A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS



Similar documents
M3S MULTIMEDIA MOBILITY MANAGEMENT AND LOAD BALANCING IN WIRELESS BROADCAST NETWORKS

An Interest-Oriented Network Evolution Mechanism for Online Communities

APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT

A Replication-Based and Fault Tolerant Allocation Algorithm for Cloud Computing

INVESTIGATION OF VEHICULAR USERS FAIRNESS IN CDMA-HDR NETWORKS

Efficient Bandwidth Management in Broadband Wireless Access Systems Using CAC-based Dynamic Pricing

A Dynamic Energy-Efficiency Mechanism for Data Center Networks

Cooperative Load Balancing in IEEE Networks with Cell Breathing

A Secure Password-Authenticated Key Agreement Using Smart Cards

Performance Analysis of Energy Consumption of Smartphone Running Mobile Hotspot Application


An RFID Distance Bounding Protocol

PAS: A Packet Accounting System to Limit the Effects of DoS & DDoS. Debish Fesehaye & Klara Naherstedt University of Illinois-Urbana Champaign

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING

Open Access A Load Balancing Strategy with Bandwidth Constraint in Cloud Computing. Jing Deng 1,*, Ping Guo 2, Qi Li 3, Haizhu Chen 1

What is Candidate Sampling

Dimming Cellular Networks

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

QoS-based Scheduling of Workflow Applications on Service Grids

VRT012 User s guide V0.1. Address: Žirmūnų g. 27, Vilnius LT-09105, Phone: (370-5) , Fax: (370-5) , info@teltonika.

AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Project Networks With Mixed-Time Constraints

A Dynamic Load Balancing for Massive Multiplayer Online Game Server

Complex Service Provisioning in Collaborative Cloud Markets

DEFINING %COMPLETE IN MICROSOFT PROJECT

iavenue iavenue i i i iavenue iavenue iavenue

Reinforcement Learning for Quality of Service in Mobile Ad Hoc Network (MANET)

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network *

A Design Method of High-availability and Low-optical-loss Optical Aggregation Network Architecture

Performance Analysis and Comparison of QoS Provisioning Mechanisms for CBR Traffic in Noisy IEEE e WLANs Environments

A role based access in a hierarchical sensor network architecture to provide multilevel security

VoIP Playout Buffer Adjustment using Adaptive Estimation of Network Delays

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

Energy Conserving Routing in Wireless Ad-hoc Networks

A Performance Analysis of View Maintenance Techniques for Data Warehouses

J. Parallel Distrib. Comput. Environment-conscious scheduling of HPC applications on distributed Cloud-oriented data centers

An Adaptive Cross-layer Bandwidth Scheduling Strategy for the Speed-Sensitive Strategy in Hierarchical Cellular Networks

Frequency Selective IQ Phase and IQ Amplitude Imbalance Adjustments for OFDM Direct Conversion Transmitters

denote the location of a node, and suppose node X . This transmission causes a successful reception by node X for any other node

A hybrid global optimization algorithm based on parallel chaos optimization and outlook algorithm

A Novel Adaptive Load Balancing Routing Algorithm in Ad hoc Networks

Analysis of Energy-Conserving Access Protocols for Wireless Identification Networks

An Integrated Approach of AHP-GP and Visualization for Software Architecture Optimization: A case-study for selection of architecture style

Politecnico di Torino. Porto Institutional Repository

A 2 -MAC: An Adaptive, Anycast MAC Protocol for Wireless Sensor Networks

QoS in the Linux Operating System. Technical Report

A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION. Michael E. Kuhl Radhamés A. Tolentino-Peña

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

An Alternative Way to Measure Private Equity Performance

A Game-Theoretic Approach for Minimizing Security Risks in the Internet-of-Things

Traffic State Estimation in the Traffic Management Center of Berlin

Cost Minimization using Renewable Cooling and Thermal Energy Storage in CDNs

IWFMS: An Internal Workflow Management System/Optimizer for Hadoop

An ILP Formulation for Task Mapping and Scheduling on Multi-core Architectures

Enterprise Master Patient Index

Rapid Estimation Method for Data Capacity and Spectrum Efficiency in Cellular Networks

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College

How To Solve An Onlne Control Polcy On A Vrtualzed Data Center

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

Load Balancing Based on Clustering Methods for LTE Networks

A New Paradigm for Load Balancing in Wireless Mesh Networks

An Introduction to 3G Monte-Carlo simulations within ProMan

Availability-Based Path Selection and Network Vulnerability Assessment

"Research Note" APPLICATION OF CHARGE SIMULATION METHOD TO ELECTRIC FIELD CALCULATION IN THE POWER CABLES *

Network Services Definition and Deployment in a Differentiated Services Architecture

FOURTH Generation (4G) Wireless Networks are developed

Efficient Project Portfolio as a tool for Enterprise Risk Management

Research of Network System Reconfigurable Model Based on the Finite State Automation

Resource Scheduling in Desktop Grid by Grid-JQA

Ad-Hoc Games and Packet Forwardng Networks

Cloud-based Social Application Deployment using Local Processing and Global Distribution

The OC Curve of Attribute Acceptance Plans

A New Task Scheduling Algorithm Based on Improved Genetic Algorithm

Load Balancing Algorithm of Switched Dynamic Iteration

Optimization Model of Reliable Data Storage in Cloud Environment Using Genetic Algorithm

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

Traffic-light a stress test for life insurance provisions

Mining Multiple Large Data Sources

Multi-Source Video Multicast in Peer-to-Peer Networks

Brigid Mullany, Ph.D University of North Carolina, Charlotte

Enabling P2P One-view Multi-party Video Conferencing

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy

Transcription:

A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS Shanthy Menezes 1 and S. Venkatesan 2 1 Department of Computer Scence, Unversty of Texas at Dallas, Rchardson, TX, USA 1 shanthy.menezes@student.utdallas.edu 2 venky@utdallas.edu ABSTRACT Next generaton networks are defned to be packet based networks that provde telecommuncaton servces to users by utlzng dfferent transport technologes, wred and wreless. In ths paper we provde a generc framework for handover decson management n next generaton networks. We show that any handover decson algorthm can utlze our framework. We show the feasblty of applyng our framework through an mplementaton example. One of the man features of NGNs s applcaton adaptablty; the applcatons desgned for NGN systems have to be adaptable to dfferent networks and dfferent devces. We propose a handover decson algorthm that utlzes our generc framework and selects the network for handover such that the qualty of experence of the user s near optmal. Through smulatons we show that our algorthm reduces unnecessary handovers and also satsfes the handover requrements. KEYWORDS Next Generaton Network, Wreless Network, Moblty, Handover Decson 1. INTRODUCTION Next generaton networks (NGNs) are expected to support a varety of the exstng wred/wreless access network technologes as well as new wreless access technologes to support hgh data rates up to 1Gbps. The network envronment envsoned for NGN s made up of the varous wred and wreless access networks connected to a common NGN core network. Fgure 1 shows a typcal settng of a NGN. Snce the evoluton towards NGNs s drven by the ever ncreasng moble subscrber base and the set of moble applcatons, moblty management (MM) s an essental requrement for NGN users. Hence the Telecommuncaton Standardzaton Sector (ITU-T) has descrbed ffteen general moblty management requrements for NGNs [1]. One of the requrements s to support polcy-based and dynamc network selecton. Network selecton s the process of choosng the access network to connect to. The network selecton process s done ntally at the tme of startng a servce (applcaton) that requres network connecton and durng handover of the servce to a new network. Durng handover, the selecton process s known as the handover decson process; that s, decdng whether to dsconnect from the current network and decdng whch network to handover to. Durng handover n NGN, the moble devce has the opton of handng over to the access network whch s of the same rado access technology (RAT) as ts current network (horzontal handover) or to an access network of a dfferent RAT (vertcal handover). Tradtonally a handover s performed when the current network does not meet the requrements of the moble devce. In the case of NGNs the handover process poses a varety of challenges due to the dfferent set of applcatons used, networks avalable, qualty of servce (QoS) avalable, and devces used. In ths envronment handover can be performed not only because the current DOI : 10.5121/cnc.2011.3204 51

network does not meet the requrements, but also due to the avalablty of a better network or devce. Servces and applcatons Cellular 2.5G New RAT NGN Core Wred xdsl WMax Cellular 3G Cellular 4G WLAN BlueTooth Fgure 1. Archtecture of Next Generaton Networks The requrements of horzontal handovers were prmarly confned to the rado lnk requrements such as rado sgnal strength and bt error rate of the channel. In the NGN scenaro, the varety of access networks avalable dffer n QoS provson, securty schemes offered, cost of servce etc. The capabltes of the moble devces also vary n many aspects such as power consumpton, applcaton support and so forth. The key goal for NGNs s to enhance the user experence. Ths requres the provson of the most sutable rado access network for any gven user. Thus the handover decson has to be managed such that the requrements of the user are satsfed optmally. In ths paper we propose a generc framework for handover decson management whch has an extensve lst of requrements for handover that optmze the moble user experence. We elaborate on actual mplementaton of our framework by utlzng the IEEE 802.21 meda ndependent handover standard. In NGNs, n order to be able to offer servces anywhere and anytme the servces and applcatons have to be desgned such that they are adaptable to the multtude of access networks avalable. We propose a handover decson algorthm that takes nto consderaton the needs of adaptable applcatons and also provdes flexblty n specfyng the user and network requrements. Through smulatons we show that our algorthm satsfes user requrements and also reduces unnecessary handovers. The rest of the paper s organzed as follows. Secton 2 dscusses related work n the area of handover decson algorthms and framework sutable for NGNs. In secton3 we present our generc framework for handover decson management. In secton 4 we show how our framework can be used to represent any exstng or future handover decson algorthm through three examples consstng of a decson algorthm for horzontal handover and two of the most cted decson algorthms for vertcal handover. Secton 5 presents our proposed generc handover decson algorthm. In secton 6 we present an example mplementaton of our generc framework and handover decson algorthm. Secton 7 provdes the smulaton detals and analyzes the smulaton results. Fnally secton 8 provdes the summary of the paper. 52

2. RELATED WORK Varous solutons [2,3,5,7,8,9,12,13,14,15] for handover management n heterogeneous network envronment have been proposed n the lterature. Several solutons [2, 3, 14, 15] utlze the servces provded by the IEEE standard 802.21 [4] to facltate a handover decson. The 802.21 framework can provde the parameters requred for a handover decson; an actual handover decson management module s not specfed n the standard. Cacace and Vollero [2] provde a detaled moblty management framework that uses the 802.21 standard to provde servces to the applcatons to adapt to the new networks avalable. Ther soluton provdes nput to the moble IPv6 protocol to mplement a handover wthout any packet loss. The moblty management frameworks provded n [2,3,5] do not address the handover decson that decdes the lnk layer handover. Yungkyu and Cho [6] provde solutons for handover between 802.16 and 802.11 networks that maxmze the energy usage and mnmze the cost requred. These solutons are not unversally applcable snce not all the obectves for heterogeneous handovers are consdered. A varety of approaches [7,8,9] based on the theory of mult-attrbute decson makng (MADM) are generc n nature. We elaborate on these algorthms n secton.4 and also compare them to our handover decson algorthm usng smulatons. An extensve framework for handover decson management s not provded n these solutons [7,8,9]. Any of the exstng handover decson algorthms can be represented by our proposed framework. In secton 4 we gve examples of a horzontal handover decson algorthm and the two most cted vertcal handover decson algorthms [7,9], and demonstrate how any handover decson algorthm can be represented usng our generc framework. Due to the vast number of choces for access networks, a large dscrepancy between the values offered by dfferent networks s a defnte possblty. The handover decson algorthm presented here addresses ths ssue wth the ad of the framework provded. Our algorthm also prevents unnecessary handovers, unlke the other generc vertcal handover algorthms. 3. HANDOVER DECISION MANAGEMENT FRAMEWORK We present a handover decson management framework as shown n Fgure 2. Any handover decson algorthm, whether for horzontal handover or vertcal can be bult upon our generc management framework. The vtal part of our framework s a handover decson algorthm. The other parts of the framework are elements that affect the handover decson, used as nputs by the handover decson algorthm. The most nfluental of these nputs are the Parameters. Parameter values are nfluenced by the handover decson, namely the access network chosen. The parameters portray the obectves set by varous enttes that get affected by handover. The enttes specfy the requrements for the parameters n terms of weghts and thresholds. The weghts specfy the mportance of the parameters to the entty. The thresholds specfy the condtons to be met for the parameters. The exstng handover decson algorthms only consder one type of threshold that affects the handover decson, whch we term mandatory threshold. In order to consder the adaptve nature of the servces offered by NGNs, we consder another threshold for each parameter called the preferred threshold. The other feature of the framework s the values offered by the varous networks for each of the parameters specfed. The bass of any handover decson algorthm s a comparson of these offered values to the parameter requrements. Also as part of the framework, are the parameters used by handover decson algorthms to avod unnecessary handovers. In the next secton we show that our framework can be used by any exstng handover decson algorthm and we llustrate the utlty through examples. If, due to future enhancements or restrctons to the NGNs the handover decson module requres nputs not represented by our framework, our framework can be expanded to nclude these nputs termed as future nputs. 53

Fgure 2. Handover Decson Management Framework Our framework can be represented n vector form as follows: 3.1. Enttes The enttes that provde the parameters are denoted by the vector <E 1,,E M >. 3.2. Parameter Requrements The parameters are denoted by <O 1,O 2,,O P >. Parameter requrements are gven n terms of weghts and thresholds. Weghts: The weghts specfy the prortes gven to each parameter. The hgher the weght of the parameter, the larger s ts nfluence n the handover decson. Each entty provdes weghts for ts own parameters by prortzng them, these weghts are called local weghts. The handover decson algorthm can prortze the enttes by provdng weghts to the enttes. The global weghts used by the algorthms are the products of the local weghts and the weghts of the respectve enttes. For example let <W 1,,W m > be the local weghts of the m parameters set by entty and <W 1,,W n > be the local weghts of the n parameters set by entty. If the weghts of and are W and W respectvely, then the global weghts are <W 1 *W,,W m *W,W 1 *W,,W n *W >. We denote the global weghts of the parameter as W = < W 1,W 2,,W P > Thresholds: In most of the handover decson algorthms n lterature [7,8,9], the threshold values of the parameters are only consdered n elmnatng networks durng the decson process. That s, f the network s offered value for a parameter does not meet a threshold, then t s not consdered for handover. Snce some of the requrements such as user requrements and servce requrements are adaptable, a sngle threshold s not applcable n the case of handover n NGN networks. We propose the use of two types of threshold values: 1. Mandatory Thresholds: For a network to be elgble for selecton, the network s offered value for a partcular parameter should satsfy these thresholds. The mandatory thresholds are represented by the vector µ=<µ 1,µ 2,,µ P >. These values could be lower sded, upper sded or exact dependng on whether the value of the parameter should be,, or equal to the threshold. 54

2. Preferred Thresholds: The preferred thresholds represent the values desred by users and servces for ther best performance. The preferred thresholds are ether lower sded or upper sded. These may not be specfed for all parameters. These are represented as vector π=<π 1, π 2,,π P > 3.3 Parameter Values The values offered for each of the parameters by the access networks specfes the contrbuton of the network for each of these parameters. The values offered by an access network s denoted as the vector R = <R 1,R 2,,R P >. From the parameter values offered by the networks, we form two sets of vectors: (1) hghest offered value H = <H 1,H 2,,H P > denotng the best offered value, and (2) lowest offered value L = <L 1,L 2,,L P > denotng the worst offered. If parameter O has a lower sded threshold, then H = max R and L = mn R. If O has an upper sded threshold then H = mn R and L = max R, otherwse H = L = R. 3.4 System Parameters for Unnecessary Handover Avodance If the handover decson s not mplemented effcently, t can result n unnecessary handovers that can also lead to the png-pong effect. Hysterss value and dwell tmers are two such parameters utlzed n horzontal handover decson algorthms. We represent the set of these system parameters by Σ. 4. UTILITY OF THE FRAMEWORK Most handover decson algorthms can be dvded nto two phases () part of the algorthm that makes the decson to perform a handover and () part of the algorthm that chooses the network to handover to. The frst phase s referred to as the handover ntaton algorthm and the second phases s referred to as the network selecton algorthm. In ths secton we lst some of the exstng handover decson algorthms and show how they can be represented as functons of our framework. Horzontal handover algorthm: One of the ntal handover decson algorthms used for horzontal handover part of The North Amercan Personal Access Communcaton Systems (PACS) combnes a hysteress value wth dwell tmer to reduce the number of unnecessary handovers [10]. Ths basc algorthm can be splt nto ntaton algorthm and network selecton algorthm. Intaton algorthm: If R c1 µ 1 and R c2 µ 2 for T amount of tme then handover to the network selected by the network selecton algorthm. The ntaton algorthm can be represented as f(r c1, R c2, T, µ 1, µ 2 ), where c represents the current network, R c1 s the RSS measured and R c2 s the bt error rate measured. Network selecton: For the network selecton, the mandatory thresholds are computed as µ 1 = R c1 + δ 1 and µ 2 = R c2 + δ 2, δ 1 and δ 2 are hysteress values used to avod unnecessary handover. If R 1 µ 1 and R 2 µ 2, for any detected neghborng network, then handover to network. The network selecton algorthm can be represented as f(r, µ 1, µ 2, Σ) Smple addtve weghng (SAW) algorthm: Varous vertcal handover algorthms are proposed n lterature based on SAW [8]. SAW s a network selecton algorthm. Accordng to SAW the selected network s the one satsfyng max M P = 1 W R, where M s the set of all elgble networks, W s weght of parameter, and R s the value offered by network for parameter. Ths algorthm can be represented as f(w,r). 55

Multservce vertcal handover: A varaton of SAW proposed by Zhu and McNar [7] s to s use the cost functon for each network gve by: C =, where C s the cost ncurred by choosng network for servce s and s gven as C s s = W s C η ( R ). The functon η s used to normalze the values offered. Only those R values that satsfy the mandatory thresholds µ are consdered. Whenever a new network s detected, the cost functon s evaluated for all networks and the network wth the mnmum cost s selected. Hence the handover selecton s done whenever a new low cost network s detected. The authors also present a per servce handover algorthm. The servces are prortzed, and each servce n the order of the prortes s handed off to the network wth the least cost. The assumpton here s that the moble devce s communcatng usng multple rado nterfaces smultaneously. Ths algorthm can be represented as f(µ, W, R). Gray relatonal analyss (GRA) algorthm: The GRA algorthm presented by Song and Jamalpour [9] lsted n Fgure 3 s based on grey relatonal analyss (GRA). GRA s used to analyze relatonshp between dfferent dscrete vectors. A grey relatonal coeffcent (GRC) s computed between each of the vectors and a reference vector known as the deal vector. The vector wth the largest GRC s selected. GRA Algorthm (µ, W, R, H, L) 1 Normalze the vectors R nto a vector =( 1,, ) 2 Calculate the GRC for each network as = 3 Select the network wth mnmum GRC Fgure 3 GRA Algorthm 1 =1 1 +1 Here R * R = H L L f parameter O has a lower sded threshold and network selecton algorthm can be represented as f(µ,w,r,h,l). R * = L L R H. Ths 5. GENERIC HANDOVER DECISION ALGORITHM Clearkt no decson algorthm consders the preferred threshold values of adaptable parameter requrements. In ths secton we descrbe a handover ntaton and network selecton algorthm that can optmze the user qualty of experence by consderng the requrements for the best operaton of the applcaton/servce used. 5.1 Defntons For a network, and a parameter, we quantfy dgresson from preferred threshold as D : D = 0 f preferred threshold s satsfed,.e f R π or R π dependng on whether the threshold s lower sded or upper sded respectvely. If the preferred threshold s not satsfed then the dgresson s gven as: D = (π - R )/π, f threshold s lower sded. D = (R - π )/π, f threshold s upper sded. 56

The dgresson for a network s gven by D = D. 5.2. Algorthm Our handover decson algorthm conssts of two phases, as descrbed n Fgure 4. P = 1 Fgure 4 Handover Decson Algorthm 5.2.1. Phase I: Handover Intaton Algorthm Handover s ntated f one of the two condtons s met: Condton 1: The current network has not met the threshold requrements for T. That s: there exsts at least one parameter (1 P), such that R c < π or R c > π for lower sded or upper sded mandatory threshold respectvely for at least T amount of tme. Condton 2: Condton 1 s not true, but one or more new networks are detected and for at least one of the new networks, D < D c, where D c s the newly calculated dgresson for the current network. 5.2.2. Phase II: Network Selecton Algorthm Calculate D for each of the detected networks and select the network wth the mnmum D value. If handover was ntated due to condton 2 n phase I, then handover to the network wth mnmum D value only f the dgresson for current network D c < D. We also propose a varaton of the handover decson algorthm descrbed above, whch ncorporate the weghts of parameters n calculatng D. Each parameter has a weght 57

assocated wth t. Usng the weghts the dgresson for network s computed as: D = P = 1 W D. 6. FRAMEWORK IMPLEMENTATION In ths secton we provde an example mplementaton of our framework. Fgure 5 depcts ths example. The handover decson algorthm resdes n the moble devce n ths partcular example, but t s possble to have the handover decson algorthm n a network element n the case of a network controlled handover. In ths example we utlze ze the IEEE 802.21 standard [4] known as Meda Independent Handover (MIH). Ths s the standard for optmzaton of handovers between IEEE 802 systems and cellular systems. Fgure 5 Framework Implementaton Example The enttes provdng the parameters for the handover decson are: MAC and PHY layer: The PHY layer parameters used are sgnal-to-nose nose rato (SNR), BER, and receved sgnal strength (RSS). Only the mandatory thresholds are consdered for the PHY layer. The MAC layer (lnk layer) parameters specfy the parameters measured by the MAC layers, such as the peak rate, packet error rate, tter, etc. The mandatory and preferred thresholds set for these parameters are dependent on the applcaton used. Due to the dfferent access networks supported, the MAC and PHY layer are specfc to the access technology (AT) of the networks, such as 802.3 or rado access technologes such as 802.11, 802.16 or UMTS. Applcaton: The applcaton entty provdes the parameters that affect the performance of the applcaton/servce used by the moble devce. In ths example we use the followng applcaton parameters: QoS Requrements: The mplementaton can utlze the QoS parameters specfed n the 802.21 standard: data rate, mnmum packet transfer delay, maxmum packet transfer delay, tter, 58

packet loss rate, and packet error rate [4]. An applcaton requres all or a subset of these parameters. The applcaton provdes thresholds and weghts for each of the parameters t specfes. We consder both mandatory and preferred thresholds for these parameters n ths example. User: The user provdes a profle that ncludes the parameters of: Cost: Ths requrement specfes the monetary cost that the user s ready to ncur due to a connecton. Securty: Wth ths requrement the user specfes the authentcaton related requrements. For example, the authentcaton requrements can be gven as dfferent levels where level 0 ndcates open authentcaton, level 1 ndcates a shared key based authentcaton wth no encrypton, and level 2 may ndcate shared key based authentcaton wth encrypton etc. Now, we elaborate on the operaton of the handover decson algorthm usng ths example framework. Fgures 6a and 6b show handover ntaton due to the PHY layer mandatory thresholds not satsfed and due to the applcaton QoS requrements not met. The handover decson algorthm s notfed that the PHY layer mandatory thresholds are not met through the event servce provded by the 802.21 MIHF. A protocol that wshes to use the MIH event servce sends the message MIH_Event_Subscrbe.request [13] to the MIHF as shown n Fgure 6a. In the case of the PHY layer mandatory thresholds, ths s acheved by subscrbng to the predctve event Lnk_Gong_Down of the current lnk. Ths event s trggered by the current MAC and PHY layer, whenever the mandatory thresholds of PHY layer parameters of ths lnk are not met. Fgure 6a. Notfcaton of PHY layer Performance In order to get notfed whenever the mandatory thresholds of the QoS requrements are not satsfed, we utlze the MIH_Lnk_Confgure_Thresholds command [4] as shown n fgure 6b. The 802.21 standard provdes the mappng of the specfed QoS parameters to AT specfc parameters. When the MIHF receves the MIH_Lnk_Confgure_Thresholds request, t maps the QoS parameter thresholds to the current lnk layer parameters, and sends the command Lnk_Confgure_Thresholds to the current lnk layer. Ths results n the lnk layer notfyng the MIHF usng Lnk_Parameters_Report event whenever the specfed thresholds are not met. 59

Fgure 6b Notfcaton of Applcaton QoS Deteroraton In Fgure 7, we show when the network selecton algorthm s trggered and how t acqures the network offered values. When the moble node (MN) frst connects to any network, t acqures nformaton about neghbor networks through the MIH_Get_Informaton.request message of the 802.21 meda ndependent nformaton servce (MIIS). The 802.21 standard specfes a MIIS server that serves a group of neghborng networks and has statc knowledge of the network parameters such as the network dentfcaton, cost of servce, and securty offered. The MIIS server responds to the nformaton request from the MIHF of the MN wth ths nformaton. Ths provdes the values offered for cost and securty parameters. As descrbed n secton 5.2, one of the condtons for handover ntaton s that a neghborng network s detected. Ths s acheved n the example through the MIHF event Lnk_detected. The other condton to ntate handover s whenever the mandatory thresholds of any parameters are not satsfed. As descrbed above the events Lnk_gong_down and Lnk_Parameters_Report notfy that the mandatory thresholds of the PHY layer parameters and the applcaton QoS parameters respectvely are not met. When handover s ntated, the handover decson algorthm acqures the network offered values of all the avalable lnks for the specfed parameters through the request MIH_HO_Canddate_Query to the MIHF. The MIHF n turn requests the nformaton from the current network whch n turn requests the nformaton from the gven neghborng networks. Wth ths nformaton, the network selecton algorthm has all the requred values to calculate the dgresson of all the neghborng networks and select a network to handover to. 60

7. SIMULATION AND RESULTS Fgure 7 Network Selecton Procedure We smulated our handover decson algorthm and the SAW algorthm as used by Zhu and McNar [7] and the GRA algorthm as gven by Song and Jamalpour [9]. 7.1. Smulaton Model For our smulatons, we used a dscrete event smulator developed by us usng the Mcrosoft Vsual Studo.Net C++ to smulate moblty and handovers. We have used a network area of 2000 square meters populated by three dfferent access networks (ANs) as shown n Fgure 8.We use shadow fadng wth 6dB standard devaton to calculate the RSS at the moble node. The specfcatons of the three ANs for ths calculaton are gven n Table 1. Table 1. Network Specfcatons for RSS Calculaton AN 1 AN 2 AN 3 Frequency (MHz) 2600 2300 1900 Transmt Power (dbm) 50 40 33 Coverage Radus (meters) 1000 500 250 61

Fgure 8 Smulaton Network Model The parameters used for handover decson are the receved sgnal strength (RSS), data rate offered, and the cost of AN. Only the mandatory threshold s consdered for the RSS parameter; for the data rate and cost parameters both mandatory and preferred thresholds are consdered. Whle the RSS value s dynamc and changes wth the locaton of the moble node, data rate and cost are statc parameters. We have used the random waypont moblty model [11], where ntally the moble node s at a random poston wthn the smulaton area. Subsequently, the MN randomly chooses a new locaton and moves to ths locaton wth a speed of 5m/s. The smulaton s done for two scenaros:. In the frst scenaro there s a large dscrepancy between the values offered by the three access networks and only one network satsfes the preferred threshold hold for the cost parameter. Table 3 shows the values offered by the networks. We can see that n the frst scenaro the dfference between the data rate offered by the networks AN 1 and AN 3 s vastly hgher than the data rate of 8Mbps offered by AN 2. Only AN 2 satsfes the preferred threshold of 0.3 for cost n ths scenaro.. In the second scenaro the dscrepancy between the values offered s less and two of the three avalable networks satsfy the preferred threshold. As can be seen from Table 3, n ths scenaro the data rates offered by the three networks dffer at most by 150 Mbps and access networks 1 and 2 both satsfy the preferred thresholds for both data rate and cost. We provde the detals of the requrements for the parameters and the values offered by the networks for each of the scenaros n Tables 2 and 3. Cost s gven as a normalzed value usng the hghest cost charged by a network. For example f the hghest cost charged s by a network s $100, then the cost value for AN 1 s 0.6 f ths network charges $60. 62

Table 2. Parameter Requrements. Mandatory Threshold Preferred Threshold Data rate Cost Data rate Cost Scenaro 1 4Mbps 0.6 8Mbps 0.3 Scenaro 2 25Mbps 0.6 50Mbps 3 Data Rate Table 3. Parameter Values Cost AN 1 AN 2 AN 3 AN 1 AN 2 AN 3 Scenaro 1 100 Mbps 8 Mbps 200 Mbps 0.6 0.3 0.4 Scenaro 2 100 Mbps 50 Mbps 200 Mbps 0.6 0.3 0.3 7.2. Results and Analyss We present the results for both the scenaros n terms of the number of handovers occurred and the percentage of handovers that satsfes the preferred thresholds. In the results gven below we are only recordng those handovers that occurred when more than one AN was avalable for selecton, and when AN 2 was one of those networks. Ths s done n order to evaluate the behavor of the algorthms at tmes when a network satsfyng preferred thresholds s avalable. Fgure 9 compares the total number of handovers and Fgure 10 gves the percentage of tmes the preferred thresholds were met after the handovers. 90 T o t a l H a n d o v e r s 80 70 60 50 40 30 20 10 Our algorthm SAW algorthm GRA algorthm 0 Scenaro 1 Scenaro 2 Fgure 9 Smulaton Results: Total Handovers 63

The results show that by consderng the preferred thresholds, our algorthm avods unnecessary handovers. In Fgure 9, we observe ths for both the scenaros, even though n the second scenaro both networks AN 1 and AN 2 satsfy the preferred thresholds. Ths s due to the fact that the MN executes handover only f the current network does not satsfy the preferred thresholds. Hence by utlzng preferred thresholds for parameters we reduce the handovers by more than 25%. Handover executon not only requres resources of the user devce and networks, but also the bandwdth whch s lmted for wreless access networks. Hence, by reducng the unnecessary handovers through our handover decson algorthm, we have acheved resource effcency. 120 S a t s f e d R e q u e s t s % 100 80 60 40 20 Our algorthm SAW algorthm GRA algorthm 0 Scenaro 1 Scenaro 2 Fgure 10. Percentage of Preferred Requests Satsfed In the frst scenaro, only AN 1 can satsfy the preferred thresholds of all parameters. The preferred thresholds are satsfed durng hundred percent of the handovers when our algorthm s used as shown n Fgure 10. Ths s due to the fact that by usng dgresson our algorthm always chooses the network that most satsfes the preferred thresholds. In the second scenaro where the two networks AN2 and AN3 satsfy the preferred thresholds, SAW based algorthm also satsfes the preferred thresholds. Avalablty of dfferent access networks offerng values that dffer largely s a defnte possblty n NGNs. We have shown that n such scenaros our algorthm satsfes the parameter requrements whenever possble. 8. CONCLUSIONS The goal of the next generaton networks s to provde servces to users utlzng a packet based core network whch s able to ntegrate a varety of transport technologes, wred and wreless. The ITU-T standardzaton body has publshed requrements for the NGNs. In ths paper we have consdered the handover decson problem, whch s complex due to the avalablty of dfferent access networks. Next generaton networks not only offer the choce of a varety of access networks but also a myrad of applcatons to users. A key feature of the applcatons desgned for NGNs s ther adaptablty to the dverse networks offered. The overall user experence depends not only on the performance of the applcatons, but also on the cost benefts and the devce performance. We have presented a generc framework for handover decson n NGNs usng the concept of preferred thresholds for the requrements of parameters that affect the handover. Through examples we showed that our framework can be utlzed by any exstng handover decson algorthm for horzontal or vertcal handover. 64

We have presented a handover decson algorthm that utlzes the concept of preferred thresholds and hence represents the adaptve nature of the NGN servces. Through smulatons we have compared our algorthm wth the vertcal handover decson algorthms based on SAW [25] and GRA [27]. We observe that our algorthm s able to satsfy the user requrements and, n scenaros where there s a consderable dscrepancy n the values offered by networks for the dfferent parameters, only our algorthm pcks the network that most satsfes the user preference durng all handovers. The results showed that n all scenaros we reduce handovers by more than 25%, thus mprovng resource effcency. REFERENCES [1] ITU-T Recommendaton Q.1706 Next Generaton Networks: Generalzed Moblty Internatonal Telecommuncaton Unon, 2006 [2] F. Cacace and L. Vollero, Managng Moblty and Adaptaton n Upcomng 802.21 Enabled Devces, Proc. 4th Int l. Wksp. Wreless Moble Apps. and Servces on WLAN Hotspots, 2006 [3] G. Lampropoulos, A. Salkntzs, and N. Passas, Meda Independent Handover for Seamless Servce Provson n Heterogeneous Networks, IEEE Commun. Mag., vol.46, no. 1, Jan. 2008. [4] IEEE Std. 802.21-2008. IEEE standard for Local and metropoltan area networks- Part 21: Meda Independent Handover [5] 3gPP TS23.060 v 6.9.0, General packet rado servce. 3GPP specfcatons [6] Y. Cho and S. Cho, Servce charge and energy-aware vertcal handoff n ntegrated IEEE 802.16e/802.11 networks, Proc. IEEE INFOCOM, 2007, pp. 589 597. [7] Fang, Z., McNar, J.: Multservce Vertcal Handoff Decson Algorthms. EURASIP Journal on Wreless Communcatons and Networkng 2006(2), 52 64 (2006) [8] Ram T, O. Salazar, Guy P. Vertcal Handoff Decson Scheme Usng MADM for Wreless Networks. In IEEE WCNC 2008 [9] Q. Song and A. Jamalpour, A network selecton mechansm for next generaton networks, n Proc. IEEE ICC, Seoul, Korea, May 2005, pp. 1418 1422. [10] G. Polln, Trends n handover desgn, IEEE Commun. Mag., vol. 34, no. 3, pp. 82 90, Mar. 1996. [11] J. Broch, D. A. Maltz, D. B. Johnson, Y. Hu, and J. G. Jetcheva. A Performance Comparson of Mult-Hop Wreless Ad Hoc Network Routng Protocols. In MobCom'98, Proceedngs of the Fourth Annual ACM/IEEE Internatonal Conference on Moble Computng and Networkng, pp 85-97. [12] Pen-ho Chang and Tsan-Pn Wang (2009) Desgn and Implementaton of an Integrated RFID and VOIP System for Supportng Personal Moblty Internatonal Journal of Computer Networks & Communcatons (IJCNC) October 2009 vol 1, no. 3. [13] Yee YC, et al. (2007) "SIP-based proactve and adaptve moblty management framework for heterogeneous networks", Journal of Network and Computer Applcatons. [14] Dutta et al. 2005) Seamless Handover across Heterogeneous Networks An IEEE 802.21 Centrc Approach Proceedngs of the nternatonal wreless submt-wreless personal multmeda communcatons (WPMC) 2005. [15] Srcharan, M.S.; Vadeh, V.; Arun, P.P. (2006) An actvty based moblty predcton strategy for next generaton wreless networks Proceedngs of IFIP Internatonal Conference on Wreless and Optcal Communcatons Networks, 2006. 65

Authors Shanthy Menezes receved the M.S. and Ph. D degrees n Computer Scence from the Unversty of Texas at Dallas n 2000 and 2010 respectvely. Her research nterests nclude wreless networks, moblty management n next generaton heterogeneous networks, and network smulaton. S. Venkatesan receved the M.S. and Ph.D. degrees from the Unversty of Pttsburgh n 1985 and 1989, respectvely and has been a faculty member n the Computer Scence Department at the Unversty of Texas at Dallas snce then. Hs nterests are n telecommuncaton networks, varous ssues n the desgn and operaton of wreless networks, dstrbuted algorthms, fault-tolerance and network forenscs and securty. Hs work has been funded by the federal, state and ndustral sponsors n the past. 66