Scribd
Upload
132 views

Human Movement Prediction Based On Telecom 'S Handoff

This document discusses using data from mobile telecom networks to predict human movement patterns. It aims to determine the probability that a person at one location will move to another location after a certain time period. The methodology involves collecting statistics on calls and handoffs from the network over time intervals and using mathematical models to calculate transition probabilities between locations. While the approach has limitations in accuracy, it can effectively predict movement for large numbers of users without compromising privacy.

Uploaded by

Surabhi Sadavat
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
132 views

Human Movement Prediction Based On Telecom 'S Handoff

This document discusses using data from mobile telecom networks to predict human movement patterns. It aims to determine the probability that a person at one location will move to another location after a certain time period. The methodology involves collecting statistics on calls and handoffs from the network over time intervals and using mathematical models to calculate transition probabilities between locations. While the approach has limitations in accuracy, it can effectively predict movement for large numbers of users without compromising privacy.

Uploaded by

Surabhi Sadavat
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 23

HUMAN MOVEMENT PREDICTION Based on Telecoms handoff Devaraj (09427) Devi .S.

Surabhi Sadavat (09470) Jyotsna Chaturvedi (09480)

O UTLINE

Objective
Motivation History

Mathematical concept
Methodology Limitations

Advantages
References

F EW Q UESTIONS
Is there a danger of spread of contagious disease? Is there a need to monitor traffic during rush hours? Is there a need to track human movement under conditions of emergency or distress? E.g. Exodus of north eastern people from Bengaluru.

O BJECTIVE
To study the use of the standard outputs (handover rates, call arrival rates, call holding time and call traffic) measured in a mobile telecom network to find out the probability of a person moving from one location to another after certain time.

M OTIVATION

To find the solution of a common problem :


what is the probability that a person at location A will move to location B after T hours ??

Prediction of human movement behavior is important for studying issues such as prediction of vehicle traffic and spread of contagious diseases, which requires tracing the movement of people. Mobile telecom network can efficiently monitor the movement of mobile users, the telecoms mobility management is an ideal mechanism for studying human movement issues.

M OBILITY M ANAGEMENT
The mobility of the users necessitates location management and handoff management, together known as mobility management of the wireless networks
MOBILITY MANAGEMENT

LOCATION MANAGEMENT

HANDOFF MANAGEMENT

H ISTORY

WIRELESS SENSOR NETWORK TECHNOLOGY

-high resolution data -limited to close proximity interactions

MOBILE TELECOM SERVICE

-use of mobility management -accuracy depends on the coverage area of one location areas

T ECHNIQUES
1) CELL ID BASED
-based on coverage of service areas - Accuracy : 500 m coverage area of 1 SA.

2) OBSERVED TIME DIFFERENCE OF ARRIVAL(OTDOA)


-Use of trilateration -Accuracy:50-150 m

3)ASSISTED GLOBAL POSTIONING SYSTEM(A-GPS)


-Use of radio access network to speed up GPS systems. -Accuracy:5-15 m

S PREAD P REDICTION M ODEL

A novel approach to study the spread problem by using the four major statistical parameters obtained from MSCs.
PARAMETERS-

1) Expected Call Holding Time (E[tc])


2) Number of handovers in and out of the cells(ij()) 3) Number of new call arrivals of the cells(i) 4) Voice/Data Traffic (in Erlangs) of the cell(i())

S IMPLIFIED M OBILE T ELECOM N ETWORK

10

M ETHODOLOGY
When a call arrives at a user or when he/she performs a handover, the activity is recorded at the MSC/SGSN.

STEP 1

STEP 2

The mobile telecom network collects the statistics of the activities for every t interval typically ranging from 15 minutes to several hours.

STEP 3

The mobile operator can then investigate these statistics (output measures) for future network planning.

11

R ELATED -P ARAMETERS

t The mobile telecom network collects the statistics of the activities for t time ranges from 15 min. to several hours. -timeslot Ri()-the average residence time before the user moves out of the cell. pij ()- transition probability of a user in time slot Pij(T) probability that a person will move from

cell i to cell j after time period T.

12

M ATHEMATICAL C ONCEPT

Based on Littles Law, a model is proposed to predict how people spread from one location to another after a period of time. LITTLES LAW - The expected number of users N in a system is arrival rate times the expected response time R that a user stays in a system.
N= R

13

B ASIC F ORMULAE

The average number () of users at cell in timeslot () = () /E[]

CASE 1- E[]>> , E[]<< () = () /min(E[], ) CASE 2- E[] t

() = ( )/ {min E[], }

14

T YPE O F U SERS
In timeslot :

TYPE 1 user moves into the cell when he/she is in phone conversation (and a handover occurs).
TYPE 2 user is not in phone conversation when he/she moves into the cell, and then has phone calls at this cell in . TYPE 3 user moves in /out of cell i without any call activity.

TYPE 4 user arrives at cell i earlier than , and then has at least one phone call at this cell in .

15

T OTAL P ROBABILITY

The average residence time () of a user arriving in cell i in timeslot is approximated as : () = () / ( () + () )

() - The number of type 1 users moving into cell i in timeslot . ( )- If Multiple-Calls-Per-Cell effect does not exist, then ( ) is the number of type 2 and type 4 users of cell i in timeslot . Pij(T) can be expressed as:

Pj(T)= P (T) (T)

Where T=T + Rk(T)

16

E FFECTS

EFFECT 1 - Multiple-Calls-Per-Cell Effect

-If a user generates multiple calls per cell : overestimates ()

EFFECT 2Movement of type 3 user is not included in () , which affects the accuracy.

17

E RROR A ND R ESULTS

18

L IMITATIONS

The concept is based on probability, so accuracy of the model is one of the main concerns. The errors of the prediction are limited to 20% and are less than 10% in most cases.

Its a mathematics based prediction model, it has some limitations regarding the assumptions made in order to generalize the results so obtained.

19

A DVANTAGES
Our approach does not need to identify individual users and therefore does not cause any customer privacy problem. This method can be effectively used for large number of users , and hence can cover larger areas.

20

C ONCLUSION

Predicting human movement behavior is a prominent issue for identifying the vehicular traffic, migration of refugee across the boundaries or spread of contagious diseases. Using wireless cellular technology we can predict the human movement behavior and this could be our next step towards more advanced cellular applications.

21

R EFERENCES

Wenchao Ma, Yuguang Fang, and Phone Lin, Mobility Management Strategy Based on User Mobility Patterns in Wireless Networks, IEEE Transactions on Vehicular Technology,56(1):322-330, January 2007. Yi Bing lin , Nabil Alrajeh, Predicting human movement based on telecoms handoff in mobile networks,IEEE transaction on mobile computing,2012. Prof. W.G.Teng, Report based on Handoffs Management, Wireless and Mobile Networks, National Cheng Kung University. Article on Human Behavior is 93 percent predictable, Website :

physorg.com

22

You might also like