Handover algorithm for multiple networks based on Bayesian decision

KONG Ling-bin, WANG Jun-xuan

(School of Communication and Information Engineering,Xi’an University of Posts and Telecommunications, Xi’an 710121, China)

?

Handover algorithm for multiple networks based on Bayesian decision

KONG Ling-bin, WANG Jun-xuan

(SchoolofCommunicationandInformationEngineering,Xi’anUniversityofPostsandTelecommunications,Xi’an710121,China)

An improved vertical handover algorithm for multiple networks based on Bayesian decision is proposed. Firstly, the handover probability distribution is established considering multiple conditions including signal strength, bit error rate, blocking probability and user demands, and accordingly the prior handover probability is calculated. Secondly, the posterior probability based on Bayesian decision algorithm is got. Finally, the optimal access network is selected according to the decision strategy based on posterior probability. Simulation results indicate that the proposed algorithm not only effectively achieves vertical handover among WLAN, WiMAX and LTE with the least number of handovers, but also keeps high average network load, which can provide the users with good service quality.

multiple networks; blocking probability; Bayesian decision; vertical handover

With the rapid development of wireless communication technology, more and more wireless access technologies have emerged. Integration of heterogeneous wireless networks, can simultaneously support many different types of business. Vertical handover can be used to ensure that mobile terminals have the continuity of high-quality communications and roaming services in multi-network session anytime, any place, any when. Thus, the research on vertical handover algorithm in heterogeneous networks has a high theoretical and practical significance.

Currently, there have been many wireless access technologies which can be applied to terminal nodes, but different access technologies have their own characteristics. For vertical handover technology in heterogeneous wireless networks a lot of related research has been made, including vertical handover algorithm based on signal strength[1], gray relational analysis (GRA)[2]and vertical handover algorithm based on signal to interference plus noise ratio (SINR)[3]. But most of the vertical handover algorithms take a single network attribute factor as the indicator of handover decision regardless of actual demand preference of the business for network. TAO Yang, et al.[4]proposed adaptive speed vertical handover algorithm based on business needs for better heterogeneous network handover in business. Ref.[5] proposed a vertical handover algorithm in heterogeneous wireless network environment based on car networks to select the existing optimal access network as target network for handover.

Although some research achievements have been obtained, due to uniqueness of a heterogeneous network environment and different preferences and service needs from different users, only considering the network state and terminal-related stat is not sufficient. But some simple weighted strategies do not reflect the actual demands of network terminals, and the research on verification of heterogeneous network environment is deficient.

To solve these problems, the terminals in wireless local area nework (WLAN), world interoperability for microwave access (WiMAX) and long term evolution (LTE)[6]heterogeneous network environments are taken as the research objects. By fully considering the actual demands of different applications, the weights of netowork attribute factors are adaptively adjusted to make effective handover decision.

1 Vertical handover algorithm based on Bayesian decision

The block diagram of the improved vertical handover algorithm based on Bayesian decision[6]is shown in Fig.1. The weights of network attribute factors are determined according to different applications and then network handover decision is made according to the weights. Afterwards, the Bayesian decision algorithm is used to calculate posterior probability. Finally, by comparing the handover probabilities, the optimal target network can be selected for network handover.

Fig.1 Vertical handover algorithm based on Bayesian decision

1.1 Network attribute factors based on business demands

To provide the users with the best service in complex heterogeneous networks, namely, vertical handover algorithm should choose optimal network to meet different user requirements. Therefore, the multi-attribute vertical decision algorithm based on business demands analyzes first the quality of service (QoS) to selecte network attributes for business evaluation and sets the threshold for the candidate networks. Then analytic hierarchy process (AHP)[7]is used to determine the weight of each network attribute factor. Finally, by means of Bayesian criteria, the optimal handover network is got.

Since different preferences and business demands lead to a variety types of business[8], to meet user requirements, the business is divided into four classes: conversational class, interactive class, streaming class and background class, as listed in Table 1.

Table 1 Requirements of network QoS

1.2 Network attribute factors definition and analysis

Network attribute factors affecting vertical handover of heterogeneous wireless networks include signal strength, transmission rate, bit error rate and network blocking probability. The detailed analysis is as follows:

1) Signal strength

Signal strength is a basic trigger condition of vertical handover and reflects the quality of signal strength of the current channel. It can be expressed as

whereK1is network transmission power,K2is network path loss factor,ddenotes the distance between the terminal and access point andu(x) is Gaussian random distribution function that obeys (0,σ).

2) Maximum transmission rate

Transmission rate is an important indicator for network selection, which directly affects business quality of the terminal. According to Shannon’s theorem, the maximum transmission rate of the channel is

whereWis bandwidth,sis average signal power andnis average noise power.

3) Bit error rate (BER)

When bit error rate of network is higher than a certain threshold, the network will not meet the current needs of the business. To calculate the bit error rate, assuming that there exists the Gaussian noise obeying random distribution, and the distance from the terminal to the base station isdk(k=1,2,…), BER is a function of signal to noise ratio (SNR) and can be expressed as

whereI(k) is interference signal strength,Q(x) obeys Gaussian distribution with parameter (0,1), namely,

4) Network blocking probability (Pbk)

This paper uses a mathematical model of network blocking probability proposed in Ref.[10], which studies the effects of two algorithms on network blocking probality in vertical handover, and it is expressed as

1.3 Multi-condition handover probability distribution

The current network handover is based on single or multiple network attribute factors, accordingly, the corresponding threshold values are set. The handover probabilities based on single network attribute factor are as follows:

1) Handover probability based on signal strength

P1=P(YRSS(d)≥η),

whereYRSS(d) is the signal strength of target network for handover andλis the minimum signal strength threshold required when the terminal has access to network.

2) Handover probability based on bit error rate

P2=P(RBER(k))<τ,

whereRBER(k) is the BER of target network for handover andτis the maximum BER which can meet terminal business needs.

3) Handover probability based on transmission rate

P3=P(VCB>VCφ),

whereVCBis the maximum transmission rate of target network for handover, andVCφis the minimum transmission rate which only meets current business needs.

4) Handover probability based on blocking probability (network load)

P4=P(Pbk<ε),

wherePbkis the network load of target network for handover, andεis the maximum blocing probability which meets the business needs.

For the above conditions, there is a correlation between each other, where the signal strength is a comparative reference condition. Based on the above analysis, a prior probability for handover can be given by

Pth=P(YRSS(d)>η,RBER(k)<τ,VCB>VCφ,Pbk<ε).

2 Vertical handover algorithm based on Bayesian decision

When making a decision for network handover, the two states for network handover are taken as random variables,φ1andφ2, and the prior probabilities of two states are expressed byPthandPfh, wherePthindicates the probability that handover occurs,Pfhindicates the probability that handover does not occur,Pth+Pfh=1. In general, ifPth>Pfh, making the decision for network handover.

If only according to the prior probability for handover decision, all the networks are classified into a state,φ1andφ2. Therefore it is necessary to introduce network priority to make common decision for handover[9-10].

2.1 Selection of target network based on Bayesian decision

In order to distinguish the multi-target networks in covered areas accurately, the rule of multi-target networks based on Bayesian decision for handover are given as follows:

Letx1,x2andx3denote the handover states of three access networksC1,C2andC3, respectively. Set network priority:C1>C2>C3, there isP(x1)>P(x2)>P(x3). Assuming thatP(x1|φ1) is the conditional probability that the terminal has access to networkx1in case of network handover andP(x1|φ2) is the conditional probability that the therminal has access to networkx1in case of no network handover, using Bayesian theorem, there is

whereP(φi|x1) is the posterior probabilities of handover states. Based on the posterior probability, the decisions are made as follows:

1) IfP(φ1|x1)>P(φ2|x1), the state of networkx1is classified asφ1and the terminal is switched to networkC1;

2) IfP(φ1|x1)

Therefore, according to the above analysis, it can be inferred as follows:

1) IfP(x1|φ1)P(φ1)>P(x1|φ2)P(φ2), the state of networkx1is classified asφ1and the target network is considered to be switched to networkC1;

2) IfP(x1|φ1)P(φ1)

2.2 Selection of target network for handover based on Bayesian decision

Assuming there are several candidate handover target networks, an optimal selection strategy based on Bayesian decision is as follows:

1) IfP(x1|φ1)P(φ1)>P(x2|φ1)P(φ1) as well asP(x1|φ1)P(φ1)>P(x3|φ1)P(φ1), networkC1is selected as a candidate for handover;

2) IfP(x2|φ1)P(φ1)>P(x1|φ1)P(φ1) as well asP(x2|φ1)P(φ1)>P(x3|φ1)P(φ1)， networkC2is selected as a candidate for handover;

3) IfP(x3|φ1)P(φ1)>P(x1|φ1)P(φ1) as well asP(x3|φ1)P(φ1)>P(x2|φ1)P(φ1), networkC3is selected as a candidate for handover.

In practical applications, this algorithm can be extended for the selection of the best candidate target network among four or more candidate networks.

3 Simulation and analysis

In order to verify the proposed algorithm, simulation scenarios must be combined with the actual network application. In this paper, the simulation movement scene is constructed to verify vertical handover algorithm for mobile terminal by using Matlab.

3.1 Experimental parameters

This paper gives a simulation scenario, as shown in Fig.2.

Fig.2 Terminal movement scene

To facilitate data acquisition and analysis, the mobile terminal moves from pointAto pointBin a straight line at the speed of 1 m/s. The network simulation properties are set, as shown in Table 2.

Table 2 Network property settings

3.2 Experimental results and analysis

In the simulation experiments, the handover probability of each network is calculated at the interval of one second in the mobile terminal for a terminal handover decision, whereP(x1|φ1)=0.4，P(x2|φ1)= 0.32 andP(x3|φ1)=0.28.

This paper focuses on the design of multi-network attributes vertical judgment algorithm, which has different performance in many aspects and are combined with the different wireless networks of LTE, WiMAX and WLAN. For the current business types in the same network environment, there are three different algorithms for comparative analysis of the handover Network. The following simulation diagram gives the specific circumstances of simulation experiments based on the current interaction class services of network handover.

Fig.3 shows the number of terminal handovers based on different decision strategies in heterogeneous networks. It can be seen that not only the number of terminal handovers based on Bayesian decision significantly decreases, but also the residence time of the terminal in the network with good properties are the longest. This not only effectively reduces “ping-pong” effect and system overhead, but also provides the user with the best network experience.

Fig.3 Terminal handover according to different decision strategies in a heterogeneous network

Fig.4 shows the relationship between average network load and blocking probability. It can be seen that the handover based on Bayesian decision makes the terminal have access to the network in good load condition as far as possible, which reduces the network blocking of hot-spot cells.

Fig.4 Relationship between average network load and blocking probability in a heterogeneous network

Fig.5 shows the handover probabilities of the mobile terminal based on different handover decision strategies when it moves at different speeds in a heterogeneous network. It can be seen that the improved algorithm can reduce the number of handovers, which ensures better network performance and provides the users with better service quality and effects.

Fig.5 Terminal handover probability with different decision strategies at different movement speeds

4 Conclusion

This paper presents a vertical handover algorithm for multiple networks based on Bayesian decision. When the terminal moves in different networks such as WLAN, WiMAX and LTE, full considering the weights of network attribute factors in different applications, the optimal access network can be selected by the proposed Bayesian decision algorithm. Simulation results show that this algorithm not only can be used for any vertical handover in heterogeneous networks, which can reduce the number of handovers, but also ensure that the target network is the current optimal network for handover. The further work will focus on how to introduce the user's access preference into the vertical handover algorithm to improve the terminal handover performance in a heterogeneous network.

[1] LIU Ming, LI Zhong-cheng, GUO Xiao-bing, et al. Performance evaluation of vertical handoff decision algorithms in heterogeneous wireless networks. IEEE Transactions on Mobile Computing-TMC, 2008, 7(7): 846-857.

[2] Mukherjee S, Ishii H. Energy efficiency in the phantom cell enhanced local area architecture. In: Proceeding of IEEE Wireless Communications & Networking Conference, Shanghai, China, 2013: 1267-1271.

[3] LIU Min, LI Zhong-cheng, GUO Xiao-bin, et al. Switch-based adaptive algorithm vertical movement trend and its performance evaluation. Chinese Journal of Computers, 2008, 31 (1): 112-119.

[4] TAO Yang, JIANG Yan-li, CHEN Lei-cheng. Speed vertical handover algorithm based on application requirements. Journal of Computer Applications, 2014, 34(5): 1236-1238, 1262.

[5] van Base P. Vertical handover method based on Bayesian decision vehicle networking. Journal of Communication, 2013, (7): 4.

[6] Pedersen K I, Barbera S. Mobility enhancements for LTE-advanced multilayer networks with inter-site carrier aggregation. IEEE Communication. Magazine, 2013, 51(5): 64-71.

[7] Fallah Y P, Khan S, Nasiopoulos P, et al. Hybrid OFDMA/ CSMA based medium access control for next-generation wireless LAN’s. In: Proceedings of IEEE Conference on Communications (ICC’08). Beijing, China, 2008: 2762-2768.

[8] Ylianttila M, Hossain E, Chow G, et al. Vehicular telematics over heterogeneous wireless networks:a survey. Computer Communications, 2010, 33(7): 775-793.

[9] Buddhikot M, Chandranmenon G, Han S, et al. Integration of 802.11 and third-generation wireless data networks. In: Proceedins of 22nd Annual Conference on the IEEE Computer and Communication Society (INFOCOM 2003), San Francisco, USA, 2003: 503-512.

[10] PAN Su, YE Qiang, LIU Su-mei. Equivalent spectral bandwidth concept in ubiquitous networks and applications in vertical handoff algorithms. Journal on Communications, 2012, 33(3): 130-136.

基于貝葉斯決策的多網(wǎng)絡(luò)切換算法

孔令斌， 王軍選

(西安郵電大學(xué) 通信與信息工程學(xué)院， 陜西 西安 710121)

現(xiàn)有的垂直切換技術(shù)通常不支持多網(wǎng)絡(luò)下切換。 為此， 提出了基于貝葉斯決策的改進(jìn)算法。 首先根據(jù)接入用戶終端的信號(hào)強(qiáng)度、 網(wǎng)絡(luò)阻塞率和誤碼率以及不同用戶業(yè)務(wù)對(duì)網(wǎng)絡(luò)的實(shí)際需求建立多條件相關(guān)的切換概率分布， 得出先驗(yàn)切換概率； 然后利用貝葉斯垂直切換決策算法計(jì)算出后驗(yàn)概率； 最后， 根據(jù)后驗(yàn)概率的決策規(guī)則選出最優(yōu)接入網(wǎng)絡(luò)。 仿真結(jié)果表明， 該算法不僅有效地實(shí)現(xiàn)異構(gòu)無(wú)線接入網(wǎng)之間的垂直切換， 避免了不必要的切換， 而且還能保持較高的網(wǎng)絡(luò)平均負(fù)載， 使用戶獲得更好的服務(wù)。

網(wǎng)絡(luò)； 網(wǎng)絡(luò)阻塞率； 貝葉斯決策； 垂直切換

KONG Ling-bin, WANG Jun-xuan. Handover algorithm for multiple networks based on Bayesian decision. Journal of Measurement Science and Instrumentation, 2015, 6(4)： 347-353. [

JournalofMeasurementScienceandInstrumentationISSN1674?8042，QuarterlyWelcomeContributionstoJMSI(http：∥xuebao．nuc．edu．cn)(jmsi@nuc．edu．cn) ※ XJMSIaimstobuildahigh?levelacademicplatformtoexchangecreativeandinnovativeachievementsintheareasofmeasurementscienceandinstrumentationforrelatedresearcherssuchasscientists，engi?neersandgraduatestudents，etc．※ JMSIcoversbasicprinciples，technologiesandinstrumentationofmeasurementandcontrolrelatingtosuchsubjectsasMechanics，ElectricalandElectronicEngineering，Magnetics，Optics，Chemistry，Biol?ogy，andsoon．※ JMSIhasbeencoveredbyAJ，IC，EBSCO，CSCD，CNKIandCOJ．

10.3969/j.issn.1674-8042.2015.04.008]

Foundation items： National 863 Project of China (2014AA01A703); Natural Science Foundation of Education Department of Shaanxi Province (2013JK1045); ZTE Forum Foundation of ZTE Corporation

KONG Ling-bin (1349409974@qq.com)

1674-8042(2015)04-0347-07 doi： 10.3969/j.issn.1674-8042.2015.04.008

Received date： 2015-09-15

CLD number： TN911.5 Document code： A