Location Management in Cellular Networks: Classification of the Most

Important Paradigms, Realistic Simulation Framework, and Relative Performance

Analysis

Author: K. Kyamakya, Klaus JobmannIEEE Transactions on Vehicular Technology,

Vol. 54, No. 2, Mar. 2005 Speaker: Jun Shen

Overview

Background Motivation Contribution Methodology Strength and Drawback of the paper Link with the class Link with project Q&A

Background

Mobile network is more and more popular • Increasing number of mobile subscribers

• the emergence of different mobile communication technology e.g. IEEE 802.11 WLAN, 3G/4G wireless cellular network, bluetooth,

• Everything on the move, e.g. laptop, PDA, mobile

Motivation (1)

Cellular network is one of most important network in daily life, almost every mobile network is based on cellular network, GSM, CDMA, UMTS, X-CDMA

How to lower the cost of system management and control scheme?

Motivation (2)

Location management (LM) is one important part of management and control scheme, one good start point

There are lots of location management schemes presented, what is the most efficient one?

Contribution

Classification of published location management methods

Presents results of a related extensive performance comparison of various location management in cellular network---LM with profile is most efficient scheme

Methodology--Some Assumptions(1)

LM scheme cost components• Paging

• Polling cycle

• Number of cells polled

• Location update

• Mobility pattern

• Call pattern

• Overlook the impact of handover because it focus on radio mobility

• This paper focus on signaling cost only

Methodology--Some Assumptions(2)

Network architecture

Methodology overview

Define&Study a universal structure of a performance analysis framework for LM methods

Introduce&Impl. a realistic user mobility model and simulation environment

A systematic comparative performance analysis of a representative sample of most important LM schemes.

Methodology—Current LM Overview

LM scheme components• Paging (cost are polling cycle and number of

polling cells sensitive)• Polling cycle---one ~ three polling cycles (with

delay constraint)

• Polling area---static/dynamic based on profile

• Location update • Static LA– cost depends on topology

• Dynamic LA --- cost depends on user mobility and call pattern

Methodology—Overview of PA

Methodology—Overview of LU

Methodology—Mobility Model (1)

Methodology—Mobility Model (2)

The paper adopts:• Activity-based approach

• stress user mobility

• More realistic

• Consider the impact of aggregate traffic on individual behavior

• Generate reference mobility profile used to develop a Markov model with history

Methodology—Mobility Model (3)

The model includes:• Space dimension

• accuracy of street segment

• data can be obtained from roadmap (e.g. GPS roadmap)

• Commercial simulation tool available--VISUM

• Simulation of aggregate traffic state profile

• Location, timing and sequencing of individual user movement

Methodology—Mobility Model (4)

The activity-based model includes:• Number of activities of interest for a user

• Time zone for each activity

• Activity duration profile

• Activity sequence profile

• Geographic location of activities

Methodology—Mobility Model (5)

The activity-based model

Methodology—Mobility Model (6)

User classification:

Methodology-Sample LM method (1)

Profile classification:

Methodology-Sample LM method (2)

LM classification (to be continued):

Methodology-Sample LM method (3)

LM classification:

Methodology-Sample LM method (3)

Brief Introduction (to be continued):• GSM classic : same as textbook mentioned

• GSM+profile : allow sequential paging rather than blanket paging

• Scourias: use profile to dynamically setting up the LA for a user

• SCOUKYA: • Enhancement of Scourias

• adopt GSM+profile fallback method

• Reduce dependence on movement history

• LA has a predefined max size

Methodology-Sample LM method (4)

Brief Introduction (to be continued):• Movement-based: refer to textbook

• Distance-based: refer to textbook

• Direction-based: LU whenever movement direction changed

• Direction-based sector method: use a sector of direction instead of a single direction

Methodology-Sample LM method (5)

Brief Introduction :• SCOUKYA2: replace type2 profile with type3 one• BIEST:

• Use type4 profile• Iteratively increase the size of LA(according to profile)

until cost of paging > cost of update

• BIEST_KYA:• Use type3 profile• LA of fixed size

• KYAMA: • LA-based + timer-based• Macro LA—actual LA + next LA

Methodology-Simulation context (1)

Overall scheme (to be continued): DB part

Methodology-Simulation context (2)

Overall scheme: functional structure

Methodology-Simulation context (3)

Geographical and aggregation data: • From the administration of town Hannover

• From the traffic planning of the university of Hannover

Radio cell structure: square size, cell diameter range [100m, 7km]

Methodology-Simulation context (4)

Timing of user movement (contd)• Activity location:

• C1-C7: data from the Hannover admin.

• C8,9: random distribution over the city

• Activity sequencing:• C1-C7: data from survey

• C8,9: random transition and duration matrix

Methodology-Simulation context (5)

Possible values for the duration, two groups:

Methodology-Simulation context (6)

Call arrival profile• Fix call numbers per day

• Distribute numbers over a day

Methodology-performance analysis (1)

Mobility characterization—simplify the designed model

Develop two metrics and a benchmark – for the purpose of comparison

Methodology-performance analysis (2)

Mobility Simplificaiton (one example) ----contd• Cell dwell time: independent of any activity

duration and transition matrix if consider logarithmic axes

Methodology-performance analysis (3)

Elements of interest • Average activity duration

• Activity location randomly distributed over the geographical surface

• Activity transition matrix can be taken random

• Radius of geographical area is R

• Average call intensity

• Average CHT

• Average cell size

Methodology-performance analysis (3)

Performance analysis• Call to mobility ratio (CTM)

CTM=Avg number of calls per day/ Avg activity duration *100

A indicator of user activity determinism

• Cost = nPA + c* nLUnPA: average number of paging

nLU: average number of locaion update

C: nLU/nPA, [5,10]

Methodology-performance analysis (4)

Performance analysis (c=5)

Methodology-performance analysis (5)

Performance analysis (c=10)

Methodology-performance analysis (6)

Performance analysis

Link between paper and class?

This paper gives a thorough review of current LM scheme

It gives an extension of standard LM scheme.

How the paper is related to my project?

The paper show a way to evaluate the efficiency of LM scheme

My project is to compare the efficiency of two LM scheme