View
226
Download
6
Category
Preview:
Citation preview
47
CHAPTER 3
PROPOSED SOLUTIONS FOR VERTICAL HANDOVER
DECISION IN 4G NETWORKS AND VANETS
3.1 INTRODUCTION
The various vertical handover decision solutions proposed by this
thesis are presented in this Chapter. A simple handover management process
based on mobile-terminals and network controlled inputs are described. The
various seamless vertical handover decision techniques such as application-
oriented, QoS-centric and congestion-aware approaches are presented.
Seamless Vertical Handover Decision in multimode mobile
terminal is the critical challenge in the 4G networks. The existing solutions
are based on Signal strength and network resources. Hence disruption of
service to the end user (hard handover) happens during the handover. The
proposed system achieves the seamless soft handover by considering the
complex metrics like momentary cost, QoS, user preferences etc. IEEE 802.21
MIHF based network scanning has great impact on improvised handover
performances. Also the game-theory and constrained Markov Decision Process
(MDP) based decision for soft handovers are presented. The seamless soft
handover is the most expected requirement in 4G heterogeneous networks.
3.2 MOTIVATION FOR SEAMLESS VERTICAL HANDOVER
Fourth Generation Networks (4G) enables the integration and
interworking of current wireless systems. With 4G/NGWNs, the user
48
dem
ands
fo
r se
amle
ss
roam
ing
acro
ss
diff
eren
t w
irele
ss
netw
orks
an
d
diff
eren
t op
erat
ors
to p
rovi
de v
ario
us l
evel
s of
bro
adba
nd s
ervi
ces
such
as
voic
e, d
ata
and
mul
timed
ia a
t ve
hicu
lar
spee
ds.
The
verti
cal
hand
over
deci
sion
bas
ed o
n th
e re
ceiv
ed s
igna
l st
reng
th a
s th
e so
le c
riter
ion
is
inef
ficie
nt a
nd im
prac
tical
in h
eter
ogen
eous
net
wor
ks. M
ore
com
plex
met
rics
like
mom
enta
ry c
ost,
band
wid
th, p
ower
con
sum
ptio
n, n
etw
ork
cond
ition
s an
d
user
pre
fere
nces
are
als
o co
nsid
ered
for
det
erm
inin
g th
e se
amle
ss v
ertic
al
hand
over
de
cisi
on.
The
seam
less
ve
rtica
l ha
ndov
er
deci
sion
fo
r 4G
hete
roge
neou
s ne
twor
ks e
nabl
es u
ser t
o se
lect
the
best
net
wor
k w
hich
off
ers
the
max
imal
qua
lity
of se
rvic
e at
min
imal
cos
t.
The
exis
ting
syst
em’s
mob
ility
mod
el d
oes
not
cons
ider
the
lan
e-
chan
ges,
emer
genc
y br
akin
g, s
peed
lim
its a
nd t
opol
ogic
al m
aps
in b
uild
ing
the
V-2
-V a
nd V
-2-I
com
mun
icat
ions
. Hen
ce t
he p
ropo
sed
syst
em p
rovi
des
seam
less
ve
rtica
l ha
ndov
er
deci
sion
fo
r V
AN
ETs
(Veh
icul
ar
Ad
hoc
Net
wor
ks) b
y co
nsid
erin
g th
e re
al-ti
me
mob
ility
mod
el w
hich
is in
turn
use
ful
in b
uild
ing
inte
llige
nt tr
ansp
orta
tion
syst
ems.
3.3
PRO
POSE
D S
OL
UT
ION
SY
STE
MS
FOR
4G
NE
TW
OR
KS
AN
D V
AN
ET
S
The
follo
win
g sy
stem
s ar
e pr
opos
ed i
n or
der
to a
chie
ve t
he
obje
ctiv
es a
nd so
lutio
ns to
the
prob
lem
s des
crib
ed in
the
sect
ion
2.6.
1.A
pplic
atio
n ba
sed
Seam
less
Ver
tical
Han
dove
r
2.A
dapt
ive
Ban
dwid
th A
lloca
tion
(AB
A)
and
Cal
l A
dmis
sion
Con
trol (
CA
C)
3.C
onge
stio
n-A
war
e V
ertic
al H
ando
ver
4.M
ulti-
Play
er G
ame-
Theo
retic
Nas
h-eq
uilib
rium
Han
dove
r
49
5.EM
DP
base
d M
ulti-
attri
bute
Wei
ghte
d-R
anki
ng H
ando
ver
6.Sp
eed
base
d R
eal-t
ime
mob
ility
sup
porte
d V
ertic
al H
ando
ver
with
Em
erge
ncy
Bra
king
In
Four
th
Gen
erat
ion
(4G
) N
etw
orks
, th
e se
amle
ss
verti
cal
hand
over
de
cisi
on
fram
ewor
k is
pr
opos
ed
usin
g IE
EE
802.
21
Med
ia
Inde
pend
ent
Han
dove
r Fu
nctio
n (M
IHF)
. Th
e ad
aptiv
e ba
ndw
idth
man
agem
ent
and
call
adm
issi
on c
ontro
l m
echa
nism
s ar
e al
so i
ncor
pora
ted
with
th
e ha
ndov
er
man
agem
ent
syst
ems.
The
gam
e-th
eore
tic
verti
cal
hand
over
dec
isio
n m
odel
inc
lude
s lo
ad a
nd c
onge
stio
n fa
ctor
s ba
sed
on
Erla
ng lo
ss fo
rmul
ae.
In V
AN
ETs,
the
gam
e-th
eory
bas
ed s
eam
less
ver
tical
han
dove
r
deci
sion
with
hyb
rid E
mer
genc
y-B
raki
ng s
yste
m i
s pr
opos
ed i
n or
der
to
achi
eve
effe
ctiv
e ve
hicu
lar
com
mun
icat
ions
. The
han
dove
r de
cisi
on m
akin
g
prot
ocol
wor
ks a
long
with
hyb
rid m
obili
ty m
odel
s of
EEB
L (E
mer
genc
y
Elec
troni
c B
raki
ng L
ight
s) a
nd ID
M (I
ntel
ligen
t Driv
er M
odel
). It
has
vario
us
emin
ent f
eatu
res s
uch
as:
Prop
agat
ion
of a
ccid
ent a
nd c
ollis
ion
war
ning
s to
am
bula
nce
or p
olic
e st
atio
ns
Red
uctio
n of
spe
ed o
f th
e ve
hicl
es b
efor
e cr
ashi
ng t
o av
oid
acci
dent
s.
50
3.4
HA
ND
OV
ER
MA
NA
GE
ME
NT
PR
OC
ESS
A
wire
less
ne
twor
k w
hich
co
nnec
ts
usin
g di
ffer
ent
acce
ss
tech
nolo
gies
is
ca
lled
a he
tero
gene
ous
netw
ork
sinc
e it
mai
ntai
ns
its
conn
ectio
ns
whi
le
switc
hing
fr
om
one
netw
ork
to
anot
her
netw
ork.
Het
erog
eneo
us n
etw
orks
util
izin
g a
dive
rse
set
of b
ase-
stat
ions
can
be
depl
oyed
to
impr
ove
spec
tral
effic
ienc
y pe
r un
it ar
ea.
H
eter
ogen
eous
netw
orks
mus
t ha
ve t
he a
bilit
y to
man
age
and
cont
rol
inte
rfer
ence
in
netw
orks
whi
ch p
rovi
des
subs
tant
ial g
ains
in th
e ca
paci
ty a
nd p
erfo
rman
ce o
f
wire
less
syst
ems i
n th
e fu
ture
.
51
The handover management process is shown in Figure 3.1. The
various strategic metrics for handover decisions are network-controlled,
mobile-terminal controlled and both network and mobile terminal controlled.
In this context, a network selection decision is made at call setup and
subsequently the decision is re-made if a handover is triggered.
The process consists of the following three main steps:
Handover Discovery - this step can play different roles:
monitoring the network conditions, listing the available
networks, predicting/estimating the characteristics for each
network, etc. and using the monitored data to trigger a HO
decision.
Handover Decision - handles the Network Selection process
and is initiated either by an automatic trigger for a HO for an
existing call or by a request for a new connection by the
mobile device. The selection of the best network is decided
based on the decision criteria provided by the device, the
application and the monitoring process. After the target
network is selected, the call is set up on the target candidate
network. Traditionally, this decision was made by the network
operators both for mobility and load balancing reasons. This
decision was based mainly on a single parameter (i.e.,
Received Signal Strength (RSS)). Whereas the proposed
system uses the Bayesian Game Theory concept for network
selection.
Handover Execution -After the target network is selected, the
connection is set up in the target candidate network. In case of
an existing connection, HO is executed and the original
52
connection is torn down and the call data is re-routed to the
new connection. If the first choice network is unavailable,
then the next listed candidate is chosen as the target network.
Connection setup (and teardown in the case of handover) will
be handled by a mobility management protocol.
3.4.1 Vertical Handover Decision Parameters
The application level vertical handover decision parameters and
triggering handoffs are explained as follows:
Decision Processing Delay is the processing time needed by a
mobile node to decide the network to which it has to
handover.
Handover Blocking Rate: It represents the percentage of
calls that do not finish their services. Lower level of blocking
rate reflects the overall system service continuity and scheme
efficiency.
Packet Loss is the amount of data lost during the handover
process. This depicts the efficiency of the scheme in terms of
availability.
Transmission Delay is the amount of time taken for
switching from a existing network to a foreign network during
handover. It is also a measure of the service continuity during
the handover process.
3.4.2 Triggering Handoffs
The Upward-Initiate begins when several beacons from the current
overlay network are not received. The Downward-Initiate begins when
53
several beacons are heard from a lower overlay’s network interfaces (NI). The
handoff starts when the lower overlay becomes reachable or unreachable, and
ends when the first data packet forwarded from the new overlay network
arrive at the MH.
The cross-layer information available in the different layers is
considered for vertical handover decision. The cross-layered handoff solution
scheme is more effective and it is explained in the subsequent section.
3.5 SEAMLESS VERTICAL HANDOVER DECISION
TECHNIQUES
There are many solution strategies for providing seamless vertical
handover support in 4G networks and VANETs which are:
Seamless integration of Mobile WiMAX in 3GPP Networks to
provide efficient seamless broadband wireless access.
Vertical Handoff algorithms for connection management and
optimal resource allocation using link-layer MIHF for
Seamless Mobility at vehicular speeds.
All-IP principle which hides heterogeneity and achieve
convergence of various networks. The IP at layer-3 and MIHF
at layer-2 make network assisted/mobile controlled seamless
mobility. If the mobile device capabilities are limited, handoff
decisions are taken by mobility agents (IDE) on the network
side.
There is an existence of handover decision taxonomy module.
There are the four following handover strategies - network-controlled,
mobile-controlled, network-assisted and mobile-assisted handover. The
54
netw
ork-
cont
rolle
d an
d m
obile
-ass
iste
d so
ft-co
mpu
ting
verti
cal
hand
over
deci
sion
(VH
D) a
ppro
ache
s are
pre
sent
ed in
the
follo
win
g se
ctio
ns.
3.5.
1 N
etw
ork
Para
met
ers B
ased
Ver
tical
Han
dove
r D
ecis
ion
The
netw
ork
disc
over
y m
odul
e re
ports
the
Net
wor
ks’
Rec
eive
d
Sign
al S
treng
th (
RSS
), ba
ndw
idth
, jit
ter,
dela
y, e
tc.,
para
met
ers
of e
very
avai
labl
e ne
twor
k in
the
cov
erag
e ar
ea o
f th
e M
obile
Nod
e (M
N).
The
hand
over
nec
essi
ty e
stim
ator
mod
ule
com
pare
s th
e ne
twor
k re
sour
ces
with
user
’s re
quire
men
t and
dec
ides
the
nece
ssity
of h
ando
ver.
The
hand
over
dec
isio
n m
odul
e de
cide
s an
d ch
oose
s th
e ne
twor
k
whi
ch o
ffer
s hi
ghes
t Q
oS a
t m
inim
al c
ost
whe
n th
e M
N i
s in
the
ove
rlay
cove
rage
of
the
netw
orks
. Th
e ha
ndov
er e
xecu
tor
mod
ule
exec
utes
the
hand
over
pro
cess
in L
2 (li
nk-la
yer)
or
L3 (n
etw
ork-
laye
r) o
r bot
h L2
and
L3
cros
s-la
yer b
ased
seam
less
Ver
tical
Han
dove
r.
The
follo
win
g ne
twor
k pa
ram
eter
s na
mel
y re
ceiv
ed s
igna
l stre
ngth
(RSS
), ba
ndw
idth
, con
gest
ion
and
cost
are
con
side
red
for
verti
cal h
ando
ver
deci
sion
. The
follo
win
g ap
proa
ches
are
use
d to
ach
ieve
the
seam
less
ver
tical
hand
over
dec
isio
n an
d al
so g
over
n th
e Q
ualit
y of
Exp
erie
nce
(QoE
) of
the
user
by
usin
g Q
ualit
y of
Ser
vice
(QoS
) par
amet
ers.
1.Se
amle
ss M
edia
Inde
pend
ent R
esili
ence
Trig
gerin
g (S
MIR
T)
2.A
pplic
atio
n-or
ient
ed M
edia
-Ind
epen
dent
Ver
tical
Han
dove
r
Dec
isio
n (A
MIV
HD
)
3.A
dapt
ive
Ban
dwid
th A
lloca
tion
(AB
A)
and
Cal
l A
dmis
sion
Con
trol (
CA
C)
4.C
onge
stio
n-A
war
e V
ertic
al H
ando
ver D
ecis
ion
(CA
VH
D)
55
The seamless vertical handover can also be achieved using soft
computing approaches which are explained in the subsequent section.
3.5.2 Soft Computing Based Vertical Handover Decision
The soft-computing techniques like neural networks and fuzzy
logic, Grey-Relational Analysis, Nash-equilibrium game-theoretic approach,
MDP based Ranking approaches are used to achieve the handover decision.
Soft computing based vertical handover approaches are:
1. Weighted – SAW and MEW Vertical handover
2. GRA – Grey Relational Analysis Vertical handover
3. TOPSIS based Vertical handover
4. Game-Theoretic Nash-Equilibrium based Vertical handover
5. Extended MDP (EMDP) based Vertical handover
6. Real-Time Mobility Models Framework for VANETs
The above mentioned network parameters based VHD and soft
computing based VHD design details and its implementations are provided in
Chapter 5, Chapter 6 and Chapter 7 respectively.
3.6 SEAMLESS MEDIA INDEPENDENT RESILIENCE
TRIGGERING (SMIRT)
The dual-mode mobile stations (MSs) which roam between
wireless local area network (WLAN) and cellular networks. The act of
transitioning from WLAN to cellular is commonly referred to as a vertical
handoff (VHO). The Seamless Media Independent Resilience Triggering
(SMIRT) framework is developed to achieve Vertical Handover Decision
56
(VHD) in Heterogeneous 4G Wireless Overlay Networks. Figure 3.1 shows
the SMIRT architecture for VHD.
A call admission control (CAC) algorithm is another key factor
that enables efficient system resource utilization while ensuring that
connection-level QoS requirements are satisfied. CAC is always performed
when a mobile initiates communication in a new cell, either through a new
call or a handoff.
3.6.1 SMIRT Framework
SMIRT framework helps to do handover for heterogeneous
networks and has the following modules as shown in Figure 3.2.
Figure 3.2 SMIRT Framework Architecture
57
Network Discovery: MT searches for reachable wireless networks.
Handover Decision:
Handover necessity estimation: It determines whether a
handover is necessary or not from the current network.
Handover target selection: It chooses the “best”
network among the available network based on a set of
criteria.
Handover triggering condition: It determines the right
moment to initiate a handover out of the currently
connected network.
3.6.2 SMIRT Algorithm
The SMIRT algorithm is explained as follows:
Step 1 : Discover the available networks using the
interface services of IEEE 802.21 MIHF.
Step 2 : Calculate the quality of the networks after
analyzing the handover necessity estimation.
Step 3 : Select the network which has highest QoS.
Step 4 : Perform make-before-break soft handover.
Step 5 : Trigger the handover at the right moment.
58
3.7 CROSS-LAYER BASED VERTICAL HANDOVER
DECISION
The proposed handover system is based on cross-layered approach.
It has four new functional modules. Monitoring agent (MA) for each protocol
layer, and profile database (PDB) are the functional modules that facilitate
efficient inter-layer communications. The MAs are the interface for each
legacy protocol layers to monitor and collect protocol specific dynamic status
information as well as to adjust the protocol controls without requiring direct
modification to the existing protocols.
Figure 3.3 Cross-Layered Handover Approach
The PDB maintains both the static and dynamic information
necessary for handover related decisions and processes, and the dynamic part
of the information in the PDB are updated by the MAs. The protocol specific
Application Side
ApplicationNetwork Side
IP Agent
Network
Data Link
Physical
Profile DB(PDB)
Decision Engine (DE)
Transport
59
dynamic status information collected in the PDB greatly facilitates the
availability of cross-layer information necessary for handover decisions. The
decision engine (DE) maintains the per-application handover processing
policies to enable seamless handover of each user session.
A set of rules to determine when to trigger the handover decision
procedure for a certain service flow is also maintained to avoid unnecessary
handover decision processing caused by redundant status reports from
multiple MAs. In making the handover decisions, the DE utilizes the
information of the predefined key parameters across the protocol layers by
obtaining the necessary static/dynamic data from the PDB. IP Agent is
responsible for the mapping of the end point addresses of ongoing sessions to
the addresses corresponding to the current location. It enables the discovery of
a peer’s current location as well as the continuity of data delivery transparent
to the mobility by tracking IP address changes of end points as shown in
Figure 3.3.
This chapter provided the literature survey along with the
limitations of the existing systems and the proposed systems which are
providing the solutions.
3.8 GAME-THEORY BASED VERTICAL HANDOVER
DECISION
The seamless vertical handover decision is based on game-theoretic
Nash-equilibrium. Vertical Handoff Decision Making problem is one of the
significant technical issues in the development of Heterogeneous wireless
networks. This paper presents a vertical handoff decision algorithm based on
game theory approach. In this algorithm, the handoff decision problem is
formulated as a non cooperative game between the mobile users and the
wireless networks that are available in the vicinity of the mobile user. It
60
considers terminal parameters such as QOS requirements of the mobile
application along with the velocity of the mobile terminal. It also considers
network parameters such as available bandwidth and cost per bit offered by
each network. The algorithm chooses the target network with maximum
network utilization that offers services at lower prices.
3.8.1 Game-Theory Based Vertical Handover in VANETs
Thesis focuses on a vertical handoff decision algorithm based on
Bayesian Evolutionary game model. When a mobile host is under the
coverage of more than one wireless network, it performs network selection
iteratively to achieve best quality of service at minimum cost. The decisions
evolve to an equilibrium point at which the quality of a network access
service is maximized and the cost of the service is minimized.
This solution is termed as Bayesian Nash equilibrium and is found
by solving pair-wise matrix. If there is no Nash equilibrium solution for the
game, the algorithm finds the sub-optimal solution. The Game Theory based
Seamless Vertical Handover Decision with hybrid Emergency-Braking in
Vehicular Network follows the following principle:
As long as the vehicles have sufficient inter-vehicular distance
the speed remains the same as set by the user.
Reduce the speed of the vehicle when the inter-vehicular
distance is lesser than safe distance.
Broadcast warning messages to other vehicles behind, when
the vehicle decelerates fast indicating a crash.
The Vehicles receiving the packet broadcasts the packet
before coming to halt.
61
Choose the best available network for vehicle to roadside
communication with respect to the changing speed values of
the vehicle using game theory and MDP concepts.
The various types of Game Theory models are as follows:
3.8.1.1 User Vs user
Co-operative:
Many nodes with one service provider.
Strategy used: Random arrival value is assigned on a first come
first serve basis. Here, the bandwidth is divided to the users based on their
arrival.
Non co-operative:
Many nodes, many service providers
Strategy used: Evolutionary Game theory
To select the least congested AP. As a node arrives into a new area
with lot of other nodes it simply selects the AP which has less number of
nodes. A weighted sum score function based on the AP’s load, the price and
the distance at which the user must travel to reach the new AP is calculated.
3.8.1.2 User Vs network
Co-operative:
Grim strategy- once the user leaves the network he never
returns back to it.
Non co-operative:
62
Bargain strategy-many nodes compete for a network by
bidding. The node with highest bid gets maximum payoff
from the service provider.
3.8.1.3 Network Vs network
Co-operative:
Bankruptcy game
The node which arrives without any bandwidth is treated as
Bankrupt node. So all the network providers combine and provide the
bandwidth.
Non co-operative:
The networks compete against each other, seeking to maximize
their individual revenues. The solution point is obtained using non-
cooperative Game-theoretic Nash-equilibrium. If multiple solutions exist then
sub-optimal solution is obtained using the ranking approach.
3.8.2 Constrained MARKOV Decision Process (MDP) Based
Vertical Handover
The rank value is calculated by using benefit and penalty values of
the each available network. The Markov chain is a common tool for decision
making. MDP can be used to handle multi objective dynamic decision-
making problem. MDP model solves network selection and VHO decision
at one time by considering both benefit f(st, at) and handover cost g(st, at).If
only f(st, at) is considered, this model tells us the best network at all the
decision epochs.
63
The MDP decision system model is presented and the governing
Equations (3.1) to (3.7) are represented as follows:
EMDP System Model
The system model and the decision approach of the EMDP model
are presented as follows:
1. Decision Epoch (DE)
DE = F x T x P (3.1)
where
F = {fixed(f), mobile(m)}
T = {voice(v), data(d)}
P = {preference: low-cost(lc), high-quality(hq)}
States (S) is defined as
S = DE x M (3.2)
M = N x B1 xD1xJ1xE1x…….x Bn xDnxJnxEn x V x L x C
where
n N
N = Set of available networks
Bn = bandwidth of the network n; Bn ={1,2,….bnmax}
Dn= delay of the network n; Dn ={1,2,….dnmax}
64
Jn = Jitter of the network n; Jn ={1,2,….jnmax}
En= error-rate of the network n; En ={1,2,….enmax}
V = {0,1,2,………vmax}
L = {1,2,3……lmax}
C={ Conversational, Streaming, Interactive, Background}
2. Let us introduce vectors for the current and next-state
Current-State, S = {i , b1, d1, j1, e1, …… b|n|, d|n|, j|n|, e|n| , v, l, c}
Next-State, S’= {j , b’1, d’1, j’1, e’1, …… b’|n|, d’|n|, j’|n|, e’|n| , v’, l’, c’}
3. Actions
A={Conversational, Streaming, Interactive, Background} (3.3)
Example: Conversational – VoIP, Streaming – Video, Interactive – HTTP,
Background –e-mail
4. Transition Probability
DEDE';0DEDE'if];e,j,d,be,j,d,Pr[bl][l'PV][V'P
a]S,[S'P nnnn'n
'n
'n
'nNnrr
r (3.4)
where a A
v’ = v + (1 + 21
1
2
' mr
m-1 m m+1max
m
m-1 m
ALTALT1+ ALT if 1
ALTP / if 2 1ALT ALT ALT
if l = lALT
ALT ALT
ll l l to m
65
ALTi – Area of LTi ; LT-Location Type
5. Policy-Reward function
fr(S, a) = fben(S, a) fpen(S, a) (3.5)
6. Optimal-policy
The optimal policy is governed to provide the maximum QoS with
minimal cost of the selected target network as per the Lagrangian’
approach
Maximize the reward function
)(S'a]S,S'[P)a;(S,rAa
max(S)
SS'r
S
(3.6)
– discount-factor (ie., 0 to 1)
The Q-learning algorithm gives the proper .
maxCCk1
zk1k )
where z varires from 1 to N
and
Minimize the cost
Ctotal = min {Cjtotal /j=1,2,…N} (3.7)
66
The sample QoS offered and requested values are presented below:
Table 3.1 Offered QoS Parameters
Table 3.2 QoS Threshold Parameters
BTh DTh VTh JTh ETh
80 100 60 50 150
Table 3.3 QoS Required Parameters
Parameter / QoS Classes
Bandwidth (Mbps)
Packet Delay(msec)
Supported Velocity (Kmph)
Jitter(msec)
BER (per 108)
UGS
(Voice – VoIP)
10 200 5 60 400
rtPS(Streaming-Video)
25 300 5 60 400
nrtPS
(Interactive-HTTP)
15 300 5 200 250
BE
(Background-Email)
20 400 5 300 250
Parameter / Network
Bandwidth (Mbps)
Delay(msec)
Supported Velocity (Kmph)
Jitter(msec)
BER(per 108)
WiFi 54 160 10 50 200 WiMAX 70 120 20 40 150 LTE-A 150 70 50 30 100
67
Table 3.4 Cost per bit Offered by Each Network
Network WiFi WiMAX LTE-A
Cost 0.2 0.4 0.6
Table 3.5 Pair-Wise Nash-Equilibrium Matrix
Network /
QoS Classes
WiFi WiMAX LTE-A
UGS
(Voice – VoIP)
0.094, 0.128, 0.170,0.6
rtPS (Streaming-Video)
0.106, 0.133, 0.129,0.6
nrtPS
(Interactive-HTTP)
0.056, 0.065,0.4 0.122,0.6
BE
(Background-Email)
0.056,0.2 0.071,0.4 0.130,0.6
It is observed from the pair-wise matrix that the equilibrium is
achieved for various traffic classes and hence the decision can be made
optimally. For example, VoIP calls select LTE-A network by using Nash-
equilibrium solution point.
68
Figure 3.4 Delay Vs Network-Utility
Figure 3.5 Delay Vs QoS-Ratio
0
0.05
0.1
0.15
0.2
0.25
0.3
200 400 600 800
Net
wor
kut
ility
DELAY
Conversation-varying Delay
LTE
WIMAX
WIFI
0
0.05
0.1
0.15
0.2
0.25
0.3
40 80 120 160 200
QoS
RATI
O
DELAY
VIDEO- QoS vs DELAY
LTE
WIMAX
WIFI
69
The performance graph shows that Network-Utilization and QoS ratio
are relatively high in LTE networks which is shown in Figure 3.4 and
Figure 3.5 respectively. The pair-wise solution matrix shows that equilibrium
solutions arrived for VoIP, Straming and Intreactive traffic classes at LTE-A,
LTE-A and WiMAX networks respectively. But there is no solution point for
E-mail services, hence the sub-optimal solution is obtained using MDP based
ranking method.
Example: Conversational – VoIP, Streaming – Video, Interactive – HTTP,
Background –e-mail
The selection of the best network is based on the highest reward
with minimal penalty value of that network in order to maximize the objective
function. In this work, a vertical handoff decision algorithm for 4G wireless
networks is presented. The problem is formulated as a constrained Markov
decision process. The objective is to maximize the expected total reward per
connection subject to the expected total access cost constraint.
3.9 SUMMARY
This Chapter presented the various vertical handover decision
solutions proposed by this thesis. A simple handover management process
based on mobile-terminals and network controlled inputs were discussed. The
various seamless vertical handover decision techniques such as application-
oriented, QoS-centric and congestion-aware approaches for 4G networks were
presented. Also the Game-theoretic and MDP approaches were applied to
vertical handover decision for vehicular networks and explored. The general
vertical handover decision frameworks for 4G networks and VANETs are
presented in Chapter 4.
Recommended