GNSS based Adaptive Monitoring for the Assistance of Persons with

Orientation Difficulties

Martin Litzenberger

, Klaus Dittrich

, Birgit Unger-Hrdlicka

, Hendrik P. Buimer

Ivan Rigamonti

, Wil Wintjens

, Matthieu Arendse

, Patrizia Murko

Matthias W.G. Zeller

and Stephanie Auer

Austrian Institute of Technology Ges. m. b. H., Giefinggasse 4, 1210 Vienna, Austria

CareCenter Software Ges.m.b.H., Franzosenhausweg 53, 4030 Linz, Austria

Vilans, Churchilllaan 11, 3527 GV Utrecht, The Netherlands

CREAGY AG, Baarerstrasse 2, 6301 Zug, Switzerland

Ivengi BV, Withuisveld 24, 6226 NV Maastricht, The Netherlands

Stitching TanteLouise, Postbus 73, 4600 AB Bergen op Zoom, The Netherlands

terzStiftung, Seestrasse 112, 8267 Berlingen, Switzerland

Kepler Universitätsklinikum Ges.m.b.H., Krankenhausstraße 7a, 4020 Linz, Austria

MAS Alzheimerhilfe, Lindaustraße 28, 4820 Bad Ischl, Austria

ivan.rigamonti@creagy.ch, wil.wintjens@ivengi.com, matthieu.arendse@tantelouise.nl

patrizia.murko@terzstiftung.ch, matthias.zeller@kepleruniklinikum.at, stefanie.auer@mas.or.at

Keywords: Geo-Fencing Technique, Safe Geo-Fence, Personal Safe Zones, Weighted Map, GNSS Position Monitoring,

Active Assisted Living Technology.

Abstract: Severe problems with out-of-home mobility are common under persons with cognitive impairment. Technical

solutions such as geo-fencing systems, based on GNSS position monitoring, exist to mitigate wandering and

getting lost of such patients. However, manually defined, fixed geo-zones effectively limit the freedom of the

person and can lead to false alarms even during normal walking behaviour. This paper presents work in

progress on a GNSS based position monitoring system that computes adaptive personalized safe zones from

the user walking routine. The system implements semi-rigid zone borders allowing the user to explore new

areas and automatically extend their personalized safe zones over time. Making use of routing services the

system can generate temporary corridors for appointments outside of the users typical walking routine. While

the technical implementation of the system is completed, it will be future work to evaluate the system's effect

on users in a field trial and a clinical study.

1 INTRODUCTION

Wandering and getting lost is one of the most

dramatic problems of persons with dementia and

Alzheimer’s disease (Yatawara, 2017). An example

would be someone who takes what would normally

be a routine walk and then suddenly is unable to find

his or her way. Sixty percent of people with

Alzheimer’s disease will wander at some point during

the progression of the disease, and half of those will

become lost and separated from a caregiver or the

responsible person (Butler, 1991). Sixty-one percent

https://orcid.org/0000-0002-2101-2188

of wanderers not located within the first 24 hours are

found deceased (Koester, 2008).

GPS tracking systems, often called "geo-fencing"

systems, are already used to mitigate the problem by

permanently monitoring the persons position and

alarming a carer if the person deviates from a

predefined geographic area (Milne, 2014). However,

such systems are not adaptable because they rely on

fixed predefined geographic zones, that need to be

manually defined by the care person. This can result

in frequent alarms, when a person is intentionally

exploring new regions or is visiting a friend, outside

the predefined "fence". Such situations are limiting

236

Litzenberger, M., Dittrich, K., Unger-Hrdlicka, B., Buimer, H., Rigamonti, I., Wintjens, W., Arendse, M., Murko, P., Zeller, M. and Auer, S.

GNSS based Adaptive Monitoring for the Assistance of Persons with Orientation Difﬁculties.

DOI: 10.5220/0010494602360243

In Proceedings of the 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2021), pages 236-243

ISBN: 978-989-758-506-7

the freedom of the person and may cause extra worry

to the caregiver (Landau, 2012).

GPS-tracking has been used to analyse the

mobility of persons with cognitive impairment

(Shoval, 2008). Geo-fencing systems with an

additional automatic calculation of the shortest

distance of the user to the closest carer in a group of

carers has been described (Pratama, 2020) and an

extension of the geo-fencing concept to overcome the

disadvantages of the rigid "fence" including tests with

a few persons have been presented (Lin, 2018).

In this paper we report on a new concept

developed in the framework of the EU AAL joint

programme, project “FreeWalker” (freewalker-

aal.eu), which tests and validates a monitoring and

warning system based on GNSS tracking that

addresses the specific deficiencies of older adults

with dementia and frailty. The FreeWalker system

monitors typical walking routes and thereby learns

the habits of the user to automatically generate an

adaptive safe-zone around the user's home.

Furthermore, the system can generate temporary safe-

corridors from calendar entries, to allow visiting

places outside the typical walking routes. Note that,

for simplicity, the cared-for person (or primary user

in the AAL context) will be called the user, and the

supervisor or caregiver (or secondary user in the AAL

context) will be called the carer, throughout this

paper.

FreeWalker will allow persons with cognitive

impairment - or who feel otherwise insecure in the

outside environment - an increased freedom and self-

determined way of moving outside. It is aiming at

giving carers piece of mind by supporting them in

tightly monitoring the user with the FreeWalker

"Care" App for mobile devices, when needed. But

FreeWalker also empowers the user by allowing to

make own decisions, by automatically reminding him

or her on unusual situations via the FreeWalker "Go"

App, even before an alarm is issued to the carer.

In contrast to previous work on personalized safe

zones that has been evaluated on GPS-tracks of three

individuals (Lin, 2018), this project plans a pilot

operation with a total of approximately 100 users

(users and carers) in the three countries Austria,

Switzerland and The Netherlands, accompanied by a

clinical study. This paper describes work in progress

as the technical concept is finalized but a field study

to evaluate if the system can improve the situation for

user and carer, is still to be done.

The paper concentrates on the technical

description of the logic engine software component of

FreeWalker that dynamically computes the safe zones

and corridors and is organised as follows: The

following section gives an overview over the system

components and the dynamic safe zone concept.

Section 3 explains the technical implementation, the

logic engines algorithms to compute safe zones and

safe corridors, as well as executing the rules to issue

information on the state of the primary user to the care

person. Section 4 presents an outlook on the future

study and is followed by a conclusion.

2 SYSTEM DESCRIPTION

2.1 Components

Figure 1 shows an overview over the system main UI

components that the users and carers are working

with. Four different components are integrated in the

system, that are directly interfacing to the users.

Two FreeWalker mobile applications: The user

app (named "FreeWalker GO") and carer app (named

"FreeWalker Care") have been implemented for

Android and iOS operating systems. The carer app

allows an overview of the users associated with one

carer, displaying their status as well as alarms and a

map to locate the users position in case of an alarm

(see Figure 4a and 4b).

The user app allows the user to actively send an

alarm via activating an alarm button but is also used

as a terminal to provide feedback to the user: Remind

them if he/she might have lost his or her way or might

have forgotten to go to a scheduled appointment. The

user app also constantly monitors the user position via

the smart phone GNSS receiver and sends it further

onwards to the system backend components. For

those users not capable of operating a smart phone

and mobile applications a GPS "dongle" can be used

alternatively. The GPS dongle can be worn as a

necklace or as a watch and provides an alarm button,

but automated system feedback to the user is not

possible with this option. Both, user app and GPS

dongle allow for a two-way voice communication

between carer and user in an emergency/alarm

situation.

Carers can view the user status and configure certain

system parameters in the web portal. The web portal

also allows to maintain the calendar of the user,

alternatively the calendar can be maintained by the

user via the user app. The portal also features a map

view page.

It allows to define a first, initial safe zone (see

next section) for a user by drawing with the mouse on

a map and to review the system generated safe zone

and temporary corridors to an appointment. The

history of alarms and how they have been handled by

GNSS based Adaptive Monitoring for the Assistance of Persons with Orientation Difﬁculties

237

the carers is also recorded and available through the

portal.

Figure 1: User roles and the system front-end components

available to them.

2.2 Dynamic Safe Zone Concept

The strength of the system and its major difference

compared to conventional "geo-fencing" approaches

comes from an automatic, periodic update of the

user's personal safe zone computed from the real-time

monitoring of the user's daily walking routines.

Figure 4b shows an example of a real life computed

safe zone in a screenshot of the carer app. The system

automatically produces two types of zones for each

user, depicted in the map as green and orange

polygons. The dynamic safe zone (or green zone as it

will be referred to in this paper) represents the

observed walking routines of the user over time. The

second exploration zone (or orange zone) surrounds

the green zone at a configurable distance (see Figure

4c). While the user can leave the green zone for an

adjustable duration without triggering an alarm (30

minutes have been defined in the co-design process),

leaving the orange zone (see Figure 4e) will initiate

an immediate alarm to the carer.

Note that one of the major differences of

FreeWalker from existing geo-fencing solutions is the

orange zone in which the user can explore without

triggering an alarm, but still his or her position is

closely monitored by the system (see Figure 4d).

This approach combines the safety of a strict geo-

fencing system with the flexibility to explore the

outside of a fixed zone and thereby extending the

green zone in consecutive walks. A moderate

extension is possible, while a sudden and large

change in walking habits indicates a potential

wandering or disorientation situation of the user.

During the exploration period the user gains

confidence in a certain area. This will empower the

user to independently walk outside and at the same

time protect him or her from getting lost. The green

and orange zone shapes are periodically re-computed

from the walking tracks of the previous day every 24

hours.

To mitigate the risk that the user is lost inside the

orange zone, e.g. walking in circles, the system

additionally monitors the time spent inside the orange

zone and raises an alarm to the carer after an

adjustable duration. To further empower the user

towards independency and self-responsibility, the

system is able to automatically ask the user via the

user app if he/she feels lost or otherwise unsafe before

the alarm is raised. Only if the user does not respond

after a further timeout, or actively asks for help, the

alarm is raised. The technical implementation details

of this logic are explained in the next section 3 of the

paper.

2.3 Safe Corridors

In addition to the periodically updated green zones

the system offers temporary safe corridors computed

from the appointments scheduled in the user's online

calendar. This allows the user to visit destinations

outside the orange zone within a predefined time

window without triggering an alarm. This corridor

extends the green zone to the destination address and

is valid only for a certain time interval on the day of

the appointment. At the same time the system

monitors the distance of the user to the destination

address and notifies the carer on late arrival or if the

user has arrived on time. Figure 4f shows a typical

corridor in a screenshot of the carer app.

It must be pointed out however, that the system

does not offer navigation functionality to the user, but

solely monitors the progress to, and on-time arrival at

the destination address.

2.4 User States

The status of the user is computed by the logic engine

component, one of the system backends (see Figure

2), from user position, zones and timing and is

communicated to the carer via the web portal and

carer app, using so called user state icons. Different

user states indicate clearly to the carer in what

situation the user currently is. Table 1 gives an

overview of the states related to the user position and

the zone concept. There are more states defined,

related to important system conditions such as

"battery low", but as this paper focuses on the safe

zone concept, these states are not discussed here.

Some of the states will be displayed to the carer

for information only, other states will be

accompanied by an alarm (e.g. "OutOfZone") that

must then be handled by the carer and is further

ICT4AWE 2021 - 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health

238

documented by the system. However, as this paper

focuses on the presentation of the dynamic zone

concept, the alarming and alarm handling logic of the

system are not detailed further in this paper.

Table 1: List of states related to the user position and

dynamic zone concept.

State Name Icon Description

Close2Home

The user is very near to

his/her home address

Outside

The user is in the green

zone, his/her usual

walking habit

ExploringZone

The user is in the

orange zone, exploring

the surroundings

OutOfZone

The user is outside of

the green and orange

zone, potentially

lost/wandering

IsReminded

The user is too long

exploring in the orange

zone, potentially

lost/wandering and is

reminded to return

AppointmentInProgress

A scheduled

appointment is due and

a safe corridor to the

appointment is active.

Close2Appointment

The user is close to the

due appointment

destination

LateForAppointment

The appointment is

due, but the user is far

from the appointment

destination

3 IMPLEMENTATION

In this section the system’s approach of green and

orange user-specific zones and states is explained. An

overview of the complete process done in the logic

engine backend from mapping the locations, finding

contours and zones to the evaluation of rules is

presented.

3.1 Location Mapping

Position data is received by the logic engine backend

(see Figure 2) and converted to corresponding pixel-

coordinates in a user-individual weighted map. Each

weighted map pixel is connected to an area on earth

in the WGS84 reference system while the centre is the

position of a user’s home location.

The following two points determined the choice

of map parameters:

Figure 2: System architecture with the user and carer apps

and web-portal front end and the logic engine backend and

the interfaces.

(1) The size of a weighted map is proportional to

the size of each pixel if the resolution stays constant.

Bigger pixels result in more edge-shaped zones which

is not consistent with the frontend-requirements.

(2) A higher resolution therefore results in higher

server workload which led to processing time

problems with a high number of system users.

For these two reasons, the size and resolution of a

weighted map were set to 10x10km

and 1000x1000

pixels.

To avoid trapezoid map shapes due to the

curvature of earth, geolocations of the weighted map

centre with values above 80° and below -80° latitude

are not allowed. Each element of the weighted map

can contain weights between 0.0 and 1.0. The lower

the value the longer the walked route lies in the past

(see Figure 3a). Due to a low transmission frequency

in relation to the walking speed the pixels must be

interpolated between the current and the last position

to obtain a continuous line of weighted pixels. The

walked routes have an adjustable width to achieve a

zone size customized to common movement patterns.

If on the one hand the width is chosen too small, it

leads to narrow zone areas when walking back and

forth on the same street which can easily activate an

alarm by receiving a GNSS signal just below the filter

threshold. On the other hand, a high value will make

areas accessible the user should not go to. The value

was set to two times the GNSS filter threshold of 30m

plus 10m additional space which is a width of 70m.

To adapt to behaviour changes a linear decay

algorithm is implemented to “forget” routes that have

not been walked for a specified duration. All weights

in a map are thereby reduced in proportion to time

GNSS based Adaptive Monitoring for the Assistance of Persons with Orientation Difﬁculties

239

elapsed since the last entered location with a cap of

0.0 at the lower end (no negative weights are

possible).

𝑤



=𝑤



−0.5∗









(1)

𝑤



is considered the weight before the decay is

applied and 𝑤



is the weight after the subtraction.

The denominator in the fraction part 𝑡



is the time

elapsed since the last entered location and the

nominator 𝑡



is an adjustable value which

specifies after what time period the starting weight of

1.0 is reduced by 0.5. For the planned study 𝑡



was set to 604.800s which corresponds to a duration

of seven days. Consequently, routes older than 14

days will not be considered anymore in the zone

creation process.

3.2 Zone Generation

In contrast to (Lin, 2018) the safe zone generation in

this paper is based on the weighted map data

described in section 3.1. To create a zone, state of the

art segmentation and contour algorithms provided by

the open source libraries OpenCV and SciPy are used.

Figure 3a shows a heatmap visualization of a

weighted map recorded over several days of test

walks around the home location. The darker the

colour of an area the longer the walked route lies in

the past. Figure 3b shows the corresponding selected

contour out of all contours found. The selected

contour will then be transformed to a polygon with an

average limit of 100 support points.

Figure 3: (a) Grey scale heatmap representation of the

weighted map with walks, (b) Zone contour found in

weighted map data.

Green Zone: Is the area limited by the locations

entered in the map and always contains the home

location. This zone indicates a save area where the

user can move freely.

Orange Zone: An orange zone is created by

applying a binary dilation algorithm on the data

shown in Figure 3a, which then is used as the basis

for a new contour search. The size can be changed per

user and is specified in meters distance from the

border of the previously processed green zone.

3.3 Corridor Generation

A corridor, or better described as temporary green

corridor, is a zone for special appointments outside

the 10x10km boundary depicted in section 3.1. The

system creates these time-dependant green zones with

a maximum map size of 30x30 km. This new type of

zone is only available for the appointment duration

plus estimated 90 minutes travel time before and after

the event.

To create a corridor the caregiver needs to enter

an appointment with a valid address, start time and

end time into one of the available frontends. The

appointment data is either pushed to the logic engine

directly or queried from the portal every 24 hours for

the appointments of the next day. The logic engine

then computes this data by requesting up to three

different routes from Google maps service with start

address at the user’s home location and end address

at the appointment’s location. These routes of

geolocations are point-wise transformed to pixel

coordinates and entered into an empty weighted map

with a corridor specific width. Because all routes have

the same starting and end point an enclosing polygon

is therefore created and detected by the same contour-

finding algorithms used in the zone creation

procedure in section 3.2. Finally, the contour is down

sampled and sent to the frontends.

3.4 Rule Engine

Interpretation of the current state of a user based on

his/her position is done in a two-step process:

Step 1:

User specific properties are updated, triggered by

each received geolocation. The properties are

classified into three different subjects:

Zone-Properties: These values define the current

state of a received location in regard to all colored

zones available for this user.

Example: location is in green/orange/red zone

(false/true), distance to home location (in meters)

Timestamps: The timestamps indicate when a

zone was left and when a zone was entered.

Appointment-Properties: The appointment

properties describe the current ongoing user-specific

appointment task.

ICT4AWE 2021 - 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health

240

Example: the distance to the appointment (in

meters), ongoing appointment (false/true).

Example: timestamp entered green zone

(seconds), timestamp left orange zone (seconds).

Step 2:

Properties are interpreted by rules which directly

define the status of a person. This is achieved by

setting status-specific conditions for all relevant

properties. All active rules are processed once per

minute for every active user in the system, regardless

if a location was received or not.

This provides up-to-date state information for all

users. Table 2 shows an example rule for the user state

“Close2Home”. In the left column the code line and

on the right the corresponding explanation is given.

All conditions need to evaluate as true to send the

defined user state to the system’s frontends.

Table 2: Code lines of an example rule.

CODE EXPLANATION

time_now -

last_loc_ts < 60

Last received location

timestamp must not be

older than 60 minutes

dist2home < 100

The distance to home

location needs to be less

than 100m

in_green == 1

Last location needs to be

inside a green zone

in_orange == 1

Last location needs to be

inside an orange zone

in_red == 0

Last location must not be

inside a red zone

state_to_set:

Close2Home

State to set by this rule

4 FUTURE WORK

The FreeWalker system's technical components have

been developed and tested by November 2020 and

will be piloted in a field trial, accompanied by a

clinical study.

4.1 Study Design

The study is designed as an open labelled,

randomized, controlled, parallel study with a

voluntary extension phase and will be conducted in

three countries Austria, Switzerland and The

Netherlands. The study will include both, users who

live at home and users who are living in nursing

homes with a focus on people with dementia. Study

participants will be randomized in a 2:1 ratio into the

intervention (FreeWalker, either Android

Smartphone App or a GPS dongle) or control (GPS

dongle with passive tracking only) arm using a pre-

defined, country specific, block randomization list.

Carers or contact persons of the users in the

intervention group will be using the FreeWalker carer

app based or the FreeWalker web portal.

Pre-screening of eligible candidates has started in

fall 2020. The core study duration is set to 16 weeks,

where the first study participants in Switzerland have

started to use the system in December 2020. The

voluntary extension is planned to terminate at latest

end of June 2021.

After the pre-screening phase, to select eligible

candidates, potential users will be evaluated for

fulfilment of inclusion and exclusion criteria. After

successful study inclusion, users will be randomized

to the intervention or control arm of the study and

users will be enrolled into the study. The enrolment

phase of the study was expected to span two months

but is currently behind schedule because of the

outbreak of the Covid-19 virus. Carers and users will

receive specific training in the operation of

FreeWalker or the GPS dongle from the care

organisations they are associated with.

After study enrolment the first study visit will be

conducted during which demographic data, health-

related data, etc. will be recorded and baseline

outcome measurements will be assessed. Eight weeks

into the core study phase a first follow up visit

(second visit) will be performed and outcome

measurements will be assessed for each user. After

completion of the 16-weeks core study phase, a third

study visit will be conducted, and outcome

measurements will be assessed. If the user volunteers

to participate in the study extension, he or she will

receive FreeWalker regardless of the initial study

assignment and additional training in the use of

FreeWalker if needed. At the end of the extension

phase, or if the user decides to terminate his or her

participation in the extension phase a fourth,

voluntary study visit will be conducted, and outcome

measurements will be assessed a last time. At the end

of the study, users are asked to return FreeWalker or

the GPS tracker. No further follow-up visits are

planned after the end of the study.

The primary outcome measurements of the study

are whether FreeWalker increases independence and

feeling of safety of the user and decreases worrying

of the carer about the whereabouts and safety of the

user. Outcome will be measured on a 5-point Likert

Scale created specifically for this field trial and

changes in the average sum will be assessed.

GNSS based Adaptive Monitoring for the Assistance of Persons with Orientation Difﬁculties

241

Figure 4: Example screenshots of the carer app in various situations of usage. (a) Dashboard overview of all users with current

states and alarms, (b) map view showing the current active user states and the dynamic ‚green‘ safe zone and ‚orange‘

exploration zone. Typical display for different user states on the map: (c) User is close to his/her home, (d) user is in the

‚exploring‘ state, (e) user is out-of-zone alarm, (f) a safe corridor to an appointment extending beyond the exploration zone.

Note that, for privacy reasons, the user position is only shown in case of an alarm and is visible here in (c)-(d) only in the test

version of the carer app.

4.2 Ethics

Permanent tracking of a person poses an inherent risk

of violating someone's privacy and thereby a fair

balance between safety of the person with cognitive

impairment and monitoring and disclosing his or her

position need to be found (Landau, 2012).

The FreeWalker system will acknowledge this by

providing the real-time position of the user to the

carer only in case of an alarm. In all other situations

the carer will be only informed on the general status

of the user (e.g. the user is 'Outside'), without

disclosing details unnecessary to understand the

situation. Furthermore, the user, if still capable of

operating a smart phone and using the user app, will

be notified on a potentially abnormal situation in a

first step by the system. Only after this step, the carer

will be informed in a next step. It is expected that this

ICT4AWE 2021 - 7th International Conference on Information and Communication Technologies for Ageing Well and e-Health

242

will result in an increased freedom and self-

determined way of moving outside.

Study participants assigned to the control group

will naturally face a disadvantage to those assigned to

the intervention group. To allow all study subjects to

use FreeWalker we will offer a voluntary extension

phase starting after the core study phase during which

all study subjects are eligible to be switched over into

the intervention group regardless of their initial group

assignment. The study design has been approved by

the ethical commissions of all three test countries.

5 CONCLUSIONS

We have presented work in progress on an adaptive,

dynamic geo-fencing solution to improve the safety

of persons with dementia and frailty in the outside

environment. The "FreeWalker" system is capable of

computing personalized safe zones by monitoring the

typical daily walking behaviour of the user with a

GNSS dongle or smart phone app, to mitigate

wandering and getting lost. Temporary safe corridors

to appointment destinations are automatically

computed from the online calendar of the user.

The system computes the personalized safe zones

using a weighted map of the users walking

trajectories from GNSS data and periodically updates

the safe zone every 24 hours. Safe corridors are

generated from queries to the Google map API

composed of three alternative walking routes from the

user’s home to an appointment destination. The

corridors are activated during the time of the

appointment only and the progress of the user to the

destination is monitored and abnormal situations are

detected.

In addition, the user will be notified on a

potentially abnormal situation in a first step by the

system via the smart-phone app. Only after this step,

if the user does not respond or actively ask for help,

the carer will be alarmed in the next step. It is

expected that this concept will lead to an increased

freedom and self-determined way of moving outside

compared to existing "fixed zone" geo-fencing

systems.

After the completion of the technical development

in fall 2020, future work is to perform a clinical study

on the beneficial effects of the system on its users.

The study is planned with approximately 100

participants in Austria, Switzerland and The

Netherlands, started in December 2020 and is planned

to be completed in June 2021, to analyse the

difference in outdoor movement behaviour of persons

with and without the system.

ACKNOWLEDGEMENTS

This project has received funding from the EU via the

AAL Joint Programme under contract no. AAL-2017-

057, the Austrian Research Promotion Agency

(FFG), the Austrian Federal Ministry for Climate

Action, Environment, Energy, Mobility, Innovation

and Technology (BMK), The Netherlands

Organisation for Health Research and Development

(ZonMw) and the Swiss Innovation Agency -

Innosuisse.

REFERENCES

Yatawara, C., Lee, D. R., Lim, L., Zhou, J., & Kandiah, N.,

2017. Getting Lost Behavior in Patients with Mild

Alzheimer’s Disease: A Cognitive and Anatomical

Model. Frontiers in Medicine, 4, 201.

https://doi.org/10.3389/fmed.2017.00201

Koester R.J., 2008. Lost Person Behavior: A Search and

Rescue, Charlottesville, Virginia: dbS Productions LLC

Butler, J. P. & Barnett, C. A., 1991. Window of wandering

Geriatr Nurs 12, 226–227.

Milne, H., van der Pol, M., McCloughan, L., Hanley, J.,

Mead, G., Starr, J., Sheikh, A., & McKinstry, B., 2014.

The use of global positional satellite location in

dementia: A feasibility study for a randomised

controlled trial. BMC Psychiatry, 14(1).

https://doi.org/10.1186/1471-244X-14-160

Shoval, N., Auslander, G. K., Freytag, T., Landau, R.,

Oswald, F., Seidl, U., Wahl, H.-W., Werner, S., &

Heinik, J., 2008. The use of advanced tracking

technologies for the analysis of mobility in Alzheimer’s

disease and related cognitive diseases. BMC Geriatrics,

8(1), 7. https://doi.org/10.1186/1471-2318-8-7

Landau, R., & Werner, S., 2012. Ethical aspects of using

GPS for tracking people with dementia:

Recommendations for practice. International

Psychogeriatrics, 24(3), 358–366. https://doi.org/

10.1017/ S1041610211001888

Lin, Q., Liu, X., & Wang, W., 2018. GPS Trajectories

Based Personalized Safe Geofence for Elders with

Dementia. 2018 IEEE SmartWorld, Ubiquitous

Intelligence & Computing, Advanced & Trusted

Computing, Scalable Computing & Communications,

Cloud & Big Data Computing, Internet of People and

Smart City Innovation (SmartWorld/SCALCOM/UIC/

ATC/CBDCom/IOP/SCI), 505–514. https://doi.org/

10.1109/SmartWorld.2018.00111

Pratama, E. R., Renaldi, F., Umbara, F. R., & Djamal, E. C.,

2020. Geofencing Technology in Monitoring of Geriatric

Patients Suffering from Dementia and Alzheimer. 2020

3rd International Conference on Computer and

Informatics Engineering (IC2IE), 106–111.

https://doi.org/10.1109/IC2IE50715.2020.9274637

GNSS based Adaptive Monitoring for the Assistance of Persons with Orientation Difﬁculties

243