Content uploaded by Sumit Mittal
Author content
All content in this area was uploaded by Sumit Mittal on Sep 22, 2022
Content may be subject to copyright.
ORIGINAL RESEARCH
An intelligent handoff optimization algorithm for network
selection in heterogeneous networks
Pooja Dhand
1
•Sumit Mittal
2
•Geeta Sharma
1
Received: 30 November 2020 / Accepted: 3 May 2021
Bharati Vidyapeeth’s Institute of Computer Applications and Management 2021
Abstract Heterogeneous wireless networks are an inte-
gration of dissimilar networks for better coverage and
spectral efficiency. These networks differ in cost, reliabil-
ity, bandwidth, coverage, security, and other related
parameters. And cell users roam across these networks and
gets connected to the new candidate network during
handoff to maintain call continuity for uninterrupted ser-
vice delivery. But the decision of selecting an optimal
network in such a diverse environment is a challenge. To
cater this, Handoff Optimization Algorithm (HOA) having
two phases is proposed-Handoff Initialization phase and
Network Selection or Decision making phase. This paper
discusses the network selection phase of Handoff Opti-
mization Algorithm (HOA) for Future Networks (FN). Six
key parameters based on network, mobile and user interest
are selected after careful analysis and understanding. Fuzzy
Analytical Hierarchical Process (FAHP) is used to assign
weights to these key parameters and then Elimination and
Choice Expressing Reality (ELECTRE) is applied to rank
different networks. Results show that decision making
phase of HOA selects the optima network from among the
candidate networks based on network, mobile or user
inputs.
Keywords Handoff Heterogeneous networks Handoff
decision making phase Fuzzy Analytical Hierarchical
Process (FAHP) Network Selection Elimination and
Choice Expressing Reality (ELECTRE)
1 Introduction
Information society has encountered a radical transforma-
tion [1] from conventional communication framework to
smart web empowered specialized gadgets in most recent
couple of decades. The fundamental main impetus of such
society is Information and Communication Technology
(ICT) which guarantees a unified model for sustainable
economic growth [2]. ICT gained momentum with the
progression of internet which had a major impact on
classical ICT. It is assessed that the development rate of the
Internet is twofold every year of its existence [3] and this
boom in Internet advancements has (a) made numerous
technological and social developments, for example,
Tumblr, Instagram, Facebook, WeChat, LinkedIn, Twitter,
and so forth (b) promotion of e-services such as e-gover-
nance, e-waste, etc., for urban welfare government,
employees and citizens. (c) Assisted in building ambience
intelligent and expert systems through sensors, artificial
intelligence, cloud computing, edge computing, pervasive
computing and ubiquitous computing. (d) Framing a safe
and secure infrastructure by using electronic surveillance
for smart and healthy ecosystem.
This high-tech revolution and its empowering impact
have connected billions of individuals on social media and
trillions of people through E-commerce, mobile gamming
and M-commerce. Consequently, it is seen that these smart
handheld gadgets and web has altered the world with its
endless features and advantages. Every citizen dependably
&Geeta Sharma
gsharma3210@gmail.com
Pooja Dhand
pooja_dhand@hotmail.com
Sumit Mittal
sumit.mittal@mmumullana.org
1
Lyallpur Khalsa College Technical Campus, Jalandhar,
Punjab, India
2
M.M. Institute of Computer Technology and Business
Management, Mullana, Haryana, India
123
Int. j. inf. tecnol.
https://doi.org/10.1007/s41870-021-00710-1
need to stay best associated with web to look for frequent
updates, inquiries and other related information. But pro-
viding continuous and ubiquitous internet connectivity to
users is putting high pressure on wireless for providing
service and call continuity by limiting QoS deterioration
[4]. To achieve this, varied types of wireless access tech-
nologies are integrated and deployed such as UMTS, LTE-
A, Wi-Fi, WiBro, Wi-MAX, etc., to form a heterogeneous
network environment [5]. When a mobile user is nomadic
across such next generation heterogeneous network,
mobility framework manages the means of communication,
computing, accessibility, and interoperability of data to
provide continuity in services. The paradigm shift from
homogeneous system to next generation heterogeneous
network is an ideal solution for giving consistent network,
expanded limit, traffic offloading and universal coverage
[6] and key domains of such network are peer-to-peer
communication (P2P), Internet of things (IoT), device-to-
device communication, Internet of Services (IoS), cloud
computing, mobile computing, social media, near field
communication(NFC), ubiquitous coverage and seamless
connectivity. What’s more, this all is conceivable if our
cell phone is best associated with the Internet all the time.
With the vision of giving ’Constantly Best Connectivity’
[7] to users, there is a need to build up a robust framework
which can give better coverage and availability.
1.1 Contributions
In this paper, an intelligent handoff decision making phase
for network selection in heterogeneous environment is
proposed [8]. In this paper’s decision making phase with
contributions two-fold is explained.
i. Firstly, six key parameters given below are selected
for handoff decision making phase.
Network based: Received signal strength indicator
(RSSI), network load, bit error rate (BER)
Mobile based: bandwidth, packet loss
User based: Battery power
ii. FAHP is applied to assign weights to these param-
eters. Before assigning weights, a priority number is
assigned on the basis of handoff initiation criteria. It
is necessary to highlight the related parameter to
understand the cause of handoff. If handoff is caused
to due to low signal strength then RSSI is given high
priority or if reason of handoff is bandwidth or any
other performance related parameter then bandwidth
is given high priority.
iii. Secondly, ELECTRE is employed to select the most
optimal network from among the candidate networks
(WiFi, WiMax and LTE-A). Again the network is
chosen on the basis of initiation cause.
1.2 Paper organization
Providing persistent connection and communication
requires a proficient handoff strategy to deal with the
mobility among heterogeneous environment. This paper
focuses on the handoff decision making phase of the
Handoff Optimization Algorithm. The paper is organized
as follows: Sect. 2focuses on background related to
heterogeneous networks and handoff procedure. Section 3
deals with Multi Attribute Decision Making (MADM)
approaches such as FAHP and ELECTRE. Section 4pro-
poses handoff optimization algorithm for HOA. Section 5
explains the empirical results of optimal network selection.
Section 6concludes with the conclusion and future work.
Section 7 gives the detail of all references used in the
paper.
2 Related work
Earlier handoff methods considered RSSI (Received Signal
Strength Indicator) as one of the prime factor for initiating
handoffs but unpredictable and uncertainty of wireless
scenario has always posed a threat and engaged researchers
in developing more quality based algorithms. Many soft
computing techniques such as machine learning, neural
networks (NN), Bayesian statistics, Multi-Attribute Deci-
sion Making algorithms (MADM), evolutionary computa-
tions, Fuzzy Logic (FL), probabilistic reasoning, and
Genetic Algorithms (GA) are being used for building up a
robust algorithm [8] for mobility management in hetero-
geneous networks.
Fuzzy logic is widely being used to predict and model
many intelligent systems for decision making due to its
flexible nature [9]. Its contribution to solve uncertainty in a
real life system is growing daily. Employing Fuzzy logic
and Fuzzy MADM in Mobility Management for the
handoff decision making process has been extensively
applied in the literature. Nguyen-Vuong et al. [10] pro-
posed a QoS based multi-criteria algorithm named Auto-
matic Network Selection (ANS) to choose the most
excellent network amongst the available during the handoff
decision making. The algorithm has been successfully
demonstrated using numerical analysis and results are
satisfactory and can benefit end users and network opera-
tors by improving network management. Liu et al.[11]
developed a multi-objective immune based mathematical
algorithm in terms of battery power and load and con-
tributed to increase the efficiency of the handoff process by
performing better in terms of remaining battery power and
dropping probability, but performs slightly less in terms of
load balance and computation time. Jailton et al.[12] pro-
vides a quality of experience based handoff framework for
Int. j. inf. tecnol.
123
heterogeneous networks. Simulation results show a better
quality for multimedia and voice service classes than tra-
ditional MIH.
Jain et al. in [13] investigated a vertical handoff
framework for reducing unnecessary handoffs by consid-
ering coverage area of the network and the velocity of the
mobile node as input. Kassar et al.[14] has proposed an
intelligent handoff management algorithm based on the
context awareness concept, fuzzy logic and AHP. The
authors have contributed in reducing unnecessary handoffs
and ping-pong effect. Yan et al.[15] has presented a com-
prehensive survey of vertical handoff algorithm based on
RSSS travelling distance, QoS based, Cost function and
fuzzy logic.
Kumar et al.[16] has employed RTOPSIS, Analytical
Network Process (ANP) and Oliver blume method for
optimizing handoff process and considers four traffic
classes, i.e. background, streaming, conversation and
interactive. Authors have determined that Oliver Blume
method is much better then ANP as the former is dynamic
and fast. Sgora et al. [17] have given a complete, thorough
detail about handoff procedure. The authors discuss and
classified the handoff prioritization schemes into different
categories such as queuing, channel reservation, subrating,
genetic and hybrid schemes.
Yang et al. [18] has proposed a MADM based handoff
algorithm based on AHP, MEW and SAW across Wireless
LAN and Wireless MAN networks. The method proposed
is better than basic handoff method as it provides smaller
handoff times and low dropping rate. To manage real time
traffic Ovengalt et al. in [19] proposed a call admission
control process using fuzzy logic and contributed promi-
nent enhancements by reducing call rejection and call
blocking probabilities in 4Gnetworks. Kaleem et al.[20]
presented the MADM based vertical handoff algorithm
which includes the handoff necessity estimation module.
This method employs AHP, FAHP, VIKOR, FVIKOR,
TOPSIS and FTOPSIS and determines the optimal time for
vertical handoff initiation. Thumthawatworn et al.[21] has
proposed an improved fuzzy based handoff mechanism
called AMHDS design II for VOIP and video streaming
applications. Authors have contributed by improving net-
work performance and reducing execution time for VOIP
and video streaming traffic.
Many attempts and efforts have been made by
researchers to explore and employ intelligent techniques
for handoff management in NGHN but still many issues
have been left for future. One of this major issue is to
consider the reason of handoff initiation before switching it
to the appropriate network. There can be many reasons for
handoff initiation such as strong RSSI, more bandwidth
availability, traffic offloading onto lighter network, etc. For
network selection, many probabilistic and analytical
solutions have been provided which select the best network
but fails to consider that whether this best network is suited
to provide the demand laid down by the current network.
3 Multi-attribute decision making methods
(MADM) in the HOA decision making
Promising service continuity and QoS in a diverse envi-
ronment is a critical issue. Heterogeneity of radio access
technologies (RAT’s) has posed many challenges in front
of researchers. Issues like underutilization of low power
nodes, imbalance between uplink and downlink coverage,
creation of coverage hole and interference problem
between different cells needs solution to provide seamless
connectivity to mobile users [6]. Many algorithms and
solutions are proposed and implemented to provide service
continuity and handle issues pertain to heterogeneous
environments [22–27]. For this next generation networks
must be capable enough to handle mobility and perform all
related functions independently of the underlying tech-
nologies [28]. To provide continuation of services while
the user is moving between different networks is the
immediate goal of next generation heterogeneous net-
works. A user switches between different networks to
maintain connection. This is called a handoff. Handoff
process is divided into three main stages named as handoff
initialization, handoff decision making and re-association
and re-authentication phase [29]. Many soft computing
techniques are used for optimal network selection in which
the information about the alternatives can be provided in
many ways such as preference ordering, preference relation
and utility function [30,31]. In this paper, Fuzzy Analyt-
ical hierarchical Process (FAHP) is used for preference
ordering in terms of network, mobile and user parameters
and ELECTRE (Elimination and Choice Translating
Reality) is used for ranking of these alternatives.
3.1 FAHP
Analytical Hierarchical Process (AHP) is a simplest tool to
assign weights to various criteria’s for decision making
[32] but it fails to deal with the vagueness of the real
physical world. Due to the random nature of real scenario,
it is difficult to map the decision to exact numbers, hence
fuzzy systems are employed to deal with the uncertain
conditions. Integrating fuzzy set theory in classical AHP
can be employed as an effective method for solving com-
plex decision making problems [33].There are many ways
to apply FAHP method on multiple criteria’s for decision
making. The earliest method for fuzzy AHP was given by
van Laarhoven and Pedrycz [34], later, Buckley [35]
implemented FAHP on the basis of the geometric mean
Int. j. inf. tecnol.
123
calculation and computing weighted matrix to calculate
fuzzy weights. Recently in 1996, Chang [36] proposed an
extent analysis method which is rather little complex than
Buckley’s method but became popular. In this paper
Buckley’s geometric mean method is used to obtain the
weight of different criteria due to its simplicity and
effectiveness.
Linguistic terms are used to represent the experts’
assessments and then triangular fuzzy numbers, are used
for evaluations which are shown in Table 1. The steps for
solving FAHP are explained as follows [27,35,37]:
Step 1: Selected criteria’s are compared via linguistic
terms for decision making as explained in Table 1. Lin-
guistic terms explains the strength of each measure and
relation with each other, accordingly TFN’s are allotted to
them. For example, if the respondent’s states ‘one measure
is as strongly important than the other measure’, then TFN
as (4,5,6) will be allotted to it and the other measure with
which it is being compared is allotted TFN as (1/6,1/5,1/4).
Pairwise comparison matrix is formed after comparing
all parameters as given in Eq. (1) where f
pk
ij indicates the
kth respondent deriving a TFN for a choice of ith criteria
over jth dimension.
~
Pk¼
~
pk
11 ~
pk
12 ~
pk
1n
pk
21 ~
pk
2n
.
.
. .
.
.
~
pk
n1~
pk
n2 ~
pk
nn
2
6
6
6
43
7
7
7
5ð1Þ
Step 2: The pairwise matrix formed in Eq. (1) explains
the preference of each respondent f
ðpk
ij Þand further these
values (l, m, u) are aggregated to form ~
pij is obtained as in
Eq. (2).
~
pij ¼PK
K¼1~
dk
ij
Kð2Þ
Step 3: After updating the pairwise values, an updated
matrix ~
Pis formed where each ~
pij value is viewed as triplet.
~
P¼
~
p11 p1n
.
.
...
..
.
.
pn1 ~
pnn
2
6
43
7
5ð3Þ
Step 4: Geometric mean is computed for a fuzzy eval-
uation matrix of each dimension as:
~
ri¼Y
n
j¼1
~
pij
!
1=n
;i¼1;2...::nð4Þ
Step 5: Compute the sum for each ~
riof the respective
dimensions.
Step 6: Calculate the inverse of the vector obtained in
step 1.1 and arrange the TFN in an ascending order.
Step 7: Multiplying each ~
riwith the reverse TFN value
obtained in step 1.2 to calculate the fuzzy weight
~
wi¼~
ri~
r1~
r2......~
rn
ðÞ
1
¼ðlwi;mwi;uwiÞð5Þ
Step 8: The de-fuzzy value of TFN is calculated by the
center of area method by using Eq. (6).
Mi¼lwiþmwiþuwi
3ð6Þ
Step 9: Normalize the non-fuzzy number using Eq. (7).
Ni¼Mi
Pn
i¼1Mi
ð7Þ
3.2 ELECTRE
ELECTRE is a MADM strategy invented in late mid sixties
by three French researchers Roy, Benayoun and Sussmann
who named it ‘‘Elimination Et Choix Traduisant la REa-
lite
´’’ [ 38–40] which means ‘‘Elimination and Choice
Translating Reality’’. For the selection of an optimal net-
work among the given set of networks using ELECTRE
requires three main components [27]:
i. Set of M alternatives such that X={1,2,……,M}
Table 1 Terms and Fuzzy Triangular Number (FTN) (Fuzzy Scale) [27]
Intensity Term Relation Fuzzy triangular number
(FTN)
1 Equally important (EI) Two activities contribute equally 1,1,1
3 Weak Importance (WI) Experience slightly favours one activity over another 2,3,4
5 Essential or strong
importance (SI)
Experience slightly more favours one activity over another 4,5,6
7 Demonstrated importance
(VSI)
Experience strongly favours one activity and its dominance is
demonstrated in practice
6,7,8
9 Absolute importance (EMI) Highly favoured one activity over another 9,9,9
Int. j. inf. tecnol.
123
ii. Set of N criteria such as Y={1, 2…N} defining
alternatives.
iii. Set of W weights such as W={w
1
,w
2
…w
m
}which
are calculated using weighting approaches AHP and
FAHP. It must be noted here that the sum of the
weights assigned to parameters in the weighted
matrix must be equal to 1.
X
m
i¼1
Wij ¼1
Steps for ranking of networks using ELECTRE are as
follows:
Step 1: Based on the value of each parameter defined for
an alternative, the set of values for i
th
alternative can be
summarized as:
Dij ¼Ai;Bi;Ci;Di;Ei;Fi
Formulating a definition matrix, which defines the each
alternative according to the value of their criteria’s, can be
written as Eq. (8) follows:
Dij
mn¼
A11 F11
.
.
...
..
.
.
An1 Fn1
2
6
43
7
5ð8Þ
Step 2:Using Dij, a normalized matrix Rij is built using
the following formula in Eq. (9)[41]:
Rij ¼Dij
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pm
i¼1D2
ij
qð9Þ
where i = 1, 2…, n and j = 1, 2…m.
Representing the output as a matrix given below:
Rij
mn¼
r11 r1m
.
.
...
..
.
.
rn1 rnm
2
6
43
7
5
Step 3: Computing a weighted and normalized matrix
Vij from the normalized matrix Rij using Eq. (10)[41]:
Vij ¼Rij Wð10Þ
Vij ¼
r11 r1m
.
.
...
..
.
.
rn1 rnm
2
6
43
7
5
w1
.
.
.
wn
2
6
43
7
5
¼
r11w1 r1m wn
.
.
...
..
.
.
rn1w1 rnm wn
2
6
43
7
5
ð11Þ
Step 4: Comparing the alternatives for worst case and
best case, concordance and discordance matrix are
computed.
i. Concordance Matrix: It defines the set of attributes
for which the one candidate network is superior to
another candidate network. It is denoted by matrix
Cij in Eq. (12) and is computed as follows [27]:
Cij ¼fj:Vij Vij gð12Þ
ii. Discordance Matrix: It is reverse of concordance
matrix and defines the set of attributes for which the
one candidate network is inferior to another candi-
date network. It is denoted by matrix Cij and is
computed in Eq. (13) as follows [27]:
Cij ¼fj:Vij\Vijgð13Þ
4 Handoff optimization algorithm
for next generation heterogeneous networks
(HOA)
The proposed Handoff Optimization Algorithm is devel-
oped for heterogeneous networks comprising of WiFi,
WiMax and LTE-A. LTE-A provides a huge coverage with
better bandwidth in urban, sub-urban and rural area. Wi-Fi
offers better services in smaller regions with less attenua-
tion and interference. WiMax provides significantly higher
data rates with multiple applications. The integration of
these networks works successfully if there is a proper
coordination between low power and high power nodes.
The proposed HOA algorithm has two modules:
1. Handoff Initiation module
2. Handoff Decision making module
In this paper, Received Signal Strength Indicator
(RSSI), Available Bandwidth (BW), Bit Error Rate (BER),
Packet Loss (PL), Network Load (NL), and Battery Power
(BP) are considered for the decision making criteria. Some
criteria’s are normalized to a value between 0 and 1 for
general and meaningful results.
HOA is based on the calculated value of RSSI from
EIRP (Effective Isotropic Radiated Power) and path loss
models such as free space, Hata and Okumara Hata model
[30].
4.1 Handoff initiation module
Handoff initiation means to trigger handoff at the right time
to avoid uninterrupted services. In order to correctly esti-
mate the need for a handoff, three fuzzy controllers have
been designed based on network, mobile and user initiation
based. Each controller has its own set of parameters to
decide that whether handoff is required or not [8]. Table 2
Int. j. inf. tecnol.
123
below is summarized to show the parameters in each fuzzy
controller.
1. Network Initiation Handoff: Handoff is initiated due
to the following reasons
i. When RSSI falls below the adaptive threshold and
neighbouring RSSI rises above the adaptive hystere-
sis margin.
ii. When Network load increases above the threshold.
iii. When the bit error rate exceeds its threshold.
2:Mobile Initiated Handoff: In this case, smart
handheld devices initiate a handoff if the current
network is not able to provide better bandwidth.
Handoff will be initiated if the file size of current
application requires higher bandwidth than available.
3. User Initiated Handoff: Handoff is initiated on the
basis of user preferences such as cost, battery power
and bandwidth. Reasons for handoff initiation are as
follows:
i. The cost of the current network is high
ii. Battery Power is low, therefore the user will look for
a network which consumes less power.
iii. Better bandwidth or data rate is a user requirement.
4.2 Handoff decision making module
Network Selection is a process of selecting the optimal
network from among the candidate networks to maintain
call continuity and avoiding unnecessary handoff by con-
sidering the exact reason of handoff initiation. If more than
one network is available in the vicinity, then the ranking of
the networks must be done to select the optimal service
provider. For this, a set of metrics is proposed in this paper
which considers network, mobile and user aspects to make
a correct decision.
Depending upon the type of handoff initiation (NIH,
MIH or UIH), a set of decision metrics is chosen and dif-
ferent weights have been assigned to them using FAHP. A
priority value of each parameter has been summarized in
Table 3on the basis of type of handoff initiation. If a
handoff is initiated on the basis of the network, then FAHP
and ELECTRE is applied by giving more priority to net-
work parameters. Similarly, if the handoff is initiated on
mobile or performance basis, then higher weight is
assigned to mobile based parameters and the same is done
for user initiated handoff. Selection of optimal network is
explained with the help of a numerical example and a case
study in Sect. 6.
5 Case study and experimental results
Multiple parameters considered in handoff initialization
phase such as network related, mobile related and user
related are assigned weights using FAHP in heterogeneous
networks. The set of alternatives and criteria’s defined by
HOA are as follows:
i. Set of M alternatives such that X={WiFi,
Wimax,LTE-A}
ii. Set of N criteria such as Y={RSSI,-
BER,NL,BW,PL,BP} defining alternatives.
5.1 FAHP application in HOA
Buckley’s FAHP method is used to assign weights to
selected set of metrics.
5.1.1 Network initiated handoff
NIH assigns higher weight to network related parameters.
Table 4shows a fuzzy pairwise comparison matrix using
linguistic variables.
Table 2 Parameters assigned to NIH, MIH and UIH
N/W Handover strategy Parameter considered
Network Initiated Handoff (NIH) RSSI, Network Load, BER
Mobile Initiated Handoff (MIH) Bandwidth and File size
User Initiated Handoff (UIH) Cost, Battery Power, data rate
Table 3 Priority assigns to the decision matrix for different initiation
criteria’s
Type of initiation RSSI BER NL BW PL BP
NIH 1 2 3 4 5 6
MIH 2 6 3 1 4 5
UIH 3 6 2 1 5 4
Table 4 Fuzzy pairwise comparison matrix using linguistic variables
for NIH
Factor RSSI BER NL BW PL BP
RSSI EI WI WI SI VSI EMI
BER EI EI WI SI VSI
NL EI WI SI VSI
BW EI WI SI
PL EI WI
BP EI
Int. j. inf. tecnol.
123
Table 5evaluates the pairwise comparison matrix i.e.
fuzzy triangular numbers based on Saaty’s scale using
Table 3.
The geometric mean for each network parameter is
computed by using Eq. (4) and then geometric means of all
parameters is computed and the summed and inverse of
values is shown in Table 6.
Using step 5, the fuzzy weight is calculated for ‘‘RSSI’’
parameter by using Eq. (5) as:
~
Wi¼½ð3:0862 0:0952Þ;ð3:7621 0:1128Þ;ð4:3645
0:1365Þ
¼½0:29380624;0:42436488;0:59575425:
From Table 7, using the fuzzy weights, Avgweight is
computed as average of the three fuzzy weights for each
parameter. Average is computed by dividing the Avgweight
column with the Total and the norm_wt vector is computed.
Hence, the obtained weights for each parameter are shown
in Table 8.
The computed fuzzy weights for each measure in NIH
are shown in Table 8which will be further used for ranking
of the networks during NIH.
5.1.2 Mobile inititiated handoff
Since handoff based on mobile based parameters is to be
computed, higher weight is assigned to mobile based
parameters, i.e., bandwidth, RSSI and network load.
Table 9shows a fuzzy pairwise comparison matrix using
linguistic variables.
Table 10 shows the pairwise comparison matrix for
mobile initiated handoff using fuzzy triangular numbers
based on Saaty’s scale.
The obtained weights for each parameter are shown in
Table 11.
5.1.3 User initiated handoff
During UIH, higher weight is assigned to users based
parameters, i.e., data rate, RSSI and network load [42].
Table 12 shows a fuzzy pairwise comparison matrix using
linguistic variables.
Table 13 shows the pairwise comparison matrix for user
initiated handoff.
Hence, the obtained weights for each parameter are
shown in Table 14.
5.2 Application of ELECTRE in HOA
HOA aids in selecting the network based on the network,
mobile and user preferences. Table 2has defined priority of
each parameter in NIH, MIH and UIH models. ELECTRE
is employed to rank the candidate networks in such a way
that if the reason of handoff is ‘low signal strength’, then
network with high signal strength is selected (NIH).If the
reason of handoff is ‘file size is high and bandwidth
available is low’, then network with high bandwidth
availability with low network load is preferred (MIH).
Reason of handoff initiation is considered utmost for
ranking and selecting the optimal network. Six key
parameters are considered for ranking such as signal
strength, bit error rate, network load, bandwidth available,
packet loss and battery power. Three alternative networks
assumed are WiFi, WiMax and LTE-A. The structure of
decision hierarchy is shown in Fig. 1.
5.2.1 Optimal network selection for NIH
In this case study, three alternative networks are compared
with respect to six parameters for selecting the optimal
Table 5 Pairwise comparison
matrix based on TFN for NIH Factor RSSI BER NL BW PL BP
RSSI (1,1,1) (2,3,4) (2,3,4) (4,5,6) (6,7,8) (9,9,9)
BER (1/4,1/3,1/2) (1,1,1) (1,1,1) (2,3,4) (4,5,6) (6,7,8)
NL (1/4,1/3,1/2) (1,1,1) (1,1,1) (2,3,4) (4,5,6) (6,7,8)
BW (1/6,1/5,1/4) (1/4,1/3,1/2) (1/4,1/3,1/2) (1,1,1) (2,3,4) (4,5,6)
PL (1/8,1/7,1/6) (1/6,1/5,1/4) (1/6,1/5,1/4) (1/4,1/3,1/2) (1,1,1) (2,3,4)
BP (1/9,1/9,1/9) (1/8,1/7,1/6) (1/8,1/7,1/6) (1/6,1/5,1/4) (1/4,1/3,1/2) (1,1,1)
Table 6 Calculating geometric mean
Factor GM
RSSI 3.0862 3.7621 4.3645
BER 1.5131 1.8086 2.1398
NL 1.5131 1.8086 2.1398
BW 0.6609 0.8327 1.0699
PL 0.3467 0.4228 0.5246
BP 0.2041 0.2308 0.2698
colsum 7.3241 8.8656 10.5084
inverse 0.136535547 0.112795524 0.095161966
Order 0.0952 0.1128 0.1365
Int. j. inf. tecnol.
123
network. Table 15 indicates the values of the three avail-
able networks for the respective parameters.
Table 16 is obtained using step 2 of Electre using
Eq. (9).
Table 17 is obtained using step 4 and Eq. (11). This
table gives the weighted, normalized matrix denoted by V.
Concordance and discordance matrices are obtained
using Eqs. (12) and (13) and values obtained are listed in
Tables 18 and 19.
Table 18 explains the superiority of one candidate net-
work over the other and Table 19 explains the inferiority of
one candidate network over the other. Table 20 gives the
ranking of the candidate networks on the basis of network
related parameters, i.e. better signal strength and it is seen
that the LTE-A is chosen by the ELECTRE as the optimal
alternative to provide good signal strength during the
decision making phase.
5.2.2 Optimal network selection for MIH
Attribute values as explained in Table 19 are used for
ranking by the networks on the basis of the mobile related
parameter. Table 21 is obtained using Eq. (9) and weights
Table 7 Calculation of
normalized weight Factor Weights Avgweight Average norm_wt
RSSI 0.29380624 0.42436488 0.59575425 1.31392537 0.43797512 0.419558
BER 0.14404712 0.20401008 0.2920827 0.6401399 0.21337997 0.204407
NL 0.14404712 0.20401008 0.2920827 0.6401399 0.21337997 0.204407
BW 0.06291768 0.09392856 0.14604135 0.30288759 0.10096253 0.096717
PL 0.03300584 0.04769184 0.0716079 0.15230558 0.05076853 0.048634
BP 0.01943032 0.02603424 0.0368277 0.08229226 0.02743075 0.026277
Total 3.1316906 1.0000
Table 8 Weights of each
parameter in NIH Factor Fuzzy weight
RSSI 0.419558
BER 0.204407
NL 0.204407
BW 0.096717
PL 0.048634
BP 0.026277
Table 9 Fuzzy pairwise comparison matrix using linguistic variables
for MIH
Factor BW RSSI NL PL BP BER
BW EI WI SI SI VSI EMI
RSSI EI WI WI SI VSI
NL EI EI WI SI
PL EI WI SI
BP EI WI
BER EI
Table 10 Pairwise comparison
on MIH Factor Band RSSI N/W Load PL Battery P BER
Band (1,1,1) (2,3,4) (4,5,6) (4,5,6) (6,7,8) (9,9,9)
RSSI (1/4,1/3,1/2) (1,1,1) (2,3,4) (2,3,4) (4,5,6) (6,7,8)
NL (1/6,1/5,1/4) (1/4,1/3,1/2) 1 1 (2,3,4) (4,5,6)
PL (1/6,1/5,1/4) (1/4,1/3,1/2) 1 1 (2,3,4) (4,5,6)
BP (1/8,1/7,1/6) (1/6,1/5,1/4) (1/4,1/3,1/2) (1/4,1/3,1/2) 1.00 (2,3,4)
BER (1/9,1/9,1/9) (1/8,1/7,1/6) (1/6,1/5,1/4) (1/6,1/5,1/4) (1/4,1/3,1/2) 1
Table 11 Normalized fuzzy weight vector for MIH
Factor BW RSSI NL PL
Factor weight 0.450736 0.244322 0.112271 0.112271
Table 12 Fuzzy pairwise comparison matrix using linguistic vari-
ables for UIH
Factor Data rate RSSI N/W load Battery P/ PL BER
DR/BW EI EI WI SI VSI EMI
RSSI EI WI SI VSI EMI
NL EI WI SI VSI
BP EI WI SI
PL EI WI
BER EI
Int. j. inf. tecnol.
123
Table 13 Pairwise comparison
on UIH Factor Data rate RSSI N/W Load Battery P PL BER
DR/BW (1,1,1) (1,1,1) (2,3,4) (4,5,6) (6,7,8) (9,9,9)
RSSI (1,1,1) (1,1,1) (2,3,4) (4,5,6) (6,7,8) (9,9,9)
NL (1/4,1/3,1/2) (1/4,1/3,1/2) (1,1,1) (2,3,4) (4,5,6) (6,7,8)
BP (1/6,1/5,1/4) (1/6,1/5,1/4) (1/4,1/3,1/2) (1,1,1) (2,3,4) (4,5,6)
PL (1/8,1/7,1/6) (1/8,1/7,1/6) (1/6,1/5,1/4) (1/4,1/3,1/2) (1,1,1) (2,3,4)
BER (1/9,1/9,1/9) (1/9,1/9,1/9) (1/8,1/7,1/6) (1/6,1/5,1/4) (1/4,1/3,1/2) (1,1,1)
Table 14 Normalized fuzzy
weight vector for UIH Factor DR/BW RSSI NL BP PL BER
Fuzzy weight 0.338379 0.338379 0.168902 0.085396 0.044547 0.024397
Fig. 1 Decision model of
network selection
Table 15 Attribute values for proposed scenario
Factor RSSI BER NL BW PL BP
Wifi -100 0.15 10 15 0.2 0.4
WiMax -80 0.2 15 20 0.4 0.3
LTE-A -50 0.25 20 10 0.6 0.5
Table 16 Weighted matrix RRSSI BER NL BW PL BP
Wifi -0.7274 0.4243 0.3714 0.5571 0.2673 0.5657
WiMax -0.5819 0.5657 0.5571 0.7428 0.5345 0.4243
LTE-A -0.3637 0.7071 0.7428 0.3714 0.8018 0.7071
Weights 0.4195 0.204407 0.204407 0.096717 0.048634 0.026277
Table 17 Weighted normalized matrix
VRSSI BER NL BW PL BP
Wifi -0.3051 0.0867 0.0759 0.0539 0.013 0.0149
WiMax -0.2441 0.1156 0.1139 0.0718 0.026 0.0111
LTE-A -0.1526 0.1445 0.1518 0.0359 0.039 0.0186
Table 18 Concordance matrix Concordance matrix (Con.)
0 0.0149 0.0539
0.0832 0 0.0718
0.2014 0.2014 0
Table 19 Discordance matrix Discordance matrix (Dis.)
0-0.0756 -0.1147
0.0111 0 0.0225
0.0359 0.0359 0
Int. j. inf. tecnol.
123
of MIH as calculated by FAHP in Table 11 are applied for
computing Table 21.
Table 22 explains the ranking of the networks and it is
seen that ranking is different on the basis of the mobile
related parameters.
5.2.3 Optimal network selection for UIH
In this case, ranking of networks is done by assuming
attribute values as given in Table 15. Table 23 is computed
using Eq. (9)
Table 24 gives the ranking of the networks on the basis
of user related parameter.
5.3 Performance analysis
Figure 2explains graphically weights assigned to six
parameters in NIH, MIH and UIH case using FAHP. It has
better performance and better decision making than the
recently proposed protocols in [43–46]. Figure 3gives the
ranking of networks in all three cases.
6 Conclusion and future work
In this paper, handoff decision making phase of Handoff
Optimization Algorithm for Next Generation Heteroge-
neous Networks has been proposed. Different set of
parameters has been discussed in different handoff initia-
tion categories and accordingly different weights have been
assigned to these sets of parameters in handoff decision
making phase. It has been concluded that RSSI can be an
important parameter for Network Initiated Handoff but for
MIH and UIH bandwidth is considered as a key factor. If a
handoff is initiated due to lack of bandwidth then selecting
network with good signal strength without seeing its
bandwidth availability is useless. In such a case bandwidth
and signal strength both must be given weight and then a
decision must be taken. The numerical results indicate that
the proposed HOA algorithm performs accordingly to the
network, mobile and user requirements thereby increasing
efficiency and end user satisfaction. In future, the algorithm
will be enhanced using machine learning and analytics for
predicting optimal network and mobility management. The
proposed algorithm will be experimented with including
input parameters such as direction and delay to optimize
the performance HOA.
Table 20 Final ranking of
networks Con Dis Rank
-0.2158 -0.2373 3
-0.0612 0.0734 2
0.277 0.164 1
Table 21 Weighted matrix
R RSSI BER NL BW PL BP
Wifi -0.727 0.424 0.371 0.557 0.267 0.566
WiMax -0.582 0.566 0.557 0.743 0.534 0.424
LTE-A -0.364 0.707 0.743 0.371 0.802 0.707
Weights 0.2443 0.0275 0.1123 0.4507 0.1122 0.0529
Table 22 Final ranking of
networks Con Dis Rank
-0.1912 -0.0975 2
0.4943 -0.2834 1
-0.3031 0.3808 3
Table 23 Weighted matrix
R RSSI BER NL BW PL BP
Wifi -0.727 0.424 0.371 0.557 0.267 0.566
WiMax -0.5820 0.566 0.557 0.743 0.534 0.424
LTE-A -0.364 0.707 0.743 0.371 0.802 0.707
Weights 0.3383 0.024 0.169 0.3384 0.0445 0.0853
Table 24 Final ranking of
networks Con Dis Rank
-0.0651 -0.2474 2
0.2734 -0.1458 1
-0.2084 0.3932 3
Fig. 2 Weights assigned using FAHP to NIH, MIH and UIH
Int. j. inf. tecnol.
123
References
1. Miller HJ (2007) Societies and cities in the age of instant access.
GeoJ Libr 88:365
2. Tongia R, Subrahmanian E, Arunachalam VS (2005) Information
and communications technology (ICT). In: Information and
Communications Technology for Sustainable Development
Defining a Global Research Agenda, pp 19–41
3. Gnanasambandam C, Gomes M (2012) Online and upcoming: the
internet’s impact on India. McKinsey Global Institute, London
4. Gani A, Nayeem GM, Shiraz M, Sookhak M, Khan S (2014) A
review on interworking and mobility techniques for seamless
connectivity in mobile cloud computing. J Netw Comput Appl
43:84–102
5. Khandekar A (2010) LTE-advanced: heterogeneous networks. In:
European Wireless Conference, IEEE, pp 978–982.
6. Damnjanovic A et al (2011) A survey on 3GPP heterogeneous
networks. IEEE Wirel Commun 18(3):10–21
7. Dhand P, Mittal S (2014) Handoff algorithms based on RSSI and
Fuzzy approach: a survey. Int J Appl Innov Eng Manag
3(2):206–216
8. Dhand P, Mittal S (2016) Smart handoff framework for next
generation heterogeneous networks in smart cities. In: Proceed-
ings of the international conference on advances in information
communication technology & computing, pp 1–7
9. Chiclana F, Herrera F, Herrera-Viedma E (1998) Integrating three
representation models in fuzzy multipurpose decision making
based on fuzzy preference relations. Fuzzy Sets Syst 97(1):33–48
10. Nguyen-Vuong Q-T, Agoulmine N, Cherkaoui EH, Toni L
(2013) Multicriteria optimization of access selection to improve
the quality of experience in heterogeneous wireless access net-
works. IEEE Trans Veh Technol 62(4):1785–1800
11. Liu F, Zhu S-F, Chai Z-Y, Qi Y-T, Wu J-S (2013) Immune
optimization algorithm for solving vertical handoff decision
problem in heterogeneous wireless network. Wirel Netw
19(4):507–516
12. Jailton J, Carvalho T, Valente W, Natalino C, Frances R, Dias K
(2013) A quality of experience handover architecture for
heterogeneous mobile wireless multimedia networks. IEEE
Commun Mag 51(6):152–159
13. Jain A, Tokekar S (2015) Application based vertical handoff
decision in heterogeneous network. Proced Comput Sci
57:782–788
14. Kassar M, Kervella B, Pujolle G (2008) An intelligent handover
management system for future generation wireless networks.
EURASIP J Wireless Commun Network 2008:791691. https://
doi.org/10.1155/2008/791691
15. Adrian M, Zen H, Othman AK (2014) A survey of vertical
handover decision algorithms in fourth generation heterogeneous
wireless networks. Asian J Inform Technol 13(4):247–251
16. Kumar R, Singh B (2010) Comparison of vertical handover
mechanisms using generic QOST rigger for. Int J Next-Gener
Netw (IJNGN) 2(3):80–97
17. Sgora A, Vergados DD (2009) Handoff prioritization and deci-
sion schemes in wireless cellular networks: a survey. IEEE
Commun Surv Tutor 11(4):57–77
18. Yang S (2008) A IEEE 802. 21 Handover design with QOS
provision across WLAN and WMAN. In: Communications,
Circuits and Systems, ICCCAS 2008. International Conference
on, IEEE, pp 548–552
19. Ovengalt CBT, Djouani K, Kurien A (2014) A fuzzy approach for
call admission control in LTE networks. Proced Comput Sci
32:237–244
20. Kaleem F (2012) VHITS: vertical handoff initiation and target
selection in a heterogeneous wireless network, Thesis, 2012.
21. Thumthawatworn T, Tillapart P, Santiprabhob P (2015)
Enhanced Fuzzy-based handover decision system design for
wireless mobile networks. In: International Conference on
Ubiquitous and Future Networks, ICUFN (2015), IEEE Computer
Society, pp 491–496
22. Goyal P, Lobiyal DK, Katti CP (2018) Dynamic user preference
based network selection for vertical handoff in heterogeneous
wireless networks. Wireless Pers Commun 98(1):725–742
23. Bellavista P, Corradi A, Foschini L (2007) Context-aware
handoff middleware for transparent service continuity in wireless
networks. Pervasive Mob Comput 3(4):439–466
24. Yilmaz ONC et al (2014) Smart mobility management for D2D
communications in 5G networks. In: 2014 IEEE Wireless Com-
munications and Networking Conference Workshops, WCNCW,
pp 219–223
25. Balasubramaniam S, Indulska J (2004) Vertical handover sup-
porting pervasive computing in future wireless networks. Comput
Commun 27(8):708–719
26. Ahmed T, Kyamakya K, Ludwig M (2006) Architecture of a
context-aware vertical handover decision model and its perfor-
mance analysis for GPRS-WiFi handover. In: Computers and
Communications, 2006. ISCC’06. Proceedings. 11th IEEE
Symposium, pp 795–801
27. Charilas DE, Markaki OI, Psarras J, Constantinou P (2009)
Application of Fuzzy AHP and ELECTRE to network selection.
Mobile Lightweight Wirel Syst 13:63–73
28. Mir U, Bhatti ZA (2018) Time triggered handoff schemes in
cognitive radio networks: a survey. J Netw Comput Appl
102:71–85
29. Tuysuz MF (2014) An energy-efficient QoS-based network
selection scheme over heterogeneous WLAN-3G networks.
Comput Netw 75:113–133
30. Dhand P, Mittal S (2015) Adaptive threshold and hysteresis for
handoff initiation in next generation networks from path loss
model. Int J Appl Res Inf Technol Comput 6:10
31. Herrera F, Herrera-Viedma E, Chiclana F (2001) Multiperson
decision-making based on multiplicative preference relations. Eur
J Oper Res 129(2):372–385
32. Saaty TL, Vargas LG (2012) Models, methods, concepts and
applications of the analytic hierarchy process. Springer Science
and Business Media, Berlin
33. Chan FTS, Kumar N (2007) Global supplier development con-
sidering risk factors using fuzzy extended AHP-based approach.
Omega 35(4):417–431
Fig. 3 Ranking of networks in all three cases
Int. j. inf. tecnol.
123
34. Van Laarhoven PJM, Pedrycz W (1983) A fuzzy extension of
Saaty’s priority theory. Fuzzy Sets Syst 11(1–3):229–241
35. Buckley JJ (1985) Fuzzy hierarchical analysis. Fuzzy Sets Syst
17(3):233–247
36. Chang D-Y (1996) Applications of the extent analysis method on
fuzzy AHP. Eur J Oper Res 95(3):649–655
37. Ayhan MB (2013) A Fuzzy AHP approach for supplier selection
problem: a case study in a Gearmotor company. Int J Manag
Value Supply Chains (IJMVSC) 4(3):11–23
38. Huang W-CC, Chen C-HH (2005) Using the ELECTRE II
method to apply and analyze the differentiation theory. Proc East
Asia Soc Transp Stud 5:2237–2249
39. Bari F, Leung V (2007) Application of ELECTRE to network
selection in a hetereogeneous wireless network environment. In:
IEEE Wireless Communications and Networking Conference,
WCNC, pp 3813–3818
40. Figueira J, Mousseau V, Roy B (2005) ELECTRE methods.
Multiple criteria decision analysis: state of the art surveys.
Springer, Berlin, pp 133–153
41. Pang J, Zhang G, Chen G (2011) ELECTRE I decision model of
reliability design scheme for computer numerical control
machine. J Softw 6(5):894–900
42. Sun Y, Feng G, Qin S, Liang YC, Yum TSP (2018) The SMART
handoff policy for millimeter wave heterogeneous cellular net-
works. IEEE Trans Mob Comput 17(6):1456–1468
43. Litake SR, Mukherji P (2020) Integration of Fuzzy logic and
ABC algorithm for optimized network selection in heterogeneous
wireless environment. Int J Sens Wirel Commun Control
10(2):248–261
44. Baghla S, Bansal S (2020) An adaptive algorithm for energy
efficient network selection in heterogeneous networks. Int J Syst
Control Commun 11(3):227–241
45. Chowdhury MZ, Rahman MT, Jang YM (2020) An analytical
hierarchy process combined with game theory for interface
selection in 5G heterogeneous networks. KSII Trans Internet
Inform Syst (TIIS) 14(4):1817–1836
46. Thakur GK, Priya B, Sharma PK (2020) Optimal selection of
network in heterogeneous environment based on fuzzy approach.
J Math Comput Sci 10(3):554–571
Int. j. inf. tecnol.
123