5
A Novel Fuzzy Logic Based Vertical Handoff Decision Algorithm for Heterogeneous Wireless Networks Peng Yang, Yong Sun, Chao Liu, Wei Li, Xiangming Wen Beijing Key Laboratory of Network System Architecture and Convergence Beijing University of Posts and Telecommunications, Beijing, P.R. China [email protected] Abstract-The next generation wireless communication systems will combine multiple wireless access networks, thus can provide mobile users with the best services. In these heterogeneous wireless networks, vertical handoff decision plays an important role to guarantee Always Best Connected (ABC) services and seamless mobility for mobile terminals. In this paper, we first introduce a speed-adaptive system discovery scheme before vertical handoff decision, which effectively improves the update rate of the candidate networks set. Then we propose a vertical handoff decision algorithm based on fuzzy logic with a pre-handoff decision method. Compared with the traditional RSS based handoff algorithm and sole fuzzy logic based handoff algorithm, the simulation results show that the proposed algorithm's performance is enhanced by reducing unnecessary handoffs, balancing the whole network resources and decreasing the probability of call blocking and dropping. Keywords-Vertical hando Fuzzy logic; Pre-hando//; Network selection; Heterogeneous wireless network. I. INTRODUCTION With the rapid development and deployment of wireless technologies, the future wireless networks are envisioned as a combination of multiple wireless access networks like WIFI, WIMAX and Universal Mobile Telecommunications System (UMTS), which can provide mobile users with Always Best Connected (ABC) services [1]. However, there are still some technical challenges to be further researched, handoff decision making is one of the most challenging problem in heterogeneous wireless networks since various parameters should be considered when executing handoff Handoff (HO) is the process of changing the mobile connection between different base stations or access points. In heterogeneous wireless networks, handoff can be divided into two categories: horizontal handoff (HHO) and vertical handoff (VHO) [2]. A HHO is made among different access networks when changing a connection from one access point () to another or one base station (BS) to another, that is to say, the networks have the same link-layer technology. While a vertical handoff happens between access networks with different Iink- layer technologies, such as changing a connection between an AP and a BS. Compared with the horizontal handoff, the vertical handoff is more valuable for research. In general, the VHO process involves three main phases [3]: 1) system discovery; 2) handoff decision; 3) handoff execution. During the system discovery phase, mobile terminals equipped with multiple interfaces have to determine the available candidate networks and the available services. In the second phase, the mobile device has to decide whether the connections should continue using current selected network or be switched to another network. In the handoff execution phase, the connections need to be rerouted om current network to new network, with authentication, authorization and the transfer of context information. In the vertical handoff procedure, the handoff decision phase is the most important phase because an inappropriate handoff decision may break off the current communication and degrade the QoS of traffic [3]. Thus this paper focuses on the vertical handoff decision phase and proposes a novel fuzzy logic based vertical handoff decision algorithm. Fuzzy Logic (FL) as a Multiple Attribute Decision Making (MADM) method is not only to evaluate and combine many parameters at the same time but also to deal with some imprecise information. The contributions of this paper can be summarized as follows. 1) The paper introduces a speed-adaptive system discovery scheme before handoff decision, which effectively improves the update rate of the candidate networks set. 2) We present a pre-handoff decision method which can eliminate the serious Ping-Pong effect and decrease the probability of call blocking and dropping. 3) The paper proposes a novel vertical handoff decision algorithm based on fuzzy logic. And the better performance of our proposed algorithm is validated via simulations. The rest of this paper is organized as follows: we briefly review the related works about vertical handoff decision algorithms in Section II. In Section III, a novel zzy logic based vertical handoff decision algorithm is proposed. Then Section IV shows the performance evaluation of our proposed algorithm. Finally, conclusions are given in Section V. II. RELATED WORK In this section, the paper simply reviews some recent works on vertical handoff decision algorithm. Aſter analyzing, we know that received signal strength (RSS) is usually used in traditional vertical handoff algorithms, and some subsequent RSS based handoff algorithms further consider hysteresis and ISSN:1882-5621/13/ ©2013 NICT 1

An efficient vertical handoff mechanism for future mobile network

Embed Size (px)

Citation preview

Page 1: An efficient vertical handoff mechanism for  future mobile network

A Novel Fuzzy Logic Based Vertical Handoff Decision Algorithm for Heterogeneous Wireless

Networks

Peng Yang, Yong Sun, Chao Liu, Wei Li, Xiangming Wen Beijing Key Laboratory of Network System Architecture and Convergence Beijing University of Posts and Telecommunications, Beijing, P.R. China

[email protected]

Abstract-The next generation wireless communication

systems will combine multiple wireless access networks, thus can

provide mobile users with the best services. In these

heterogeneous wireless networks, vertical handoff decision plays

an important role to guarantee Always Best Connected (ABC)

services and seamless mobility for mobile terminals. In this paper,

we first introduce a speed-adaptive system discovery scheme

before vertical handoff decision, which effectively improves the

update rate of the candidate networks set. Then we propose a

vertical handoff decision algorithm based on fuzzy logic with a

pre-handoff decision method. Compared with the traditional RSS based handoff algorithm and sole fuzzy logic based handoff

algorithm, the simulation results show that the proposed

algorithm's performance is enhanced by reducing unnecessary

handoffs, balancing the whole network resources and decreasing

the probability of call blocking and dropping.

Keywords-Vertical handofJ; Fuzzy logic; Pre-hando//;

Network selection; Heterogeneous wireless network.

I. INTRODUCTION

With the rapid development and deployment of wireless technologies, the future wireless networks are envisioned as a combination of multiple wireless access networks like WIFI, WIMAX and Universal Mobile Telecommunications System (UMTS), which can provide mobile users with Always Best Connected (ABC) services [1]. However, there are still some technical challenges to be further researched, handoff decision making is one of the most challenging problem in heterogeneous wireless networks since various parameters should be considered when executing handoff.

Handoff (HO) is the process of changing the mobile connection between different base stations or access points. In heterogeneous wireless networks, handoff can be divided into two categories: horizontal handoff (HHO) and vertical handoff (VHO) [2]. A HHO is made among different access networks when changing a connection from one access point (AP) to another or one base station (BS) to another, that is to say, the networks have the same link-layer technology. While a vertical handoff happens between access networks with different Iink­layer technologies, such as changing a connection between an AP and a BS. Compared with the horizontal handoff, the vertical handoff is more valuable for research.

In general, the VHO process involves three main phases [3]: 1) system discovery; 2) handoff decision; 3) handoff execution.

During the system discovery phase, mobile terminals equipped with multiple interfaces have to determine the available candidate networks and the available services. In the second phase, the mobile device has to decide whether the connections should continue using current selected network or be switched to another network. In the handoff execution phase, the connections need to be rerouted from current network to new network, with authentication, authorization and the transfer of context information.

In the vertical handoff procedure, the handoff decision phase is the most important phase because an inappropriate handoff decision may break off the current communication and degrade the QoS of traffic [3]. Thus this paper focuses on the vertical handoff decision phase and proposes a novel fuzzy logic based vertical handoff decision algorithm. Fuzzy Logic (FL) as a Multiple Attribute Decision Making (MADM) method is not only to evaluate and combine many parameters at the same time but also to deal with some imprecise information. The contributions of this paper can be summarized as follows.

1) The paper introduces a speed-adaptive system discovery scheme before handoff decision, which effectively improves the update rate of the candidate networks set.

2) We present a pre-hand off decision method which can eliminate the serious Ping-Pong effect and decrease the probability of call blocking and dropping.

3) The paper proposes a novel vertical handoff decision algorithm based on fuzzy logic. And the better performance of our proposed algorithm is validated via simulations.

The rest of this paper is organized as follows: we briefly review the related works about vertical handoff decision algorithms in Section II. In Section III, a novel fuzzy logic based vertical handoff decision algorithm is proposed. Then Section IV shows the performance evaluation of our proposed algorithm. Finally, conclusions are given in Section V.

II. RELATED WORK

In this section, the paper simply reviews some recent works on vertical handoff decision algorithm. After analyzing, we know that received signal strength (RSS) is usually used in traditional vertical handoff algorithms, and some subsequent RSS based handoff algorithms further consider hysteresis and

ISSN:1882-5621/13/ ©2013 NICT 1

Page 2: An efficient vertical handoff mechanism for  future mobile network

dwelling-time. The first policy-enabled handoff strategy [4] proposes the cost function to select the best available network in the decision making. Optimized cost function is used to select the target network by introducing tradeoff between user satisfaction and network efficiency. In [5], a vertical handoff decision algorithms for providing optimized performance in heterogeneous wireless networks is proposed, which can optimize the combined cost function according to the battery lifetime of the mobile nodes.

MADM is usually used to select a target network from a set of candidate networks that have many attributes to consider. The most popular traditional MADM methods are: 1) SAW (Simple Additive Weighting); 2) TOPSIS (Technique for Order Preference by Similarity to Ideal Solution); 3) AHP (Analytic Hierarchy Process); 4) GRA (Grey Relation Analysis). In [6], vertical handover decision schemes using SAW for network selection is proposed, it used the overall score of a candidate network to select target network, and the score determined by the weighting sum of all the parameters values. In [7], a TOPSIS based handoff algorithm is proposed, the target network is the one which is the closet to the ideal solution and the farthest from the worst case. In [8], AHP is used to decompose the network selection process into several sub­processes and assigns a weight value for each sub-process. And grey relation analysis for heterogeneous wireless networks [9] is presented, GRA is used to rank the candidate networks and select the one with the highest ranking.

Otherwise, there are some other vertical handoff algorithms have been put forward. In [lO], congestion-aware proactive vertical handoff decision using coalition game is proposed to maximizing the utilization of the available resources and meeting QoS requirement of mobile users. In [11], a handoff threshold function and a handoff cost function are used to evaluate the vertical handoff decision, it can dynamically adapt to the network environment changes. The classical MADM methods cannot efficiently solve a decision problem that should consider imprecise parameters. However, fuzzy logic can handle this problem well, and can evaluate and combine multiple attributes at the same time. In [12], a fuzzy logic based handoff algorithm in IEEE 802.16 and IEEE 802.11 hybrid networks is proposed, which can adapt itself with the dynamic conditions of the candidate networks.

Despite various vertical handoff decision algorithms have proposed before, they still have some drawbacks. Some algorithms which are complex could spend more time, and some algorithms that use the simple mechanism like just considering RSS may not select a suitable network and may have serious Ping-Pong effect, others can not reflect the dynamic variations of wireless network parameters. Moreover, previous algorithms seldom consider the update rate of the candidate networks set in terms of mobile terminal's speed and rarely use a pre-handoff decision method. To make the handoff decision algorithm better, our algorithm is described in Section III, which contains three parts.

III. THE PROPOSED SCHEME

In this section, we begin by describing a speed-adaptive system discovery scheme. Then we introduce a pre-handoff

decision method, which can quickly filter the candidate network set. Finally the fuzzy logic based vertical handoff decision algorithm is proposed.

A. Speed-adaptive System Discovery Scheme

In the system discovery phase, mobile terminals (MTs) need to determine which networks can be reached and used. This phase should be periodically invoked as the users are mobile and the available networks depend on the location of users. Hence the paper proposes a speed-adaptive network discovery scheme, which dynamically changes the update rate of candidate network set.

Recently, the GPS (Global Positioning System) technology is widely used in MTs. We can use this technology to get the instant velocity of the MT at a given time. Because the MT's movement is variable, we should compute the average velocity of the MT in fixed time duration. First we get N samples of

instant velocity of the MT in time duration Ts and then

compute the average value. The average velocity of MT is Va .

1 N

� =- LV; (i=l.. . . . . N) N ,�l

( 1)

Where V; represents the instant velocity of the i'th sample.

We denote IlT as the update time of the candidate network

set, which related to the average velocity Va'

IlT = (Tmax -Tmin )(1- r)" + Tmin (2)

(3)

Where Tmax and Tmin represent the maximum and minimum

update time respectively that users can set, n represents the exponential factor that related to the service and r is the

adaptive factor, Vmax represents the maximum available

velocity of the MT. We can know that when Va is bigger, and

then r is bigger, thus IlT is smaller. That means the update time of the candidate network set increases when MT moves quickly, it can effectively improves the network discovery time.

B. Pre-Handoff Decision Method

In this part, we describe a pre-handoff decision method with a quick evaluation function, which can quickly filter the candidate networks set according to the users' preference. Different services need various combinations of parameters, including RSS, handoff time delay, velocity of the MT, monetary cost and so on. We consider five important parameters after careful deliberation in this part, they are RSS, available bandwidth, handoff time delay, bit error rate (BER) and monetary cost. The pre-handoff decision method filters the candidate networks set depends on those parameters.

F; = F(RSS, -RSS'h)X F(B, -B'h)X F(D'h -D,)

xF(BER'h -BER)x F(C'h - C) (4)

Equation (4) is a minimum guarantee function, which determines whether the MT can access to the candidate

Page 3: An efficient vertical handoff mechanism for  future mobile network

network i . In this equation, RSSpBpDpBERi and Ci represent the values of the five parameters we mentioned above.

In addition, RSS,h,B,h,D'h,BER,h,C'h denotes the predefined

thresholds of RSS, available bandwidth, time delay, BER and monetary cost of the requested service. The users can set these thresholds according to their requirement.

The function F(*) is a unit step function, whose value is

one for positive input and zero for negative input. For this reason, only when the values of RSS and available bandwidth are larger than their thresholds, and the values of time delay, BER. monetary cost are lower than their thresholds, the output of the minimum guarantee function has one value. Thus the network i with one output value will be added to the candidate

networks set, otherwise the related network i is not considered as a candidate network. Because Equation (4) is simple to calculate, time consumption of the pre-handoff decision method is very low. Apparently, through this pre-handoff decision method, our method can efficiently eliminate the serious Ping-Pong effect and decrease the probability of call blocking and dropping.

After the process of the pre-handoff decision method, based on the size of the filtered candidate networks set, it will falls into three circumstances: 1) if the set is empty, the MT stays connect to the current network; 2) if there only have one candidate network, MT decides to make the handoff procedure to the sole network; 3) moreover, there is more than one candidate network in the set, we will choose the best network by using the decision algorithm in the next part.

C. Vertical HandojJ Decision Algorithm

The paper introduces a new vertical handoff decision algorithm based on fuzzy logic in this part. Four sub­procedures are showed in Fig.I.

Fig.l. Handoff decision algorithm flowchart

The parameters are first normalized, and then calculated by the fuzzification procedure. After that we make membership value evaluation. Finally we make a handoff decision according to the performance evaluation values.

In the proposed algorithm, we use normalized available bandwidth (B), RSS and monetary cost(C) as the inputs. The reasons why we select these three parameters for fuzzy processing are as follows. First, RSS is a crucial factor in

handoff, available bandwidth is an important parameter to measure networks performance, and monetary cost is related to the users' consideration. Second, if there are more parameters in this process, the system's load will be larger and it will cost more operation time.

The parameters are first normalized to balance the candidate networks with different access technology. The normalized function is given by equation (5):

F(x) = x-xmin

Where x = B,RSS,C.

1) Fu==ification

xma.x -xmin (5)

Every normalized input parameter has three fuzzy sets according to their membership function. The membership functions are shown in Fig.2-4.

1

11

Low Medium High

o'L---�------�------�----�l--� �B T," F(S)

Fig.2. Membership function ofB

Low Medium High

oL----L---L--�------�--�--� �RSS T,RSS 1 F(RSS)

Fig.3. Membership function of RSS

11

1 Low Medium High

oL----L�----�------�--��� F(e)

Fig.4. Membership function of C

Page 4: An efficient vertical handoff mechanism for  future mobile network

The fuzzy sets are: Low, Medium and High. From the figures, the membership degrees of B, RSS and C are:

Where i represents the i'th network, x = B,RSS,C.

2) Membership Value Evaluation

(6)

For the sake of evaluating the membership values of B, RSS and C, we distributed an impact factor to each parameter.

(r r r )=(FJx)-T;X F,(x)-T;X F,(x)) (7) I-L' I-M' I-H �x ' Tt _�x ' Tt

Based on impact factors and the membership degrees, we

can get the membership values of B, RSS and C for the i'th candidate network.

Where x = B,RSS,C.

3) Handoff Decision

The performance evaluation value (PEV) of the i' th network can be obtained by combining the three membership values. And every input parameter has its own weight that has to reflect the relationships and importance of the parameters. The weights are set by (9):

(9)

For the i'th candidate network, the PEV is:

PE� =W.�T (10)

Where �=(�B,�RSS,�C) and represents the

i'th candidate network.

The handoff decision mechanism is: first we choose the target network which has the largest PEV among the candidate networks; then compare the PEV of the target network with

that of the current network; if PE�arget -PEf/"urrent > PE�h'

make handoff to the target network; otherwise, stay connecting with the current network.

IV. PERFORMANCE EVALUATION

In order to make more realistic performance evaluation, our simulation scenarios and models are developed using NS-2 simulator, and the implementation of the proposed algorithm is completed by the MA TLAB.

In our simulation scenario, we suppose that there are an overlapping service network area which covered by WIFI, WIMAX and UMTS. The UMTS base station covers the whole simulation area, and the radius of WIMAX and WIFI network is 500m and 30m respectively. One WIMAX coverage area includes several WIFI networks. For simplicity, we define some important simulation parameters values in TABLE I. And for the non-real time services, the weights W= (0.4, 0.3, 0.3); for the real time services, W= (0.4, 0.4, 0.2).

TABLE I. Simulation Parameters

Parameter Value Parameter Value

Bmax 30 Mbps

RSSmax -60 dBm

Cmax 10 Cent

�B ,T2B 0.3, 0.7

�c ,T2C 0.25, 0.75

Bn11n

RSSmin

Cmin RSS RSS � ,T2

PE�h

o Mbps

-100 dBm

1 Cent

0.2, 08

0.1

In the simulation area, we assume that there are 20 mobile nodes (MNs), which are uniformly distributed and move freely in different speed. For the sake of figuring out how our algorithm improves the handoff results, we compare our algorithm with traditional RSS based handoff algorithm (Algorithm 1) and sole fuzzy logic based handoff algorithm (Algorithm 2) in three aspects: handoff times, load balancing degree and probability of call blocking and dropping. We run the simulation for 10 times and every time spends 10 minutes.

If> 1)

.§ � 0

'0 §

::r:

20����--�����======� _ Algorithm 1

15

10

1

5

0

c:::J Algorithm 2

_ Our algorithm

1

2 3 4 5 6 7 8 9 10 Order of simulations

Fig.5. Comparison of hand off times

Fig.5 shows the handoff times of different simulation times, we can see that our algorithm has fewer handoff times than other two algorithms. It illustrates our algorithm can reduce unnecessary handoffs, and thus eliminate the serious Ping-Pong effect. That is because the proposed algorithm considers many different parameters, not only considering RSS like Algorithm 1, and adds the pre-handoff decision method compared with Algorithm 2.

In Fig.6, the load balancing performance is compared when the number of MN increases from 20 to 100. The load balancing degree is calculated as:

N d = L 11 - (B�cCltPied / B:ota1 ) / (B occupied / Btotal) 1 (11)

i=l

Where i represents the i' th network, B represents bandwidth. The algorithm which has lower degree could have better load balancing performance. It is visible that the

Page 5: An efficient vertical handoff mechanism for  future mobile network

proposed algorithm outperforms other two algorithms because it can distribute the MNs averagely in the simulation area when the candidate networks can all provide the desired service.

0.4

� 0.35 .... � � 0.3 c

'0 � 0.25 � "0 g 0.2 ....l

bIJ c

'0. 0-2

"0

0.15

"0 § 0.25 bIJ c

:i2 <.J

� 0.2

� "'" o

, ...... "

-+-, Algorithm 1

--+--, Algorithm 2

--+--, Our algorithm

.-----...... . •• ..t-----•..••.

.......... -----•...

40 60 80 100 The number ofMNs

Fig,6, Comparison of load balancing degree

• Algorithm 1

D Algorithm 2 .Our algorithm

:E 0.15

1 I II 0...

0.1 L- -la------2- ' ----

3 4 5 6 7 8 9 1 0 Order of simulations

Fig,7, Comparison of probability of call blocking and dropping

Call blocking and dropping is another critical issue that should be considered. Fig.7 shows us the comparison of probability of call blocking and dropping in 10 simulations. It is indicated that our algorithm could decrease the probability of call blocking and dropping obviously compared with other two algorithms. Based on the results above, we can conclude that the proposed algorithm effectively improves the performance of handoffs and can provide users with high quality of services.

V. CONCLUSION

Handoff decision making is one of the main challenges in vertical handoff process because there are different types of wireless network and plenty of parameters that should be taken into account. In this paper, we have proposed a vertical handoff decision algorithm based on fuzzy logic which considers many parameters including available bandwidth, RSS, monetary cost,

BER and time delay. The simulation results above shows that the new algorithm outperforms the traditional RSS based handoff algorithm and the sole fuzzy logic based handoff algorithm in several aspects including reducing unnecessary handoffs, balancing the whole network resources and decreasing the probability of call blocking and dropping.

ACKNOWLEDGMENT

This research was supported by the National Natural Science Foundation of China (No.61101106) and the Fundamental Research Funds for the Central Universities (No.2011RC0107). Thanks for the Open Research Fund of Beijing Key Laboratory of Network System Architecture and Convergence, Beijing University of Posts and Telecommunications .

REFERENCES

[I] E, Gustafsson and A. Jonsson, "Always best connected," IEEE Wireless Communications Magazine, vol. 10, no, I, pp, 49-55, Feb 2003,

[2] e. Gu, M, Song, Y. Zhang, L. Wang, and J. Song, "Novel network selection mechanism using AHP and enhanced GA," 7th Annual Communication Networks and Services Research Conference (CNSR), pp, 397-401, 2009,

[3] J. McNair and F, Zhu, "Vertical han doffs in fourth-generation multi­network environments," Wireless Communications, vol. II, no, 3, pp, 8-15, Jun 2004,

[4] H, Wang, R, Katz, and J. Giese, "Policy-enabled handoffs across heterogeneous wireless networks," Second IEEE Workshop on Mobile Computing Systems and Applications (Proceeding WMCSA '99 ), pp, 51-60, 1999,

[5] S, Lee, K. Sriram, K, Kim, Y. Kim and N, Gohnie, "Vertical handoff decision algorithms for providing optimized performance in heterogeneous wireless networks," IEEE Transaction on Vehicular Technology, vol. 58, no, 2, pp, 865-881, Feb 2009,

[6] K. Savitha and e. Chandrasekar, "Vertical handover decision schemes using SAW and WPM for network selection in heterogeneous wireless networks," Global Journal of Computer Science and Technology Volume II Issue 9,May 2011.

[7] S, Liu, S, Pan, M, Xu, "An improved TOPSIS vertical handoff algorithm for heterogeneous wireless networks," IEEE International Conference on Communication Technology (ICCT), pp, 750-754, Nov 2010,

[8] K. Radhika and Dr. A, Venugopal Reddy, "AHP and group decision making for access network selection in multi-homed mobile tenninals," International Journal on Computer Science and Engineering(UCSE), ISSN: 0975-3397, vol. 3, no, 10, Oct 2011.

[9] K. Savitha and e. Chandrasekar, "Grey relation analysis for heterogeneous wireless networks," European Journal of Scientific Research, ISSN 1450-216X, vol. 54, no, 4, pp, 560-568, 2011,

[10] S,V, Saboji and e.B, Akki, "Congestion-aware proactive vertical handoff decision using coalition game," International Journal of Soft Computing and Enginneering (USC E), ISSN, 2231-2307 Volume I Issue 6, Jan 2012,

[II] A. Hassawa, N, Nasser and H, Hassanein, "Tramcar: A context-aware cross-layer architecture for next generation heterogeneous wireless network," Proc, IEEE ICC, pp, 240-245, Jun 2006,

[12] J. Nie, L. Zeng and J, Wen, "A bandwidth based adaptive fuzzy logic handoff in IEEE 802,16 and IEEE 802, II hybrid networks," International Conference on Convergence Infonnation Technology, pp.24-29, Nov 2007,