Content uploaded by Ayadi Mohamed
Author content
All content in this area was uploaded by Ayadi Mohamed on Oct 09, 2015
Content may be subject to copyright.
Fuzzy MADM Based Vertical Handover Algorithm
for Enhancing Network Performances
Aymen Ben Zineb
Higher School of Communications
(Sup’Com) Tunis-Tunisia
aymen.benzineb@supcom.tn
Mohamed Ayadi
Higher School of Communications
(Sup’Com) Tunis-Tunisia
m.ayadi@supcom.rnu.tn
Sami Tabbane
Higher School of Communications
(Sup’Com) Tunis-Tunisia
sami.tabbane@supcom.rnu.tn
Abstract— One of the most challenging topics for next
generation wireless networks is vertical handoff concept since
several wireless technologies are assumed to cooperate in a single
heterogeneous environment. Several parameters must be
considered in vertical handoff process along with traditional
received signal strength (RSS) information, they are related to
user preferences, terminal capabilities and QoS, such as data rate,
tolerated delay. In this context, Classical Multiple Attribute
Decision Making (MADM) methods, like the Technique for Order
Preference by Similarity to Ideal Solution (TOPSIS) were used to
make suitable handover decisions. Our approach called Fuzzy
TOPSIS, is based on the combination of TOPSIS method with
fuzzy logic in order to reduce the decision delays, blocking
probability and the number of executed handoffs. Performance
results of the proposed system are also compared with those of the
classical TOPSIS method.
The proposed algorithm is able to determine properly whether
a handoff is necessary or not, and selects the best candidate radio
access technology (RAT) considering the input parameters and
fuzzy inference rules. Simulation results show an improvement of
decision time by 40 % comparing to classical approach.
Keywords—Vertical handoff; MADM; TOPSIS; Fuzzy Logic;
Handoff delay; Complexity analysis
I. INTRODUCTION
In recent years, increasing demand of wireless services has
forced diverse technologies to cooperate since mobile users
desire to be able to get both real-time and non-real time services
with minimum cost and optimum quality. The next generation
networks technologies will coexist altogether in a heterogeneous
environment that aims to improve network performances, and
the guarantee of high Quality of Service (QoS) and ubiquitous
broadband access. Mobile terminals deployed in such a
heterogeneous network structure need to be aware of all the
networks around and to switch its connection to the appropriate
one. Accordingly, cognitive radio that is capable of connecting
different wireless technologies with perception and adaptation
attributes seems to be a remedy for these requirements.
In order to provide seamless mobility between various
wireless technologies overlapped, vertical handoff algorithms
need to be developed for next generation wireless structures.
Several interworking mechanisms have been proposed, they are
based on the optimization of used radio resources and enabling
the best technology for each new or current connection. The
decision for a vertical handover is complicated since it depends
on an increasing number of criteria, MADM has been proposed
to solve this problem.
In literature, many MADM algorithms can be found, one of
the most important is the Technique for Order Preference by
Similarity to Ideal Solution (TOPSIS)[1]. It have been widely
implemented in vertical handover decision since it presents
advantages of simplicity and finding the optimal RAT in real
time conditions [2, 3]. However, with the increase of RATs
numbers, decisional criteria and QoS parameters, this solution
becomes inefficient since decisional time will be important and
not convenient for real time services.
Moreover, computational-intelligence based algorithms [4,
5] are very performing for decision-making thanks to the use of
artificial intelligence and prior existence of human knowledge
about the solution. Among existing intelligent methods, we can
find fuzzy logic that can be considered as a white box model
applied to deal with imprecise information of radio link and user
preferences. Fuzzy logic uses an inference rules to get output
fuzzy decisions function of system inputs like handover criteria
In this study, we propose a fuzzy TOPSIS based vertical
handoff decision approach that combine advantages of both
MADM and Fuzzy logic methods. The goal is to have low delays
decisions, so that seamless connection can be made between
different wireless technologies consistently in real time.
The developed algorithm is called Fuzzy-TOPSIS algorithm.
It will exploit the advantage of fuzzy logic for intelligent fuzzy
decisions when imprecise information is provided as input, and
MADM methods to deal with the increasing number of handover
criteria and available networks. Performances of the new
algorithm are measured and compared to original TOPSIS
method in terms of decision delays during a handover process.
The remainder of this paper is organized as follows: Section
II gives a state of art about existing MADM TOPSIS method and
its performances. Section III defines the fuzzy logic algorithm
and the building of fuzzy controllers. In section IV, we present
simulation of fuzzy algorithms according to TOPSIS. In fact, we
iterate performances of our proposed model for handover
decision process, class of service weight parameter and network
selection process and decision time. The last section will
conclude our work with some observations and comments.
SS2 - 65783 - 1609 © SoftCOM 2015
II. CLASSICAL MADM (TOPSIS) APPROACH
Many classical MADM methods exists in literature, one of
the most known is TOPSIS (Technique for Order Preference by
Similarity to Ideal Solution). It was first developed by Hwang
and Yoon basing on the idea that the chosen alternative should
have the shortest distance from the Positive Ideal Solution (PIS),
and on the other side the farthest distance of the Negative Ideal
Solution (NIS). The Positive Ideal Solution maximizes the
benefit criteria and minimizes the cost criteria. Whereas the
Negative Ideal Solution maximizes the cost criteria and
minimizes the benefit criteria [1,2].
In TOPSIS, the selected decision is the closest to the ideal
solution (and the farthest from the worst solution). The ideal
solution is obtained by using the highest values of given
by:
(1)
and denote the separation measurements of each
network for the positive and negative ideal solutions
respectively given by:
(2)
(3)
Where:
denotes context criteria j of candidate network i given by:
(4)
Wj is the weight vector
best is the maximum value of for a benefit criteria or
a minimum value of for a cost criteria
worst is the minimum value of for a benefit criteria or
a maximum value of for a cost criteria
In our case, with TOPSIS method the problem returns to
calculate the network score () and network having
highest value will be the one considered. Inputs are context
criteria and simulations will be realized for i networks with j
criteria.
III. FUZZY TOPSIS APPROACH
In our work, we propose as vertical handoff decision a fuzzy
TOPSIS method composed of two parts: The first part is to
convert the fuzzy data issue of different inputs of the 4 fuzzy
controllers. These inputs are: the received signal from different
RATs, QoS criteria (delay and data rate) requested by the user
application, mobile speed and mobile battery level. The second
part is to use the calculation of TOPSIS which inputs are the
output of the fuzzy controllers, to determine the ranking score of
the candidate networks. The network selected will be the one
with the highest score. In our developed approach, we will
consider cellular radio networks (3G/4G) and wireless networks
(Wi-Fi/ WiMAX) coexisting in a same environment.
A. System description
The proposed fuzzy algorithm uses a set of fuzzy-logic
controllers based on the principle of a Mamdani system.
Proposed in 1975 by Mamdani, this system consists of four main
parts: fuzzifier, inference engine, fuzzy rules and defuzzifier. [6]
The fuzzifier allocates a value for each input corresponding
to the degree of membership of this input. Fuzzy sets are
represented by a mathematical function called membership
function, they relate fuzzy sets to 3 linguistic variables that
model the problem that are (“Low, Medium and High”).The
inference engine will use fuzzy rules to indicate the suitable one
from available RATs to be selected. Each output is a set of four
triangular membership functions that represent the four
linguistic variables (“Not acceptable NA”, “Probably not
acceptable PNA”, “Probably acceptable PA” and “Acceptable
AC”). Linguistic variables are used to establish the fuzzy
inference rules set resuming human knowledge.
Each controller has N outputs that describe the likelihood of
user to be assigned to one of the N candidates networks. For
defuzzification, method of centroids will be used. It represents
most accurate method and it did not neglect the forms of output
membership functions. In addition, it does not suffer from the
problem of ambiguity because there are precisely defined
conditions [7].
Fig. 1. Algorithm diagram of fuzzy approach
The decisional block is composed of a combination of a
fuzzy block and an MADM method. The fuzzy block is
composed of a set of fuzzy controllers, their role is to make
fuzzification of input variables in terms of criteria analysis. Each
SS2 - 65783 - 1609 © SoftCOM 2015
of the fuzzy controllers presents constructional unit of the joint
parallel fuzzy system whose output values are inputs to the next
step. The outputs of the different fuzzy controllers serve as
inputs of the TOPSIS method with different weightings
calculated by AHP method. A structural diagram of the overall
approach is shown in figure 1.
The system first monitors indicating parameters and
measures the RSS of serving network. If it goes below a given
threshold then the algorithm enables the 4 fuzzy controllers.
Each controller calculates the score of every candidate RAT. An
overall score is calculated using the AHP weights vector to give
the final score. The network that has the highest score is then
selected as given by (1).
B. Fuzzy Controllers
The fuzzy block consists of four fuzzy controllers: fuzzy
controller that refers to the received signal strength, fuzzy
controller that refers to the speed of mobile station, fuzzy
controller related to battery level of mobile user and a fuzzy
controller related to the demanded quality of service (QoS) by
the user. The following section details the design and building
of the fuzzy controllers for RAT selection between 4G and
WiMAX technologies context using MATLAB fuzzy toolbox.
1) Received signal strength (RSS) controller
This system controls the variation of the signal strength
received by the mobile terminal and then allocates users to the
appropriate technology.
It takes as a first input the signal strength that refers to the
4G network (SS-4G), and secondly the signal strength that refers
to WiMAX (SS- WiMAX), both received by the terminal. The
output resulting from this system decides which network the
user should be assigned to.
The range of values consists of all possible values according
to guidelines and operational instructions of existing networks
(LTE) and (IEEE 802.11/802.16) [8]. Each value range is
described by three linguistic variables, such as “Low”,
“Medium” and “High”. Linguistic variable "low" implies poor
reception of radio signal, the variable "high" implies a strong
reception level, while the variable "medium" indicates
intermediate signal level. The shape of the membership
functions in the space of values is presented in Figure 2.
Fig. 2. Membership functions of RSS controller inputs
Fuzzy-logic rules are established to assign the user to a
network that has the best radio performance and better strength
of received signal.
2) QoS Controller
As the QoS is a tradeoff among different parameters,
maintaining the QoS for different kinds of applications is a
critical task in VHO decision making.
Fuzzy controller related to the quality of service required by
user takes into consideration tolerated delay and data-rate
variables, which are required by the service for its proper
functioning. QoS criteria reflects the requirements of services
and applications within the algorithm for access network
selection. QoS fuzzy controller has two inputs, requirement for
and data-rate (D-R) and tolerated delay (T-D) as independent
inputs. In addition, two outputs relating to the index values of
access networks considered in the system (4G and WiMAX in
our case), as shown in Figure 3.
Fig. 3. Membership functions of QoS controller inputs
Values of the two input variables are determined taking into
account the most sensitive services for each variable. Every
value is defined by three linguistic variables (“High”, “Medium”
and “Low”). Value range of tolerated delays starts with values
between 0 and 50ms, which is typical for conventional voice
services in real time, and ends with a delay of more than 200ms,
which is appropriate for background non-real time services.
Value range for the data rate starts with a request for bit rate of
less than 500kb/s, which is typical for services with low bit rate
and ends with the bit rate of more than 7Mb/s typical of packet
based traffic with high speed data transfer or video traffic with
high resolution (for a mobile terminal). Four QoS classes are
defined, each one of them has its restrictions in terms of required
bandwidth and delay for the satisfaction of offered service type
offered by the concerned class, some examples are shown in
table I.
TABLE I. QOS CONTROLLER FUZZY INFERENCE RULES
Rule N°
D-R (Mb/s)
T-D (ms)
S(4G)
S(WiMAX)
1
Low
High
PA
AC
4
Medium
High
PNA
PA
6
Medium
Low
PA
PNA
8
High
Medium
PA
PNA
9
High
Low
AC
PNA
For every QoS class we have a different set of fuzzy rules
and so a different configuration of the controller. Fuzzy system
consists of 9 fuzzy rules. Each rule is set in order to satisfy the
application and service requirements of the system for assigning
stages.
3) Battery level Controller
Battery level is an important criterion especially when the
user is in a public space and does not have access to energy
source. WiMAX connection does not consume much power as
4G connection. We note that when battery level of the mobile
terminal is low it is more likely to connect to WiMAX networks.
-100 -90 -80 -70 -60 -50
0
0.2
0.4
0.6
0.8
1
SL-WiMAX
Degree of membership
Low Medium High
-100 -90 -80 -7 0 -60
0
0.2
0.4
0.6
0.8
1
SL-4G
Degree of membership
Low Medium High
0 5 10 15 20
0
0.2
0.4
0.6
0.8
1
D-R.(Mb/s)
Degree of membership
Low Medium High
020 40 60 80 100
0
0.2
0.4
0.6
0.8
1
T-D.(ms)
Degree of membership
Low Medium High
SS2 - 65783 - 1609 © SoftCOM 2015
When battery level is high, 4G connections are more preferred
since they are much energy consuming.
4) Mobile speed Controller
Connecting slow moving users to access networks with less
coverage, such as WiMAX networks, and users moving with
higher speeds in networks with broad coverage, reduces the
number of unnecessary handovers and conserves radio resources
[9]. Hence, this controller takes decision based on the speed of
the mobile user choosing a radio access technology. The speed
(m/s) has 3 membership functions: low, medium and high. On
the other hand, the output is the same as other fuzzy controllers.
C. Decision with Fuzzy approach
Normal approach consists on the use of classical TOPSIS
method for the decisions. We defined 5 inputs which are: RSS,
delay, bandwidth, speed and battery level. For each input, we
calculate the corresponding weight value according to the target
QoS class. We note that weights are calculated according to
AHP method. Their value will change only for delay and
bandwidth such they are QoS parameters for the rest of inputs
weights are independent from QoS class. We can note that in
case of adding new QoS criteria or other user profile or network
indicators, a new configuration is needed. In fact, for every new
added parameter, a weight should be assigned to and new
TOPSIS matrix is needed. This operation will add complexity to
network deployments and even a more decision delay caused by
computational operations. Performances of this approach will be
compared with those of our developed fuzzy approach.
The fuzzy approach consists on the use of fuzzy controllers.
In fact, handover criteria are grouped in 4 different controllers
as explained before. The QoS controller contain the 2 QoS
criteria, namely delay and data rate. So, the same inputs in the
normal approach will be considered as inputs of the fuzzy
approach and fuzzy controllers. Outputs of fuzzy controllers are
inputs of TOPSIS method and the classification function will be
composed of the outputs of the 4 fuzzy controllers with their
associated weights, determined by AHP method.
The decision matrix for each method will always have a
fixed size and fixed weights values, even when adding new
criteria. In fact, adding new parameters will consists on adding
an inference rule in the corresponding fuzzy controller. This can
reduce complexity and keep always low delays for decisions,
especially when increasing the number of handover criteria or
even the number of RATs. Simulation is realized to prove fuzzy
approach performances comparing to normal approach.
D. AHP weighting method
Analytic Hierarchy Process (AHP) [10-11] is a popular and
a widely used method to resolve decision situation problem
when multi-criteria are involved. The procedure defined by AHP
consists of building a hierarchy model based on context criteria.
To make pair-wise comparison we need to judge context criteria
two by two and indicate how many times a criterion is more
important than the other one. Weight priorities are established
by calculating the principal eigenvector of the pair-wise
comparison matrix as given by:
(5)
With A is the comparison matrix, is the largest
eigenvalue of A and W is the corresponding eigenvector. The
eigenvector are then normalized to become the priorities vector.
For every QoS class we define an AHP weighting vector:
(6)
In our case, 5 different criteria are chosen and they are
ranked according to their impact on a handover process, 2 of
them are related to QoS. For the different QoS classes these
criteria can have different impact according to service requested
by the user and network conditions. Therefore, for every QoS
class we define an AHP input ranking matrix as shown in table
below. The following tables summarizes the AHP input ranking
matrix for the streaming QoS class. The following matrices
summarizes AHP pair-wise comparison respectively for the
classical MADM (AM) and the Fuzzy-MADM (AF) methods
according to the streaming QoS class.
(7)
(8)
Vectors () and () show results and weights values for
the 2 approaches: fuzzy approach with 4 inputs and normal
approach with 5 inputs. Note that, for this last one we will have
4 different weights vectors referring each one to a QoS class.
Table II explains the difference between the 2 weight vectors.
TABLE II. WEIGHTS TABLE FOR STREAMING QOS CLASS
RSS
QoS
Speed
Batt-lev
B-W
Delay
Vector 1 (5 inputs)
0.4691
0.201
0.201
0.0862
0.0427
Vector 2 (4 inputs)
0.5650
0.2622
0.1175
0.0553
(9)
(10)
IV. SIMULATION RESULTS
In this section, we present simulation results and
performance comparison of two MADM methods: TOPSIS and
Fuzzy-TOPSIS. We have considered two access networks
detected by a mobile user: WiMAX (Worldwide Interoperability
for Microwave Access) and LTE (Long Term Evolution)
networks. Available features for these networks are qualities of
context criteria that are measured to give an indication of
whether or not a handover is needed. During the simulation,
measures of every context criteria for candidate network are
randomly varied using Matlab/Simulink and Network
Simulalator NS2, according to parameters ranges defined by
each network as shown in Table V.
SS2 - 65783 - 1609 © SoftCOM 2015
(a) (b)
Fig. 4. Results for QoS class (a) Conversational (b) Background
A. Handoff decision
To evaluate fuzzy approach we will see its performances
when taking decision according to real inputs during a
simulation process. For every QoS class we will compare
decisions of the 2 approaches and see if fuzzy approach gives
same roaming decisions results as classical approach. In fact, the
aim of this test consist in proving that our approach do not give
erroneous handover results and the choice of the target network
might be the same as classical approach that was validated in
previous works. During simulation, random input values are
used from RSS, delay and bandwidth may vary for each
technology. Speed of mobile and battery level are chosen in
manner that a user is considered in normal speed with 60% of its
battery. The mobile user required context can be varied form a
given class of service to another. For each QoS class we
calculated the ranking function using classical TOPSIS
considering 5 inputs, Fuzzy-TOPSIS using the 4 fuzzy
controllers.
From different sources in literature [12], we determined
parameter ranges, especially QoS inputs corresponding to each
used technique. Both HSPA (3G) and LTE (4G) are 3GPP
technologies, WiFi and WiMAX are IEEE technologies. For
each technology many versions and upgrades of norms exists.
We have considered in table III most general values.
TABLE III. QOS PARAMETERS RANGES
Technology
Throughput
(Mb/s)
Data Rate
(Mb/s)
Delay (ms)
RSS
(dbm)
Jitter
(ms)
Latency
(ms)
HSPA (3G)
0.1 - 14
0.1 - 2
50 - 80
5 - 10
25 - 50
LTE (4G)
1 - 100
31.4 - 100
25 - 100
-105 / -50
3 - 10
50 - 80
WiFi
1 - 54
5.9 - 30.5
100 - 150
10 - 20
100 - 150
WiMAX
1 – 60
20 - 60
60 - 100
-105 / -50
3 - 10
60 – 100
Figure 4 and table IV gives an overview about the decisions
made by each method for every QoS class. Numerical results are
shown in table below, the RAT having the highest ranking value
among the RATs candidates, is the one chosen.
TABLE IV. RANKING RATS SIMULATION RESULTS
F-TOPSIS
TOPSIS
Conversational
4G
0.6078
0.5812
WiMAX
0.3922
0.4978
Streaming
4G
0.6790
0.6085
WiMAX
0.3210
0.4258
Interactive
4G
0.6790
0.6238
WiMAX
0.3210
0.3896
Background
4G
0.4110
0.8794
WiMAX
0.5890
0.8995
We can notice that the technology chosen is always the same
for both classical and fuzzy TOPSIS approaches. This means
that our approach do not take wrong decisions and it is as
performant as classical methods. We can also see that the
decision depends on the QoS class. For conversational class:
methods choose 4G network, but for background class they
choose WiMAX network. This shows that methods are
optimizing the use of radio resources.
B. Handoff fuzzy approach evaluation
The decision delay parameter is essential especially for real
time handover since it has a direct influence on QoS. In fact a
later decision can causes QoS degradation, an early handover
decision can be judged not useful and causes an over signaling.
In this part of simulation, we are interesting to evaluate the
decision time of each method to elaborate the handover decision
when increasing the number of inputs or the number of detected
networks. In addition, we will compare the handoff blocking
probability between the two methods.
We fixed the number of detected networks and varied the
number of inputs. More inputs were randomly added (Jitter,
Latency, User class…). Two possible technologies are available
(4G and WiMAX). For fuzzy approach each input is considered
as an independent controller like the (user class), so a new
controller is added. However, in some other cases, some inputs
can be grouped into one controller as for QoS controller (Jitter,
4G WLAN
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Availables RATs
Ranking Score
Fuzzy-Topsis
TOPSIS
4G WLAN
0
0.2
0.4
0.6
0.8
1
Available RATs
Ranking Score
Fuzzy-Topsis
TOPSIS
SS2 - 65783 - 1609 © SoftCOM 2015
Latency…) which can decrease the complexity and as a
consequence, the decision time. However, for classical approach
for every new input added a new configuration is needed for
weighting AHP matrix and decision calculations.
Fig. 5. Decision delay for different input criteria
From figure 5, we can depict that, when increasing the
number of inputs, the decision delay by TOPSIS method
increases when number of inputs increase to reach 0.07s (x7
times) for 15 inputs, however for Fuzzy TOPSIS, it still
relatively constant (about 0.01) for 15 inputs (increases only by
10%). Therefore, when the number of inputs increases the
TOPSIS algorithm takes much time because of the increase in
matrices size and complex calculations. By using fuzzy
controllers, we reduced complexity and so the decision delay
since when increasing inputs we add new inputs for the
controllers with keeping the same number of controllers without
increasing matrices size.
In the second scenario, we defined 5 inputs (received signal,
delay, data rate, mobile speed and battery level) and we varied
the number of detected networks (RATs), to see the delay for
each method to take a handover decision. As shown in figure 6
TOPSIS method which has the highest decision delay for 6
available RATs (about 2.5x10-2), however Fuzzy TOPSIS delay
is about (1.3x 10-2). Which means ½ of delay of Topsis method.
Fig. 6. Decision delay for different RATs
Classical TOPSIS method have higher values of decision
time, since they need to compute new weight matrices for every
available network. However, Fuzzy-TOPSIS can be considered
as low complex method because of the use of knowledge a priori
and inference rules where its interest especially in real time
scenarios.
The proposed fuzzy TOPSIS algorithm is then compared to
classical TOPSIS algorithm in term of handoff blocking
probability. Handoff blocking probability is one of the main
metrics in handoff procedure. If a handoff algorithm triggers
handoff frequently, the handoff block probability will increase.
In addition, the handoff blocking probability increases when the
number of users increases as can be seen in figure 7. The handoff
blocking probability of proposed algorithm is lower than the
classical method.
Fig. 7. Handoff blocking probability
Fig. 8. Number of handovers
The last part of simulation is a comparison between TOPSIS,
Fuzzy-TOPSIS and RSS based handover decision algorithms.
The number of executed handovers is an important parameter,
since a high number indicates instability in the network and
possible Ping-Pong effect. That is why it is desired to have
minimum number of handoff while the requirements of user,
terminal and network are fulfilled.
Figure 8 shows that by using an MADM methods (TOPSIS,
Fuzzy-TOPSIS), the number of handovers is reduced
considerably when compared to RSS based algorithm. The
combination TOPSIS with fuzzy logic adopted by our approach
(Fuzzy-TOPSIS), has the lowest number of handovers thanks to
the inclusion of more criteria and the introduction of fuzzy
knowledge in decision process. We can deduce from these
results, consequently, that artificial intelligence fuzzy logic use
is much more appropriate for vertical handoff decision.
510 15
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
Number of Inputs
Decision Delay (s)
Fuzzy-TOPSIS
TOPSIS
2 3 4 5 6
0
0.005
0.01
0.015
0.02
0.025
0.03
Available RATs
Decision Delay (s)
Fuzzy-TOPSIS
TOPSIS
020 40 60 80 100
0
10
20
30
40
50
60
Number of Users
Handoff blocking probability (%)
TOPSIS
Fuzzy TOPSIS
0
2
4
6
8
10
TOPSIS Fuzzy-TOPSIS RSS based
Number of executed handovers
SS2 - 65783 - 1609 © SoftCOM 2015
V. CONCLUSION
In this work, we proposed a fuzzy logic based vertical
handoff decision algorithm that is capable of switching between
different radio access technologies. According to obtained
simulation results, the proposed vertical handoff decision
algorithm is able to determine the select the best candidate
access network in lower delay with less complexity.
It is also observed that, artificial intelligence based vertical
handoff algorithms as well as newly proposed approach,
noticeably reduce the number of handoff compared to classical
algorithm. Together with the number of handoff, handoff
latency, algorithm complexity, and handoff blocking probability
are another important metrics that affect the performance of the
system.
As future work, we plan to extend the MADM algorithms
furthermore with taking into account the priority information of
context metrics and defining a framework for QoE management
including dynamic weighting for best handover decision.
REFERENCES
[1] K. Savitha and Dr. C. Chandrasekar, “Vertical Handover decision
schemes using SAW and WPM for Network selection in Heterogeneous
Wireless Networks”, Global Journal of Computer Science and
Technology Volume 11 Issue 9 Version 1.0 pp. 19-24, May 2011.
[2] Wang, Y-J and Lee, H-S, “Generalizing TOPSIS for fuzzy multiple-
criteria group decision-making”, Computers & Mathematics with
Applications, 53(11), 1762-1772, 2007
[3] K. Savitha and Dr. C. Chandrasekar, “Vertical Handover decision
schemes using SAW and WPM for Network selection in Heterogeneous
Wireless Networks”, Global Journal of Computer Science and
Technology Volume 11 Issue 9 Version 1.0 pp. 19-24, May 2011.
[4] L. Giupponi, R. Agust´ı, J. P´erez-Romero, and O. Sallent, “A novel joint
radio resource management approach with reinforcement learning
mechanisms,” in Proceedings of the24th IEEE International Performance,
Computing, and Communications Conference (IPCCC ’05), pp. 621–626,
Phoenix, Ariz, USA, April 2005.
[5] R. Agust´ı, O. Sallent, J. P´erez-Romero, and L. Giupponi, “A fuzzy
neural based approach for joint radio resource management in a B3G
framework,” in Proceedings of the 1stInternational Conference on Quality
of Service in HeterogeneousWired/Wireless Networks (QSHINE ’04), pp.
216–224, Dallas, Tex, USA, October 2004.
[6] J. Martin, A. Eltawil, “Towards a Unified Wireless Network Involving
Reconfigurable Devices”, Clemson University, May 2010.
[7] Aleksandar Tudzarov and Toni Janevski, “Efficient Radio Access
Technology Selection for Next Generation Wireless Networks”,
International Journal of Research and Reviews in Next Generation
Networks, March 2011
[8] IEEE Std 802.16m-2011, IEEE Standard for Local and metropolitan area
networks, Part 16: Air Interface for Broadband Wireless Access Systems,
Amendment 3: Advanced Air Interface, May 2011.
[9] LI Ming-ju, XU Ning, WANG Ying and ZHANG Ping, Handover with
resource pre-allocation in the OFDMA systems, The Journal of China
Universities of Posts and Telecommunications, Volume 14, Issue 4 pp.
27-31, December 2007.
[10] T.L. Saaty, How to make a decision: The Analytic Hierarchy Process,
European Journal of Operational Research 48, pp. 9-26, North- Holland
1990.
[11] T.L. Saaty, Decision making with the analytic hierarchy process,
International Journal of Services Sciences (IJSSCI), Vol. 1, No. 1, pp. 88-
98, 2008.
[12] ITU-T G.1010
[13] Sassi Maaloul, Afif Meriem and Sami Tabbane, “Context Awareness and
Class of Service Satisfaction for Modeling Handover Decision-Making”,
International Journal of Computer Applications Volume 47– No.20, June
2012
[14] Aymen, B. Z., Ayadi, M., & Tabbane, S. (2014, June). A fuzzy logic
algorithm for RATs selection procedures. In Networks, Computers and
Communications, The 2014 International Symposium on (pp. 1-5). IEEE.
SS2 - 65783 - 1609 © SoftCOM 2015
