Vertical Handover Techniques Evaluation

Size: px

Start display at page:

Download "Vertical Handover Techniques Evaluation"

Loreen Atkins
5 years ago
Views:

1 Vertical Hadover Techiques Evaluatio BOJAN BAKMAZ, ZORAN BOJKOVIC, MIODRAG BAKMAZ Uiversity of Belgrade SERBIA Abstract: - Hadover techiques are essetial compoets of the heterogeeous wireless etworks architecture developmet. Providig quality of service i coditios of seamless mobility is the major prerequisite for vertical hadover high fuctioality ad scalability. I this paper, the perspective vertical hadover techiques, based o radio lik quality ad merit fuctio of access etwork, are systematically preseted ad aalyzed i terms of efficiecy ad implemetatio complexity. Key-Words: Heterogeeous wireless etworks, Mobility, Vertical hadover, QoS. 1 Itroductio The priciple of Always Best Coectivity (ABC) represets the basic cocept of the ext geeratio commuicatio systems, which provides to the users access to services over optimal etwork. Followig the priciples of heterogeeous etworkig, a mobile termial may choose amog multiple available coectivity alteratives based o the criteria related to etworks performaces, users prefereces ad services requiremets [1]. This process makes a importat elemet i the complex hadover process. The eed for a hadover i the traditioal wireless etworks usually occurs whe the termial, due to the movemet of the users, leaves the access router coverage area. I a heterogeeous eviromet, hadover is more frequetly used i order to improve commuicatio ad rider i order to maitai coectios [2]. A great umber of techiques related to the hadover iitiatio ad optimal access etwork selectio are proposed i the ope literature. The suggested techiques are usig differet metrics ad heuristics for solvig the above metioed problems. Ufortuately curretly proposed vertical hadover techiques do ot meet all the requiremets i terms of fuctioality ad efficiecy. This paper is orgaized as follows. After brief itroductio, a comprehesive survey of hadover iitiatio ad optimal etwork criteria is provided. The, we cotiue with hadover techiques evaluatio. Hadover techiques based o radio lik quality ad techiques based o merit fuctio are systematically preseted ad aalyzed from the aspects of efficiecy ad implemetatio complexity. Theoretically discussio cocludes the paper. 2 Hadover Iitiatio ad Optimal Network Selectio Criteria There are four groups that are ofte aalyzed as the criteria i the vertical hadover techiques, ad those groups are related to the etities - the participats i hadover decisio: Network-orieted metrics (coverage, lik quality, badwidth, etc.), Service-orieted metrics (QoS level, security level, etc.), User-orieted metrics (users prefereces, perceived QoS, etc). Termial-orieted metrics (velocity, eergy cosumptio, etc.). Received Sigal Stregth (RSS) is the most widely used criterio because of its measure simplicity ad close correlatio to the lik quality. There is a close relatioship betwee RSS readigs ad the distace form the mobile termial to its Poit of Attachmet (PoA). Badwidth represets the importat idicator of the traffic performaces i the access etwork ad the trasparet parameter for the curret ad future users of the multimedia services. The QoS level ca be defied through the metric values of delay, jitter, package loss, etc. ad it ca be declared by the service provider o the basis of ISBN:

2 ITU recommedatio Y.1541, which defies the upper bouds of QoS parameters for specific applicatios or classes of services. By declarig the QoS level i this way, we will avoid a complex examiatio of QoS parameters by users as well as the additioal load of user termials ad other etwork elemets. Security level, as well as the previous criteria, may be declared by the service provider, ad it represets the security measure for the iformatio trasfer i the etwork. For most users, depedig o the applicatio, security plays a great part i makig a decisio o the adequacy of a etwork for trasferrig the desired cotet. Cost of service ca sigificatly vary from provider to provider, but i differet etwork eviromets. I some cases cost ca be the decidig factor for optimal etwork selectio, ad it icludes the traffic costs ad the costs of roamig betwee heterogeeous etworks. I some cotext cost of service is i tight relatio with available badwidth, QoS level, security level, but i ext geeratio wireless eviromet, cost of service is fast time differetiable fuctio depedable of may others parameters. The above metioed metrics are affectig the momet of the hadover iitiatio ad optimal access etwork selectio. The umber of criteria, ad dyamic variability of some parameters sigificatly icreases the complexity of the hadover techique, ad because of that, the choosig of adequate criteria is of great importace. 3 Hadover Techiques Performace Evaluatio Quatitative performace aalysis of the hadover techiques ca be performed through the determiatio of mea ad maximum hadover delays, umber of hadovers, umber of hadovers failed due to the icorrect decisios, hadover failure probability, resource utilizatio, etc [3, 4]. Hadover delay refers to the duratio betwee the iitiatio ad completio of the hadover process. It is related to the complexity of the applied hadover techique. Reductio of the hadover delay is especially importat for delay-sesitive voice ad multimedia sessios. Reducig the umber of hadovers is usually preferred, as frequet hadovers would cause wastage of etwork resources. A hadover is cosidered as superfluous whe a mobile termial back to the previous PoA is eeded withi certai time duratio ("pig-pog" effect), ad such hadovers should be miimized. A hadover failure occurs whe the hadover is iitiated, but the target etwork does ot have sufficiet resources to complete it, or whe the mobile termial moves out of the coverage area before the process is fialized. I the first case, the hadover failure probability is related to the resource availability (e.g. chael availability) of the target etwork, while i the secod case, it is related to the termial mobility. Resource utilizatio is defied as the ratio betwee the mea amout of utilized resources ad the total amout of resources i a system. I the case of efficiet chael utilizatio, the ratio betwee the mea umber of chaels that are beig served ad total umber of chaels i a system is take ito accout. 4 Hadover Techiques Based o Radio Lik Quality Traditioal horizotal hadover techiques are basically aalyzig metrics trough the variats of comparig RSS of the curret PoA ad cadidate etwork PoA. I combiatio with threshold ad hysteresis, RSS metrics represet a satisfyig solutio for a homogeeous etwork eviromet. I a heterogeeous eviromet RSS metric is ot sufficiet criterio for iitiatig a hadover, but i combiatio with other metrics it could be applied as ultimate coditio. Oe of the proposed techiques [5] aalyzes the hadover iitiatio betwee 3G ad WLAN etworks, i the case of the cell overlappig. RSS is used as a basic metric i combiatio with a estimated lifetime (expected time after which the mobile termial will ot be able to maitai its coectio with the WLAN) ad available badwidth of cadidate WLAN PoA. The velocity of the termial, hadover delay ad Applicatio Sigal Stregth Threshold (ASST) were used as secodary parameters. Accordig to the first preseted sceario, whe the mobile termial moves away from the coverage area of a WLAN ito a 3G cell, a hadover to the 3G etwork is iitiated. The hadover is triggered uder coditios that average RSS of the WLAN PoA falls below a predefied threshold (MOT WLAN ) ad the Estimated Lifetime (EL) is less tha or equal to the hadover delay. The mobile termial cotiually calculates average RSS usig movig average method. ISBN:

3 I the secod sceario, whe the mobile termial moves towards WLAN cell, the hadover is triggered to the WLAN if the average RSS is larger tha a threshold (MITW LAN ) ad the available badwidth of the WLAN is sufficiet for the curret applicatio. Algorithm of the hadover techique based o RSS is show i Fig. 1. Fig. 1. Example of RSS based hadover techique By usig the EL metrics the applicatio requiremets are icluded i terms of quality of service, as aspects of user mobility, which largely reduces the umber of uecessary hadovers. I this case the termial is usig the resources (badwidth) of the WLAN eviromet as much as possible, which icreases the perceived quality of service by user. The disadvatage of this techique is reflected i the icreasig of packet delay ad complexity of implemetatio. Other represetative techique, based o the radio lik quality, is related to the use of Sigal to Iterferece ad Noise Ratio (SINR) metric i hadover betwee WLAN ad WCDMA (Wide Code Divisio Multiple Access) etwork [6]. The SINR calculatio of the WLAN sigals is modulated to a equivalet SINR to be comparable with the SINR of the WCDMA chael. Hadover is iitiated if SINR of target etwork is larger tha SINR of curret etwork. The SINR metrics applicatio ca provide users higher throughput tha RSS based techiques sice available badwidth directly depedet o the SINR, ad results i balaced load betwee the WLAN ad WCDMA etworks. Nevertheless, such a techique may also itroduce superfluous hadovers with the variatio of SINR causig the hadover back ad fourth betwee two PoAs. 5 Hadover Techiques Based o Merit Fuctio Merit fuctio, based o which a decisio i the process of vertical hadover is made, combies several criteria for optimal access etwork selectio. The evaluatio of the criteria i order to form a uique fuctio is implemeted by usig oe or more mathematical tools. For solvig a problem, cost fuctio, Multi-Criteria Decisio Makig (MCDM), artificial eural etworks, fuzzy logic, etc., ca be applied. Vertical hadover techique that relies o a cost fuctio is proposed i [7]. Cost fuctio for each available etwork is calculated by C Ws jqs, j s j =,, E s, j 0, (1) where Q s, j is the ormalized QoS provided by etwork for parameter j (e.g. badwidth, delay) ad service s (e.g. video, voice). W s, j is correspodig weight coefficiet which idicates the impact of the QoS parameter o the user or the etwork, ad E s, j is the etwork elimiatio factor, idicatig whether the miimum requiremet of parameter j for service s ca be met by etwork. Hadover is iitiated towards the access etwork with the miimum value of cost fuctio. The algorithm of the proposed techique is show i Fig. 2. Fig. 2. Example of cost fuctio based techique I [7], algorithm modificatio by which, for the each active sessio the optimal access etwork ca be determied is proposed. The fudametal beefit ISBN:

of cost fuctio usage ad hadover idepedet iitiatio for differet services is reduced failure (blockig) probability.

Similar to the previous techique, the article [8] aalyzes the applicatio of cost fuctio i the process of evaluatig the qualitative performace of potetial target etworks.

security, P i power cosumptio, D i etwork coditio ad F i etwork performace, while the w c, w s, w p, w d ad w f are weights for each of the etwork parameters which are proportioal to the sigificace

4 of cost fuctio usage ad hadover idepedet iitiatio for differet services is reduced failure (blockig) probability. However, parameter ormalizatio ad weights coefficiets determiatio techiques are ot discussed. Similar to the previous techique, the article [8] aalyzes the applicatio of cost fuctio i the process of evaluatig the qualitative performace of potetial target etworks. By usig the ormalizatio ad weights distributio methods, cost fuctio determies a etwork quality factor 1 1 Q i = f ( wc, ws Si, wp, wd Di, w f Fi ), (2) C P i where C i is the cost of service, S i security, P i power cosumptio, D i etwork coditio ad F i etwork performace, while the w c, w s, w p, w d ad w f are weights for each of the etwork parameters which are proportioal to the sigificace of a parameter to the vertical hadover decisio. Due to the heterogeeity of the parameters, it is ecessary to make a ormalizatio of the fuctio. High overall throughput ad user s satisfactio ca be regarded as major advatages of this heuristics. A techique that is usig the artificial eural etwork for the optimal access etwork selectio is proposed i [9]. Applied feedforward eural etwork topology, which cosists of iput, hidde ad output layer, is show i Fig. 3. i the same cost fuctio as i [8]. The performed simulatios have show high accuracy ad reliability of the model while selectig the optimal etwork. The lacks of the algorithm are reflected to the complexity of the system ad to icreased hadover delays due to the traiig process. To associate to a ature of problem, a great umber of the hadover techiques available i the ope literature are based o MCDM [10-12]. Hadover techique proposed i [12] represets iterestig ad promisig solutio while combiig the heuristics of the fuzzy logic systems ad MCDM (Fig. 4). Fig. 3. Topology of the eural etwork as vertical hadover techique The iput layer is made of the h odes represetig differet criteria for optimal etwork selectio, while the hidde layer cosists of the odes that represet the available access etworks. Logistic sigmoid activatio fuctio f(x) = 1/(1+e -x ) is applied to determie the cost fuctio. For the traiig process error backpropagatio algorithm is used. Output layer is formed by a ode that geerates the idetificatio of the optimal access etwork. Durig the simulatio, the authors adopted Fig. 4. Example of hadover techique based o fuzzy logic system ad MCDM I the process of iitiatig hadover proposed techique uses fuzzy logic aalyzig the criteria such as: RSS, badwidth (B), etwork coverage (NC) ad termial velocity (V). Based o 4 related fuctios ad 81 predefied rules, a system determies whether hadover is ecessary or ot. By applicatio of Aalytic Hierarchy Process (AHP) method ad Saaty's scale o criteria such as cost of service, preferred iterface, battery status ad QoS level, the optimal access etwork is determied. ISBN:

5 Table 1. A comparative survey of represetative hadover techiques Heuristics Criteria Advatages Disadvatages Radio lik quality [5, 6] RSS, EL, SINR. Restrictive umber of criteria. Reduced umber of superfluous hadovers Short hadover delay Simplicity Icreased hadover failure probability Ureliability Cost fuctio [7,8] Cost of service, security, eergy cosumptio, badwidth, lik quality. Icreasig of criteria umbers does ot reflect to the complexity. Low hadover delay Low blockig probability Sigificat complexity Icreased itesity of sigalig traffic Artificial eural etworks [9] Icremet of criteria umbers icreases complexity. Reduced umber of superfluous hadovers High reliability Sigificat complexity Icreased hadover delay MCDM [10-12] RSS, badwidth, coverage, termial velocity, cost of service, preferred iterface, battery status, QoS level. Icremet of criteria umbers icreases complexity i the segmet of hadover iitiatio. Reduced umber of superfluous hadovers High reliability Sigificat complexity Iflexibility of the impact of user's prefereces O the other had, by applyig fuzzy logic i decisio-makig process the umber of uecessary hadover are reduced, as well as sigalig traffic ad hadover delays. Iflexibility of the impact of user s prefereces o the system is basic lack of applied AHP method, which could possibly be exceeded by usig some other MCDM techiques (e.g. TOPSIS). A qualitative compariso of the aalyzed hadover techiques i terms of fuctioality ad complexity of implemetatio is give i Table 1. 6 Cocludig Remarks Desigig of the efficiet hadover techiques is the crucial research directio i the field of heterogeeous wireless etworks. Curret techiques do ot fully comply with requiremets regardig the techology coverage, the adequacy of the aalyzed parameters, complexity of implemetatio ad itegratio of all the etities i the selectio of appropriate access etwork. Basically, techiques based o the quality of radio liks are represetig simple solutios because of the umber of aalyzed criteria ad etwork techologies. Techiques based o the merit fuctio are aalyzig the impacts of etwork performaces, service requiremets ad users prefereces. However, because of its complexity they are oly represetig theoretical solutios, which are curretly i the focus of may researchers, but for a umber of deficiecies are ot yet applicable i real eviromets. Refereces: [1] J. Akhtma, L. Hazo, "Heterogeeous Networkig: A Eablig Paradigm for Ubiquitous Wireless Commuicatios", Proceedigs of the IEEE, vol. 98, o. 2, Feb. 2010, pp [2] E. Hossai, Heterogeeous Wireless Access Networks: Architectures ad Protocols, (edited volume), Spriger, NY, USA, [3] A. Sgora, D. D. Vergados, "Hadoff Prioritizatio ad Decisio Schemes i Wireless Cellular Networks: a Survey", IEEE Commuicatios Surveys ad Tutorials, vol. 11, o. 4, Fourth Quarter 2009, pp [4] X. Ya, Y. A. Şekercioĝlu, S. Narayaa, "A Survey of Vertical Hadover Decisio Algorithms i Fourth Geeratio ISBN:

6 Heterogeeous Wireless Networks", Computer Networks, vol. 54, o. 11, Aug. 2010, pp [5] A. H. Zahra, B. Liag, A. Saleh, "Sigal Threshold Adaptatio for Vertical Hadoff i Heterogeeous Wireless Networks", Mobile Networks ad Applicatios, vol. 11, o. 4, Aug. 2006, pp [6] K. Yag, et al., "Combied SINR Based Vertical Hadoff Algorithm for Next Geeratio Heterogeeous Wireless Networks", Proc. IEEE GLOBECOM 07, Washigto, DC, USA, Nov. 2007, pp [7] F. Zhu, J. McNair, "Multiservice Vertical Hadoff Decisio Algorithms", EURASIP Joural o Wireless Commuicatios ad Networkig, vol. 2006, o. 2, Apr. 2006, pp [8] N. Nasser, A. Hasswa, H. Hassaei, "Hadoffs i Fourth Geeratio Heterogeeous Networks", IEEE Commuicatios Magazie, vol. 44, o. 10, Oct. 2006, pp [9] N. Nasser, S. Guizai, E. Al-Masri, "Middleware Vertical Hadoff Maager: A Neural Network-based Solutio", Proc. IEEE ICC 07, Glasgow, Scotlad, Jue 2007, pp [10] F. Bari, V. C. M. Leug, "Automated Network Selectio i a Heterogeeous Wireless Network Eviromet", IEEE Network, vol. 21, o. 1, Ja/Feb. 2007, pp [11] B. Bakmaz, Z. Bojkovic, M. Bakmaz, "Network Selectio Algorithm for Heterogeeous Wireless Eviromet", Proc. IEEE PIMRC 2007, Athes, Greece, Sep. 2007, pp [12] M. Kassar, B. Kervella, ad G. Pujolle, "A Itelliget Hadover Maagemet System for Future Geeratio Wireless Networks", EURASIP Joural o Wireless Commu. ad Networkig, Article ID , 12 pages, ISBN:

Implementation of Fuzzy Multiple Objective Decision Making Algorithm in a Heterogeneous Mobile Environment

Implementation of Fuzzy Multiple Objective Decision Making Algorithm in a Heterogeneous Mobile Environment Implemetatio of Fuzzy Multiple Objective Decisio Makig Algorithm i a Heterogeeous Mobile Eviromet P.M.L. ha, Y.F. Hu, R.E. Sheriff, Departmet of Electroics ad Telecommuicatios Departmet of yberetics, Iteret