfor cellis closest to the generated random n#m%er&Th#s" cellis selected as the starting cell& >nce the starting cell is selected" the net cell is
determined #sing %ne7t& ' random n#m%er is generated %etween B6@1C& The neigh%or cell of
ha!ing%ne7tclosest to the generated random n#m%er is selected& +rom +ig#re ;&; cell )has
pro%a%ility closest to the random n#m%er& Th#s" cell )is selected as the net cell&
08
5igure 4.26 '7amp*e o/ a row o/ ':trace scenario matri7
5igure 4.6 Start ce** se*ection
7/23/2019 Exploiting Tal
28/46
4.2 ':trace scenario
4.3 Aggregated !ata -rom U(trace Scenario
E*trace scenario matri is generated #sing start cell and net cell pro%a%ilities which are
%ased on cell load and cell hando!er data respecti!ely& The cell load and hando!er from the
E*trace scenario needs to %e aggregated to match the original data&
The aggregated cell load uiof a cell iis the ratio of total time spent %y Es in cell ito the time
d#ration of the E*trace scenario matri& Similarly the aggregated cell hando!ers from cell ito
k is the n#m%er of times Es mo!e from cell i to cell kin the E*trace scenario matri& In
+ig#re ;&5 if it is s#pposed that Dt 0 and T 03 then the aggregated cell load data is ui
3&5" u 3&: and uk 3&8& The hando!er data is h i 0 and h k 0&
4.4 /ule o- Thum* TAL
-#le of th#m% T'L design is proposed in ?:@& It offers a way of designing T'L witho#t the need
of E traces& -#le of th#m% is #sed for performance e!al#ation of E*trace %ased T'L design&
To eplain r#le of th#m% T'L design" consider a network with only two cells iand& The T'L of
cell ican either %e FiG or ?i, & nder the ass#mption that T's can only ha!e one cell& +or the
first case where TAL(i" ! ?i@all E performing hando!er from to iwill perform an #pdate&
The res#lting #pdate o!erhead will %e cuh i & There will %e no paging o!erhead since there is
only one T' in T'L& +or the second case where TAL(i" ! ?i, " there will %e no #pdate o!erhead
for Es mo!ing from cellto ias %oth cells are in the T'L of Es" howe!er there will %e paging
o!erhead since the network needs to search for called E in cells iand& The res#lting paging
o!erhead is gi!en as 0cpui& To minimi.e o!erhead res#lting from cell i" a T'L design is chosen
that gi!es a lower o!erhead& In other words" if cu
h i H 0cp
uithen TAL(i" ! ?i ,and TAL(i" ! ?i
0B
5igure 4.46 Ne7t ce** se*ection
7/23/2019 Exploiting Tal
29/46
4.4 cp
u i6 otherwise
($.3"
+or a large*scale network" r#le of th#m% T'L design pro!ides a local minim#m and not the
glo%al minim#m since each cell is optimi.ing T'L with respect to its own perspecti!e #sing its
data witho#t taking into acco#nt the effect of other cells on their modified T'L&
09
7/23/2019 Exploiting Tal
30/46
Chapter ! "esign and Simu*ation
7/23/2019 Exploiting Tal
31/46
!.1 Introduction
hi n#m%er of cell hando!er from cell
ito
uk cell load of cell kN total no& of cells
The a!erage n#m%er of mo!es 8>mo3es) from the original data is 0 mo!e per E& This is #sed as
the mean of the -ayleigh f#nction& +ig#re 8&0 shows the -ayleigh distri%#tion f#nction #sed to
generate E tracesA
7all d#ration for Es is determined #sing 4oisson distri%#tion with? 8 (a!erage calld#ration)& The distri%#tion is shown in +ig#re 8&5&
01
5igure !.16 + 19
0 1 0 2 0 3 0 4 0 5 0 6 00
0 . 1
0 . 2
0 . 3
0 . 4
0 . 5
0 . 6
0 . 7
0 . 8
X : 1
Y : 0 . 7 1 5 4
N u m b e r o f m o v e s
P
ro
b
a
b
ility
moves
7/23/2019 Exploiting Tal
32/46
!.1 Introduction
'lgorithm I eplains the algorithm #sed for generating a E*trace scenario matri& ' random&
M is the E*trace scenario matri&>mo3esis the a!erage n#m%er of mo!es calc#lated as in (8&0)&
Algorit$m I 7 Generating *E Traces
fori = 1:N calculate Pstart(i) using (4.1)
end
fori = 1:N
forj = 1:N calculate Pnext(i) using (4.2)
end
end
forx = 1:USERS
generate random numer et!een " and 1# dra! cell $rom Pstartusing random numer# add dra!n cell to ue%trace scenario matrix & x1#
end
trace'lengt = "#
forx = 1:USERS dra! num'mo*es using Ra+leig'random(,mo*es)#
whilenum'mo*es - "
forj = 2:trace'lengt
re*ious cell = &x(j%1)# ifando*er $rom re*ious cell to cells - "
dra! cell $rom Pnext(i/j) using random numer# add dra!n cell to ue%trace scenario matrix &x i#
num'mo*es = num'mo*es 0 1# else
53
5igure !.26 %oisson distri)ution /unction 8 ? + !9
7/23/2019 Exploiting Tal
33/46
!.1 Introduction
&xj = &x(j%1)#
end
end
end
end
50
7/23/2019 Exploiting Tal
34/46
!.2 Mo)i*it# Mode* @eri/ication
.2 Mo*ilit" Model eri-ication
The aggregated cell load and hando!er data from E*trace scenario matrices can %e #sed to
!erify against original data& 'ggregated cell load of cell iis defined as the n#m%er of times
cell iappears in E*trace scenario matri di!ided %y the total n#m%er of time inter!als T
t .
+ig#re 8&; compares aggregated cell load gi!en %y (8&5) from one of the E*trace scenario to
original cell load data& +ig#re 8&; compares aggregated cell hando!er from E*trace scenario I
to original cell hando!er data&
u i =ni t
T ().2"
where"
ni n#m%er of times cell iappears in E traces
Dt time d#ration of inter!als
T total time d#ration for E*traces
55
5igure !.6 Comparison o/ ce** *oad distri)ution /or ':trace scenario I with origina* data
7/23/2019 Exploiting Tal
35/46
!.2 Mo)i*it# Mode* @eri/ication
The res#lts show that altho#gh there are some !isi%le differences" aggregated data to some
etent resem%les the original data&
.3 Con,entional Tracking Area !esign
+or con!entional T' design" aggregated cell load and hando!er from the fi!e E*trace
scenario matrices is #sed& The data is #sed as inp#t to 74LEJ ?00@ to generate optim#m
con!entional T' designs for each scenario& The model #sed %y 74LEJ is defined in ?02@&
.4 U(Trace *ased TAL !esign
To generate E*trace %ased T'L design" the optim#m con!entional T' design is #sed as the
initial T'L& Since the optim#m con!entional T' design gi!es the lowest possi%le signalingo!erhead for con!entional case" it is a good starting point& To eplain how initial T'L is formed
from con!entional T' design consider a TA1s#ch that TA1 ! ?a,)where aand )are cells&
The T'L of cell a and ) %ecomes ?a,)& The T'L is optimi.ed %y addingKremo!ing cells
repeatedly and calc#lating signaling o!erhead after each change& If the change in T'L res#lts
in a lower o!erhead than con!entional case (which ser!es as the starting reference)" the
change is kept otherwise it is discarded& The new lower o!erhead is set as the new reference&
The process is repeated for T'L of each cell and a final T'L for each cell is o%tained&
5:
5igure !.46 Comparison o/ ce** hando3er distri)ution /or ':trace scenario I with origina* data
7/23/2019 Exploiting Tal
36/46
!.4 ':Trace )ased TAL "esign
Algorit$m II 7 TAL esign 4ase on *E3traces
sa*e x ue'traces as art'ue'traces#calculate total o*eread con*'3 $or con*entional 5 design#
sa*e con*'3 as re$erence o*eread re$'3sa*e con*entional 5 design con*'5 as otimum 56 ot'56#
fori = 1:N
forj = 1:N
ifj is not in 56 o$ i add j to 56 o$ i#
sa*e modi$ied 56 as ne!'56#calculate o*eread ne!'3 using ne!'56 $rom art'ue'traces#
ifne!'3 7 re$'3sa*e ne!'3 as re$'3#
sa*e ne!'56 as ot'56#
else
remo*e j $rom 56 o$ i#
end
else
remo*e j $rom 56 o$ i#
sa*e modi$ied 56 as ne!'56#calculate o*eread ne!'3 using ne!'56 $rom art'ue'traces#
ifne!'3 7 re$'3sa*e ne!'3 as re$'3#
sa*e ne!'56 as ot'56#else
add j to 56 o$ i#
end
end end
end
The optimi.ation of E*trace %ased T'L design in!ol!es applying changed T'Ls to the E*
trace scenarios and determine total o!erhead& The T'L is optimi.ed #sing mo%ility and call
pattern for different percentage (7) of total E traces (sample E traces)& The reason forconsidering part of the E traces (sample E traces) is that data needed to generate E traces
has an associated cost as eplained in Section ;&0& Therefore m#ltiple T'L config#rations
optimi.ed #sing different n#m%er of sample Es are generated to see how T'L performance
!aries for different n#m%er of sample Es&
5;
7/23/2019 Exploiting Tal
37/46
!.! *ock "iagram o/ Simu*ation Setup
. lock !iagram o- Simulation Setup
The %lock diagram of the sim#lation set#p is shown %elowA
+ollowing is a description of the different %locks in the %lock diagramA
0& The original data is from real*life network of Lis%on consisting of cell load and cell
hando!er
58
5igure !.!6 Simu*ation setup 8)*ock diagram9
7/23/2019 Exploiting Tal
38/46
!.! *ock "iagram o/ Simu*ation Setup
5& 'lgorithm I is #sed to generate a E traces scenario from the original data
:& 'ggregated cell load and cell hando!er are calc#lated from E*trace scenario matri
;& sing aggregated data and E*trace scenario the con!entional T' design is optimi.ed
#sing 74LEJ ?00@
8& The con!entional T' design is applied to the E traces and the signaling o!erhead is
calc#lated
B& Different n#m%er of E traces (sample E traces) are selected
9& E*trace %ased T'L design is generated for different n#m%er of sample E traces
2& Signaling o!erhead is calc#lated %y applying E*trace %ased T'L designs to E traces
1& 7ost of data collection is calc#lated for different n#m%er of sample E traces
03& sing E*trace scenario" signaling o!erhead is calc#lated for E*trace %ased T'Ls
00& -#le of Th#m% T'L design is generated as descri%ed in Section ;&;
.# umerical /esults
E*trace %ased T'L design algorithm and the r#le of th#m% T'L algorithm are implemented in
,'TL'< -5331% ?0:@& The performance of E*trace %ased T'L design is compared with %oth
the con!entional T' and the r#le of th#m% T'L designs& The performance e!al#ation is done
%y applying the T'L designs to 8 different E*trace scenarios& The signaling o!erhead is
calc#lated %y #sing E traces for the 8 scenarios& Each time a E makes a hando!er to a cell
that is not in its T'L" cuis added to the o!erall o!erhead& When a E is paged" a cost e$#al to
(L t a *:1)cp " whereL t a *is the length of E T'L" is added to the o!erall signaling o!erhead& '
called E does not contri%#te to T' d#ring the call d#ration&
5B
7/23/2019 Exploiting Tal
39/46
!.B Con3entiona* TA
.% Con,entional TA
The performance e!al#ation of con!entional T' design is done %y applying the design to E*
trace scenarios& Ta%le 0 gi!es the signaling o!erhead for the 8 E*trace scenarios& The res#lts
show that the o!erall signaling o!erheads for the 8 E*trace scenarios are !ery close& The
reason is that the E*trace scenarios are all %ased on the same original data (cell load and
hando!er data)&
%E Trace Scenario TA% /aging Total
1 ##6.6 23).2 122).2
2 #52.6 23#.5 1261.5
88).6 25$.5 11$#.5
4 #36.6 2)3.$ 1183.$
! 8#2.6 2$.6 1155.6
Ta)*e 16 Signa*ing o3erhead /or the con3entiona* TA design
59
7/23/2019 Exploiting Tal
40/46
!.& ':Trace )ased TAL design
.' U(Trace *ased TAL design
E*trace %ased T'L config#rations for different E*trace scenarios are generated& T'L
config#rations are generated for different n#m%er of E traces& The different n#m%er of E
traces #sed were 083" :33" 983" 0333" 0583" 0983" 5333 and 5833& 'll the res#lts are not
presented& Ta%le : presents the res#lts are for a case with 0583 E traces and compares them
to the corresponding con!entional T' config#ration&
%E'Trace ScenarioCon3entiona* TA TAL
TA% /aging TA% /aging
I )6 #@) 1 16@5
II 3$ )@# 1 5@6
III 3# 12@1 2 13@1
I@ 3$ 11@) 2 13@1
@ $8 )@) $ 5@$
Ta)*e 26 Signa*ing o3erhead /or ':trace )ased TAL design using 12!; 8!9 ' traces
+rom Ta%le 8 it is seen that E*trace %ased T'L design offers considera%le red#ction in T'
o!erhead compared to the con!entional T' config#ration for the collected E traces& There is
howe!er a slight increase in paging o!erhead& To e!al#ate performance for large*scale
networks" E*trace %ased T'L config#rations for different percentage of E traces are applied
to all E traces for the 8 E*trace scenarios& Ta%le : presents T' o!erhead res#lts and
compares them to the con!entional T' config#ration&
%E Trace Scenario
um-er of sample %E traces
onentional TA;esign1)6 366 )6 1666 12)6 1)6 2666 2)66
I 826 53 )1 )3# )6$ $)1 $16 358 ##6@6
II 8$) 13 58$ 536 )$# )61 $$) #52@6
III #5 5 )## )8$ )$6 $# $8# 3## 88)@6
I@ 88$ 8)$ 2$ 55$ )8) )63 $85 $$ #36@6
@ 862 2 5$2 )#5 ))3 $#5 $) $1# 8#2@6
Ta)*e 6 TA o3erhead o/ the ':trace )ased TAL design
52
7/23/2019 Exploiting Tal
41/46
!.& ':Trace )ased TAL design
+rom Ta%le : it is seen that T' cost is red#ced as the n#m%er of sample E traces increase
i&e& when data from more E traces is #sed in impro!ing the E*trace %ased T'L design& This is
!alid for all the 8 E*trace scenarios& E!en #sing data from 083 E traces" the E*trace %ased
T'L design yield a lower T' o!erhead compared to con!entional T' design for all scenarios&
The paging cost for E*trace %ased T'L design and con!entional T' design for 8 different E
trace scenarios is calc#lated and presented in Ta%le ;&
%E Trace Scenarioum-er of sample %E traces onentional TA
;esign1)6 366 )6 1666 12)6 1)6 2666 2)66
I 238@2 23@# 2$)@ 2$$@) 2$5@3 2$1 23#@3 2$2@1 22)@
II 2$3@# 2$) 2$$@3 2$@5 2)1@8 2)6@# 2))@2 2)2@8 233@
III 258@3 26@ 22 26@ 23@5 2 21@ 283@2 2)2@)
I@ 2)3@$ 2))@# 251@3 25$@2 25$@1 25)@# 258@2 2)3@1 2$1@#
@ 25@$ 283@3 28) 2#6 2#)@$ 363 2#8@) 366@2 258@)
Ta)*e 46 %aging o3erhead /or the ':trace )ased TAL design
The res#lts show that paging o!erhead increases slightly for E*trace %ased T'L design& Thecost of T' is 03 times more than paging& E*trace %ased T'L config#rations are optimi.ed
for T' more than paging& There is a trade*off %etween #pdate and paging o!erheads i&e& if
one increases the other decreases and !ice !ersa& E*trace %ased T'L design red#ces T'
o!erhead considera%ly while there is a small increase in paging o!erhead& owe!er" the total
o!erhead is still considera%ly lower than con!entional T' design&
.+ Cost o- !ata Collection
's descri%ed in Section ;&0" E traces are collected #sing special T'L config#ration& This T'L
config#ration is #sed for certain Es (sample Es) for a certain amo#nt of time& The Es
perform #pdate whene!er they change a cell& This ena%les to track E mo!ement for the time
period& The location of sample Es is known to the gran#larity of cell& ence there is no
paging o!erhead and the cost of data collection is the T' o!erhead generated %y sample
Es& This cost m#st also %e considered when e!al#ating E*trace %ased T'L design since E
traces are prere$#isite for designing it& Ta%le 8 present the o!erall o!erhead and the
associated model cost #sing E*trace scenario I&
51
7/23/2019 Exploiting Tal
42/46
!. Cost o/ "ata Co**ection
um-er of Sample %Es onentional TA
;esign Total 9er+ea,
ost of ;ata
ollection
1!;
121)@
16)8@2 $
;; 1666@# #6
B!; 815@ 218
1;;; 83@) 2#1
12!; )6@3 358
1B!; 5#2 )1
2;;; 5$#@3 )#)
2!;; 516@1 3
Ta)*e !6 Cost o/ data co**ection 8':trace scenario I9
+rom Ta%le 8 it is seen that generating E*trace %ased T'L #sing more sample E traces offers
more red#ction in o!erall o!erhead& owe!er" the cost of data increases as n#m%er of sample
Es ad!ersely impacting the performance of E*trace %ased T'L design compared to
con!entional T' design& ence" it is not ad!antageo#s to collect E traces #sing T'L&
.1 /ule o- Thum* TAL
-#le of th#m% T'L design as eplained in Section ;&; is also #sed to generated T'L for the 8
E*trace scenarios& Ta%le B presents the res#lts& The res#lts are compared to %oth
con!entional T' and E*trace %ased T'L designs&
Sc+eme Scenario Scenario Scenario Scenario = Scenario =
7/23/2019 Exploiting Tal
43/46
!.11 Conc*usion
.11 Conclusion
This chapter compares the performance of E*trace %ased T'L design to con!entional T'
scheme& The n#merical res#lts show that T'L schemes ha!e the potential to considera%ly
red#ce signaling o!erhead compared to the con!entional T' scheme& E!en #sing as small as
083 E traces to generate a T'L config#ration and applying to a large*scale network yields a
lower signaling o!erhead compared to the con!entional T' scheme& sing more E traces to
design T'L f#rther red#ces signaling o!erhead& The draw%ack howe!er is the cost associated
with collecting E traces& The high cost makes T'L an infeasi%le approach for collecting E
traces and red#ces its performance& The compared r#le of th#m% approach for designing T'L
performed %etter&
7/23/2019 Exploiting Tal
44/46
Chapter $ Conc*usion
Chapter & Concl on
The aim of the thesis work has %een to eploit T'L concept in LTE networks& The T'L concept
is a relati!ely new concept and this thesis work was aimed to eplore how T'L concept can
red#ce total o!erhead of a network& 'n approach for designing T'L #sing traces was pro!ided
and e!al#ated against con!entional T' scheme and T'L design %y r#le of th#m% approach&
The res#lts show that T'L pro!ides %etter performance (lower signaling o!erhead) compared
to the con!entional T' design for the #sed network scenario& The draw%ack howe!er is thecost for collecting E traces& -#le of th#m% is a fast and simple approach for designing T'Ls
for large*scale networks&
The need to red#ce o!erhead is a key factor in today/s mo%ile networks d#e to increasing
n#m%er of #sers& This is a hot research topic for researches& The introd#ction of LTE has made
this research more interesting d#e to the flei%ility it offers in designing a network compared
to pre!io#s generations& The concept of T'L is an interesting research area s#%=ect %#t !ery
limited research on the topic is a!aila%le as the concept is $#ite new& T'Ls ha!e the potentialto red#ce o!erhead and also allow for easy re*config#ration of the network&
+#t#re work can %e directed towards generating a %etter algorithm to design T'L
config#rations& +#rther red#ctions can %e made in o!erhead %y pro!iding E specific T'Ls i&e&
assigning each E entering a partic#lar cell a T'L that is designed considering E mo!ement
and call history&
:5
7/23/2019 Exploiting Tal
45/46
%eferen
?0@ :644& TS 5:&;30 M !2&B&3" 6eneral 4acket -adio Ser!ices (64-S) enhancements for E!ol!edni!ersal Terrestrial -adio 'ccess Network (E*T-'N) access" :6 -elease 2" 5331&httpAKKwww&:gpp&orgKftpKspecsKarchi!eK5:seriesK5:&;30K
?5@
7/23/2019 Exploiting Tal
46/46
?0B@ Naor" V& Le!y" & " P,inimi.ing the wireless cost of tracking mo%ile #sersA an adapti!e thresholdscheme"P IN+>7>, /12& Se!enteenth 'nn#al Roint 7onference of the IEEE 7omp#ter and
7omm#nications Societies& 4roceedings& IEEE" !ol&5" pp&953M959 !ol&5" 51 ,arM5 'pr 0112&
?09@ 'kyildi. I& +&" o R& S& ,& Lin & ct 0119&
?50@ Wang W& 'kyildi. I& +& St%er 6& L& 7h#ng ptimi.ation and 4erformanceE!al#ationU Licentiate Thesis No& 0;9:" Linkping ni!ersity" 5300&