Inte rnatio nal Jo urnal o f Sc ie ntific & Eng inee ring Re se arc h, Vo lume 3, Issue 1, January -2012 1
ISS N 2229-5518
An Overview on Decision Techniques for Vertical Handoffs across Wireless Heterogeneous Networks
Akhila S, Jayanthi K Murthy, Arathi R Shankar, Suthikshn Kumar
—————————— ——————————
The main attraction of wir eless communication lies in the ability to communicate and exchange information on the move. The demand for the available services anytime anywher e is acceler ating at a very high r ate which calls for an integration of the var ious w ir eless access technologies. With the curr ent technologies varying widely in their bandw idths, latencies, fr equencies and access methods, the next generation systems w ill allow global r oaming among a range of mobile access netw or ks .
This calls for a seamless tr ansfer of the Mobile Terminal
(MT) to the best access link among all available candidates with no per ceivable interruption to an ongoing conver sation[1]. It should also pr ovide an end-to-end optimization that takes into account var iables such as throughput optimization, r outing optimization, delay pr ofiles and economical pr ofitability. The actual tr end is to
integr ate complementary wir eless technologies w ith over lapping cover age, to pr ovide the expected ubiquitous
_ _
Akhila S :Assista nt Professor, Dept of Electronics & Communica tion Engineering, BMS CollegeofEngineering,India (Ema il:bmsa khila@yahoo.co.in)
Ja yanthi .K. Murthy: Assista nt Professor, Dept of Electronics &
Communica tionEngineering,BMSCollegeofEngineering,India( Ema il:jayanthi. ece@bmsce.a c.in)
Ara thi R Sha nka r: Assista nt Professor, Dept of Electronics & Communica tionEngineering,BMSCollegeofEngineering,India( Ema il:ara thi.ec e@bmsce.ac.in)
Suthikshn Kuma r:Professor,Dept of Informa tion Science, P E S Institute of
Technology, Ba nga lore, India (Ema il:suthikshn.kuma r@pes.edu
coverage and to achieve the Always Best Connected (ABC) advantage [2]. The Always Best Connected concept should enable a user , to choose among a host of networ ks that best suits his or her needs and to change when something better becomes available. It r equir es a framewor k that supports mobility management , access discovery and selection, authentication, security and pr ofile server . This calls for an efficient Vertical Handoff Decision(VHO) scheme which involves a tr adeoff among several handoff par ameters such as networ k conditions, system per formance, application types, pow er r equir ements, mobile node conditions, user pr efer ences, secur ity cost and the Quality of Ser vice(QoS). These parameters may have varying levels of importance in the decision pr ocess [3]. Also, the handoff solution should be networ k-layer -transpar ent and infrastructur e- modification-fr ee so that existing Inter net ser ver and client applications can painlessly survive the r apid pace of wir eless technology evolution [ 4].
IJSER © 2012 http :// www.ijser.org
Inte rnatio nal Jo urnal o f Sc ie ntific & Eng inee ring Re se arc h, Vo lume 3, Issue 1, January -2012 2
ISS N 2229-5518
Fig. 1: Hor izontal and ver tical handoff
The handoffs ar e classified into tw o main str eams, Hor izontal Handoff (HHO) and Vertical Handoff (VHO). Figur e1 illustr ates hor izontal and vertical handoff. The main distinction between Ver tical Handoff and Horizontal handoff (HHO) is symmetry.
Table 1: Differ ence betw een Vertical and Horizontal
Handoff.
While HHO is symmetr ic or an intr a-technology based pr ocess, VHO is an asymmetr ic or an inter-technology based pr ocess in which the MT moves betw een two differ ent netw or ks with differ ent character istics [5]. The vertical handoff pr ocess involves thr ee main phases [6] [7], namely system discovery phase, decision phase and execution phase.
Dur ing the system discovery phase, the MT scans for
available candidate networ k for connection which may
include several par ameter s like the supported data r ates and QoS par ameters. This phase needs to be invoked periodically , since the users ar e mobile.
In the decision phase, the mobile ter minal deter mines whether the connections should continue using th e existing networ k or be switched to another netw or k depending on various par ameters like the type of the application (e.g., conver sational, str eaming), minimum bandwidth available, delay constraints, cost, transmit power and the user ’s pr efer ences.
In the execution phase, the connections of the mobile
terminal ar e handed over to the new networ k in a seamless manner . Authentication, author ization, and transfer of a user ’s infor mation is done dur ing this phase.
Handover discovery and decision phase can som etimes over lap, since some situations may r equir e mor e additional pr obing of the networ k condition. A delay in handoff pr ocess can be differ entiated into thr ee main mechanisms [10].
Discovery Time (t d): Dur ing this per iod, the mobile
terminal per ceives its new wir eless networ k range either through the tr igger-based r outer solicitation or waits to r eceive a r outer advertisement fr om an access r outer in the
visited networ k and gets its r outer advertisement (RA)
fr om the new access r outer .
Addr ess Configur ation Interval (t c): Dur ing this per iod, the
mobile device r eceives the Router advertisement and updates its r outing table. A new car e-of- addr ess (CoA) w ill be based on the pr efix of the new r outer that is obtained fr om the RA.
Networ k Registration Per iod (t r): This is the per iod dur ing which the binding update(i.e., the association of home addr ess with a car e-of addr ess) to the home agent as w ell as the corr espondent node is sent and first packet fr om the
corr espondent node is r eceived. Since the binding acknow ledgement fr om corr espondent node is elective, optimizing IP-level ver tical handoff delay would involve minimizing the discovery time and networ k r egistration per iod. The decision phase is the most impor tant phase in VHO since it determines how meaningful the handoff is to the user . This needs an extensive r esear ch to find accurate ways of discover ing pr ecise decision techniques w hich may include one or mor e parameters. The obj ective of this paper is to show how decision par ameters or policies affect VHO. A br ief survey of the var ious decision making techniques used has been pr ovided.
IJSER © 2012 http :// www.ijser.org
Inte rnatio nal Jo urnal o f Sc ie ntific & Eng inee ring Re se arc h, Vo lume 3, Issue 1, January -2012 3
ISS N 2229-5518
FIG 2: System Ar chitectur e of Vertical Mobility
Figur e 2[9] shows the technical featur e of Vertical handoff which descr ibes thr ee main categor ies- Mobility Engineer ing, Resource Management and Service Management .
Resour ce management consists of tw o m
[13].Stevens-Navarr o E, W ong V,‛ Compar ison b etw een
Vert ical Handoff Decision Algorithms for Heter ogeneous Wir eless Networ ks‛, in Pr oc. of IEEE VTC'06, pp. 947 -951 ain parts i.e. dir ect and indir ect r esource allocation in heter ogeneous wir eless networ ks. Dir ect r esource allocation
r efer s to channel and bandw idth allocation wher e as indir ect r esour ce allocation r efer s to networ k capacity and per formance optimization. QoS is dir ectly based on r esour ce allocation wher eas end-to-end QoS needs other managements such as packet’s prior ity in router using header compr ession on wir eless netw or k and packet’s buffer ing in r outers and ter minal. Mobility Engineer ing consists of heter ogeneous netw or ks and services such as mobility management, design and implementation of multiple pr otocols, middlewar e solution in OSI pr otocol stack layer [ 8],[9].Service management offer s interactive mobile applicat ions, location management, mobile services and ser vice life cycle via over-the-air (OTA) pr ovisioning functions used for upgrading and downloading the ser vices.
With the thr ee phases available for per forming a handoff in heter ogeneous netw or ks, the MT will have a choice of several networ ks to which it can connect to. But the outcome of the decision phase of the VHO, which is dependent on several par ameters like available bandw idth, battery power status of the mobile terminal, cost, r eceived signal str ength (RSS), etc. will decide on the netw or k to which a connection will be made. The per formance of the networ k connection also depends, in part , on the signal str ength which also depicts the power pr esent in the r eceived signal. Betw een a MT and access point (AP), the wir eless signal str ength in each dir ection determines the total amount of netw or k bandw idth available along that connection. RSS has a gr eat r ole in the horizontal decision pr ocess due to its compatibility betw een the curr ent attachment point and that of the candidate attachment points. But in VHO, the RSSs ar e incomparable due to asymmetr ical natur e of the heter ogeneous netw or ks. How ever , it can be used to determine the availability as w ell as the condition of differ ent netw or ks . If mor e than one candidate netw or k is available, the MT should associate itself w ith the one having the str ongest RSS as it does in HHO.
Considerable w or k has been done in literat ur e to determine
the appr opr iate parameters that can be consider ed in the decision process for VHO. In [11], the author s have pr oposed a vertical handoff decision (VHD) algorithm that maximizes the overall battery lifetime of the mobile terminal in the same coverage ar ea and also aims at equally distr ibuting the traffic load acr oss the netw or ks. This algor ithm when implemented in multiple Vertical Handoff Decision Contr oller s (VHDC) located in the access networ ks can pr ovide the VHD function for a r egion covering one or multiple APs or BSs. The decision inputs for the VHDCs ar e obtained over the Media Independent Handoff Function (MIHF) of IEEE 802.21. This MIHF facilitates standar ds based message to be exchanged betw een the various access netw or ks (or attachment points) to shar e information about the curr ent link layer conditions, traffic load, netw or k capacities, etc. Though the per formance r esults which ar e based on detailed simulations show that the pr oposed algorithms per form comparatively better , the r eceived signal str ength which is a maj or indicator of the quality of service, should be consider ed to establish the superior ity of the algor ithm. It would have been a better option to study the pr obability of the number of unnecessary handoffs taking place or the pr obability of the number of handoffs being missed leading to incr ease in the pr obability of call dr opping.
In[12], a decision method called ALIVE –HO(adaptive lifetime-based vertical handoff) is pr oposed which is based
on the Received Signal Str ength (RSS).This parameter is
IJSER © 2012 http :// www.ijser.org
Inte rnatio nal Jo urnal o f Sc ie ntific & Eng inee ring Re se arc h, Vo lume 3, Issue 1, January -2012 4
ISS N 2229-5518
used to estimate coverage of the w ir eless networ k and the best netw or k is selected using vertical handoff algor ithms. An adaptive handoff based on the availability of the bandw idth till the time the MT stays in the networ k is
consider ed.
ALIVE- HO algor ithm dynamically adopts to
the Mobile Terminals (MT)velocity to decr ease the
unnecessary number of handoffs and ping pong effect but
the pr obability of handoff incr eases w ith the distance fr om
the AP. It is also established that the number of unnecessary handoffs using ALIVE handoff algorithm is less than that of algor ithms based on tr aditional RSS hyster esis. Accor ding to the authors, the simplest method to incr ease RSS is to incr ease the tr ansmit pow er , which needs further investigation , since an incr ease in transmit pow er might lead to an incr ease in inter fer ence leading to a decr ease in QoS. This might be a practical solution only in open ar eas and may not be feasible in urban ar eas due the clutter ed envir onment , in which case additional parameters need to be consider ed in the decision pr ocess. Stevens et. al [13] have selected parameters such as bandw idth, delay, j itter and bit err or rate (BER) to conduct their compar isons of some of the pr ominent decision algor ithms in literatur e, that is, simple additive w eighting (SAW ), technique for or der pr efer ence by similarity to ideal solution (TOPSIS), multiplicative exponent w eighting (MEW ) and the gr ey r elational analysis (GRA).Good per formance impr ovement of SAW and GRA over several vertical handoff decision algorithms has been obtained. The GRA decision algorithm pr ovided a slightly higher bandw idth and low er delay for inter active and backgr ound traffic classes w hile MEW , SAW and TOPSIS pr ovided almost similar per formance.
The available bandwidth and delay encounter ed has been consider ed as decision parameters by Chuanxiong et al [10]. The per formance of the algorithm is evaluated against throughput and unnecessary handoff rate that is exper ienced dur ing a handoff pr ocess. Unique featur e of this w or k is the capability of r eacting to roaming events pr oactively and accurately with a small handoff delay. The pr oposed system r eacts to r oaming events proactively and accur ately, and also maintains the connections ’ continuity seamlessly . The wor k of [14] employs the use of a cost function involving bandwidth, pow er consumption and financial cost for demonstrating t he per formance of the wor k with r espect to the handoff latency encounter ed. All algor ithms that employ cost functions r equir e manual inputs by the user . This could be a disadvantage since the algor ithm needs to cater to the users r equest as one of the input par ameter and could r esult in poor handoff in the event of any fault in the input . In the dynamic decision model pr oposed by Pramod and Saxena[15] , dynamic factor s like the RSS and velocity of the mobile, and static
factor s like cost, bandw idth and pow er consumption of the mobile terminal has been taken into consideration for making a decision to handoff. This model has been developed using a thr ee phase appr oach namely the prior ity phase, the normal phase, and the decision phase. Selecting the best networ k based on the dynamic factor s is per formed in the pr iority phase. A networ k with highest differ ence between the RSS and the thr eshold RSS is given prior ity. In the normal phase, cost function for each static parameter like cost, bandwidth and power is r ecorded based on their w eight factor s . Then the networ k w ith highest w eight factor is selected. In the decision phase, decision as to which is the best netw or k to handoff is made by obtaining a scor e function i.e., by multiplying the prior ity from the first phase w ith the cost function fr om the normal phase for each of the candidate networ k. The networ k w ith the highest scor e function w ill be the candidate networ k. This model aims at combining both static and dynamic par ameter s to per for m a handoff. Though a r eduction in the number of unnecessary handoff has been established this model is a simple model and is mor e suitable for soft vertical handoffs. How ever the author s need to elaborate on the inter val over which RSS is calculated and how velocity of the mobile is calculated.
Algorithms dealing w ith both hor izontal and vertical handoff scenarios with minimal changes in infr astr uctur e which r equir es deployment of handoff ser vers only on the Internet was pr oposed by Ling-Jyh Chen et al[4]. The Univer sal Seamless Handoff Ar chitectur e (USHA) is an upper layer solution and pr ovides a seamless handoff instead of using new tr ansport pr otocol or new session layer thr ough the middlewar e design str ategy. The handoff, either vertical or hor izontal, occurs on ly on overlaid networ ks using soft handoff technique. USHA may lose connectivity to upper layer application, if the coverage fr om multiple access methods fails to overlap.
Though the concept of vertical handoffs enables integration of netw or ks with complementary featur es, like wide coverage ar ea with low data r at e or limited cover age ar ea with high data r ate, they still have some limitations:
It handles all the connections in same manner .
When all TCP/IP connections ar e automatically
transfer ed fr om one inter face to another , then, only one w ir eless inter face, normally the one w ith end user specific application (the best one) is used at that instant . The second limitation is the need for same networ k inter face. All w ir eless inter faces must be used as part of the same Mobile IP and
IJSER © 2012 http :// www.ijser.org
Inte rnatio nal Jo urnal o f Sc ie ntific & Eng inee ring Re se arc h, Vo lume 3, Issue 1, January -2012 5
ISS N 2229-5518
DNS infrastr uctur e since mobile nodes and peers must be able to r each the Mobile IP and DNS server .
Some networ ks may contain nodes that may not be a part of dir ect peer to peer infr astr uctur e due to which, the ver y importance of w ir eless diversity is lost. Usually the peer to peer connections ar e most efficient because sometimes they offer shortcuts for slow and expensive infrastr uctur e.
Most of the Vertical Handoff decision algor ithms developed use tw o or mor e paramet ers for the per formance analysis of the algor ithms. Some ar e based on the static parameters like access cost, secur ity featur es and power consumption rate since these par ameters maintain a r elatively constant value over a period of time. The algor ithms which ar e based on continuously changing values like available bandwidth, data r ate, Received Signal Str ength and Bit err or r ate ar e said to use dynamic parameters. A decision algorithm gives a better per formance w hen several parameters ar e consider ed, mor e so when a combination of static and dynamic parameters ar e consider ed. But the tradeoff is w ith the incr ease in decision time and complexity of the algor ithm. Consider ing few er decision par ameters, might cause an inaccurate decision leading to poor per formance. Very few algor ithms ar e available in the liter atur e which calculates the pr obability of unnecessary handoffs or handoffs being missed that ar e encounter ed with each of the metr ic consider ed and which of these parameter s when consider ed can lead to r educed number of wr ong decisions to handoff. These conflicting r easons or r equir ements calls for algor ithms which ar e neither too complex nor too simple. Since ther e ar e several algorithms available which ar e based on several par ameters, deciding the best parameters which provides seamless mobility, r educed delay, r educed numb er of unnecessary handoffs’ and deciding on the best algorithm that can be practically implemented becomes difficult. Though RSS is a maj or parameter that needs to be consider ed for obtaining a good QoS, this cannot be easily measur ed owing to the rapid variations in its value over short distances or even small intervals.
The ability to r oam among the available networ ks w ith minimum modification in the infr astr uctur e w ill be the futur e of w ir eless communication for which an efficient Vertical Handoff algor ithm is an essential featur e. It is the decision phase which has maximum r esponsibility for the
whole pr ocess since it is this pr ocess that decides on the number of parameter s to be consider ed along w ith the optimum time that should be allow ed to make a decision. The system capacity and service quality impr ovement can be achieved thr ough an efficient decision algor ithm cost effectively. A survey of the curr ent handoff decision techniques developed w ith various parameter s consider ed and the effect of these par ameter s on the decision pr ocess has been made in this paper . The general implementation structur e of VHO is given and several impor tant static and dynamic parameters that ar e consider ed in the decision pr ocess of Vertical handoff has been emphasized. It can be concluded that a better decision can be obtained by employing as many measur able decision parameters possible, be it static or dynamic.
[1]. Xiaohuan Yany Y. Ahmet S,_ekercio_gluy Sathya Narayanan ‚A Survey of Ver tical Handoff Decision Algorithms in Fourth Generation Heter ogeneous W ir eless Networ ks‛, Elsevier 2010.
[2]. E. Ar un, R.S Moni, ‚A Novel Decision Scheme for Vertical Handoff in 4G W ir eless Networ ks‛, Global Journal of Computer Science and Technology, Vol. 10 Issue 5 Ver .
1.0 July 2010.
[3].YawNkansah-Gyekye, Johnson I Agbinya, ‚Vertical Handoff Decision Algor ithm for UMTS-WLAN‛, The 2nd International Confer ence on Wir eless Br oadband and Ultra Wideband Communications, (AusWir eless 2007).
[4].Ling-Jyh Chen, Tony Sun, and Mar io Ger la ‛USHA: A Practical Vertical Handoff Solution‛ Multimedia Services Access Networ ks, 2005. MSAN '05.
[5].Nishith DT, Nortel JHR, Hugh FV.‛ Handoff in cellular systems‛, in IEEE Per sonal communications, 1998 .
[6]. J. McNair and F. Zhu, ‚Vertical Handoffs in Four th- generation Multinetw or k Envir onments,‛ IEEE Wireless Communications, vol. 11, no. 3, June 2004.
[7]. W. Chen and Y. Shu, ‚Active Application Or iented Vertical Handoff in Next Gener ation Wir eless Networ ks,‛ in Proc. of IEEE ,WCNC’05,.
[8] Xiaoxin Wu, Mukherj ee B & Ghosal D,‛ Hier archical arc hitectur es in thethir dgeneration- cellular Netw or k‛, IEEE Wir eless Communications 2004
[9] Mika Ylianttila, “Vertical handoff and mobility system ar chitectur e and transition analysis”,
IJSER © 2012 http :// www.ijser.org
Inte rnatio nal Jo urnal o f Sc ie ntific & Eng inee ring Re se arc h, Vo lume 3, Issue 1, January -2012 6
ISS N 2229-5518
Univer sity of Oulu, 2005
[10].Raj iv Chakravorty ,Pablo Vidales ,Kavitha Subr amanian , Ian Pratt , Jon Cr owcr oft ‚Per formance Issues with Ver tical Handoffs – Exper iences fr om GPRS Cellular and W LAN Hot -spots Integration‛, Per vasive Computing and Communication s, 2004. Per Com 2004.
[11].SuKyoung Lee, Kotikalapudi Sr iram, Kyungsoo Kim, Yoon Hyuk Kim, and Nada Golmie,‛Vertical Handoff Decision Algor ithms for Pr oviding Optimized Per formance in Heter ogeneous Wir eless Networ ks‛, IEEE tr ansactions on vehicular technology, January 2009
[12]. Muhammad Amir Latif, Abid Aziz Khan‛ quality of ser vice during vertical handoff in 3g/4g wir eless networ ks ‛, Blekinge Institute of Technology, 2009
[14]. W ang HJ, Katz RH, Giese J Policy,‛Enabled Handoffs acr oss Heter ogeneous W ir eless Networ ks‛, Proc. of ACM W MCSA. ‘99.
[15]. Pr amod Goyal, and S. K. Saxena ‚A Dynamic Decision Model for Vertical Handoffs acr oss Heterogeneous W ir eless Networ ks‛, W orld Academy of Science, Engineering and Technology Vol. 31, July 2008.
IJSER © 2012 http :// www.ijser.org