Tool Support for Building Graph Construction

Paul Heiniz, Karl-Heinz Krempels, Christian Samsel, Kai Schwarz and G¨okay Bodur

Informatik 5, RWTH Aachen University, D-52074 Aachen, Germany

Keywords:

Landmark, Indoor Navigation, Building Graph, Mobile, Positioning.

Abstract:

Landmark-based navigation is one promising approach among the group of indoor navigation solutions. Using

logical connections between areas in the building, users are able to follow a path via step-by-step instructions.

This approach provides universally applicable navigation to complex buildings. However landmark-based

indoor navigation requires a solid and reliable data structure which depicts regarded buildings. In this paper

we present a tool for construction of such data structure. Using ﬁndings from prior experiments and well-

established design principles and patterns, we were able to signiﬁcantly improve the usability of the tool and

lower the cognitive load of the user. Furthermore, the proposed improvements were integrated into landmark-

based indoor navigation software to enhance the user experience and display a more familiar presentation

during the navigation task.

1 INTRODUCTION

Mobile devices of all kinds penetrate our daily life

and change it signiﬁcantly. The build-in hardware be-

comes more complex and consequently, the range of

application steadily grows.

Navigation is one of the many tasks that nowadays

devices are capable of. The satellite-based Global

Positioning System (GPS) became the standard for

outdoor positioning in recent time. There exist not

only stand-alone navigation solutions but many of to-

day’s sold mobile phones include the required soft-

and hardware for outdoor navigation. However, there

are limitations in the ﬁeld of application for such de-

vices. Even though navigation tasks on street level

are very precise, the GPS approach reaches its limit

at the entrance of a building. Since nowadays archi-

tectonic styles are changing to more individual and

often more complex structures, it is indisputably im-

portant to support visitors in such complex buildings

as airports, railway stations, hospitals, and museums.

Completely new navigation approaches are required

to navigate a person inside unfamiliar buildings. For

this task exist several hardware-based solutions as

well as approaches working solely with logical con-

straints of the building and human perception.

Landmark-based indoor navigation (Heiniz et al.,

2012) is one of such solutions. Compared to navi-

gation approaches based on precise geographic co-

ordinates which are mapped on a ﬂoor plan of the

building, landmark-based navigation relies heavily on

the data structure that represents the covered build-

ing. This data structure, a so called building graph,

consists of nodes and directed edges. Subdivided

into logical areas within the building, such a build-

ing graph is an abstract representation of the human

mental map that he subconsciously builds for indoor

environments. It is crucial to design such a graph in

a suiting, consistent, and precise way to cover most

special cases within rambling building structures such

as stairs, elevators, complex intersections, and large

open spaces.

This paper is structured as follows: Section 2

brieﬂy describes the aim of our work. Section 3 in-

troduces several navigation models and presents tech-

niques for indoor positioning and navigation. In Sec-

tion 4, we describe landmark-based indoor naviga-

tion in detail. Additionally, Section 5 covers the data

structure which is the foundation of this navigation

approach and depicts the implementation and archi-

tecture of our system. In Section 6, we present the

evaluation of our system. Finally, in Section 7, we

summarize this paper and future steps for the system

development are revealed.

2 APPROACH

This paper describes the design of a tool for creation

of building graph data structure. The majority of

265

Heiniz P., Krempels K., Samsel C., Schwarz K. and Bodur G..

Tool Support for Building Graph Construction.

DOI: 10.5220/0004510402650272

In Proceedings of the 10th International Conference on Signal Processing and Multimedia Applications and 10th International Conference on Wireless

Information Networks and Systems (WINSYS-2013), pages 265-272

ISBN: 978-989-8565-74-7

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

this work is based on the navigation approach and

resulting software from the work by (Heiniz et al.,

2012). We extend and simplify the existing approach

for graph creation. Sequential views from the initial

implementation are extended by the wizard metaphor

(Tidwell, 2010). This technique guides users through

the creation of the graph by locking the user into a

sequence of several well-deﬁned views. Therefore,

graph creation is done step-by-step in a logical order.

This way we are able to lower user’s cognition load

and avoid misleading handling of the tool.

Furthermore, an additional view is introduced

which is based on the ﬂoor plan of the building. This

view adapts well to the touch interface of the applica-

tion and allows a more natural handling of the graph.

During the editing, users have an overview over the

created graph and may update and delete data ele-

ments more easily. In case of missing ﬂoor plan for a

building, the proposed procedure switches into a de-

fault setting, where the user places elements of the

building graph on a white canvas. This option allows

a touch-based visual handling in a symbolic space.

Additionally, the application for indoor navigation

introduced in (Heiniz et al., 2012) is extended with

the presented approach. Users often want a more con-

ventional overview of the route while navigating with

the application. Therefore, they are able to switch

from the initial step-by-step view to an overview of

the building to observe the passed steps and the route

ahead. This way we introduce an additional view that

supports users which experience difﬁculties with pure

sequential instructions. Using this enhancement, we

were able to combine the advantages of landmark-

based indoor navigation (e.g., best-effort positioning

and navigation approach based on human cognition)

and the more common information from a building

ﬂoor plan.

3 RELATED WORK

The following approaches base on the idea of support-

ing positioning through the data structure. Thus, the

system is more independent from the existing techni-

cal infrastructure in the building.

Landmark-based indoor navigation (Heiniz et al.,

2012) does not rely on additional sensors in the build-

ing or the navigation device to lead the user to his des-

tination point in a building. This approach includes

the user into the navigation task by letting him sub-

consciously collect visual impressions about his sur-

roundings. The user has to perceive his surroundings

and compare certain areas of the building to presented

depictions in the application. He then actively in-

forms the system about his position in the building by

selecting certain areas from a list of preselected posi-

tions. Using physical constrains of the building and

the human power of observation, it was possible to

design a reliable navigation system. Landmark-based

indoor navigation serves as the starting point to this

work.

The navigation approach presented by (Chowaw-

Liebman et al., 2010) provides an advanced data

model for buildings. Users follow generated textual

instructions and thus are guided through the graph

structure of the data model. The position of the user is

monitoredvia the devicewhispering approach (Krem-

pels et al., 2011) during the navigation.

(Baras et al., 2010) presented an approach that

leads users through a building without any hardware-

based positioning. In this work, a model of the tar-

get building provides the route based on area iden-

tiﬁers such as room names or special locations. Ob-

jects which base on these identiﬁers are logically con-

nected. Users are following the sequence of locations

and reach their destination. However, the presented

system provides very sparse information which lacks

details. Furthermore, all information is presented as

text, therefore, users need to be familiar with the

building to follow the route.

Another approach working with imprecise posi-

tioning was introduced by (Jensen et al., 2009). The

presented system encloses areas to logical objects

which are connected in a building graph. Human

movement is tracked by a technique based on RFID

signal recognition. Even though this positioning ap-

proach shares the drawbacks of all hardware-based

positioning solutions, the graph on its own provides

strong constrains for possible actions within the build-

ing and thus the introduced navigation approach is

still reliable. A proper building structure enables

ﬂawless navigation for this approach.

Apparently, it is possible to navigate a person

through a building without precise hardware-based

positioning. A proper underlying data structure is cru-

cial to achieve a reliable indoor navigation approach.

However, most of the introduced approaches still use

ﬂoor plans to depict user location even though this po-

sition lacks precise coordinates. We will use this ﬁnd-

ing to evaluate human mapping of points on a map to

areas in the building.

Usability aspects play an important role for de-

sign of the building graph construction tool. We re-

course to a certain set of design principles and de-

sign patterns to create an accessible and user-friendly

experience. To optimize the existing graph creation

views, we adapted design patterns from (Shneider-

man, 2003). Additionally, we implemented the wiz-

WINSYS2013-InternationalConferenceonWirelessInformationNetworksandSystems

266

ard pattern for construction views from (Tidwell,

2010) to guarantee certain sequence of steps and sim-

plify the usage of our tool.

4 LANDMARK-BASED INDOOR

NAVIGATION

This section describes the approach for landmark-

based indoor navigation in detail. First, we explain

the idea behind the navigation approach, followed by

the navigation data model of our system.

4.1 Navigation Approach

The concept of the developed navigation system fol-

lows the human cognitive navigation process. Sub-

consciously, the human brain constructs a unique cog-

nitive map from the starting point to the endpoint

of the route which is divided into single route sec-

tions of manageable sizes characterized by waypoints

and landmarks known or communicated to the user

(Downs and Stea, 1982). A landmark is a unique rec-

ognizable reference point in a section used for orien-

tation and positioning of the user, whereas a waypoint

is a special kind of a landmark, namely the starting or

endpoint of a route section. Hence, the route consists

of a sequence of waypoints which the user needs to

pass in a predeﬁned order. Each of these waypoints

is connected to one or several landmarks which de-

pict this exact position visually unique in the context

of routing. During the human navigation process, a

mental depiction of the route, the cognitive map, is

continuously compared to the surroundings.

This subconscious procedure is modeled in the

navigation system by (Heiniz et al., 2012). However,

this approach does not only model the human navi-

gation process but also human instructions in case of

asking other people for the direction to a destination

point. People tend to describe the route by provid-

ing two to three landmarks which are located on the

way and build the directions using these unique ar-

eas. Landmark-based navigation system is based on

the same principle. A route along waypoints is com-

puted after deﬁning a starting and an end point. The

navigation system displays successively the next way-

point the user has to pass until he reaches the end-

point. To facilitate the navigation for each waypoint,

textual instructions are attached which describe how

to reach the next waypoint from the current position.

Additionally displayed landmarks allow a continuous

veriﬁcation of the current position along the route.

The user is actively integrated into the navigation pro-

cess by conﬁrming his arrivalat the target waypoint to

be navigated to a successor located on the route.

4.2 Navigation Model

In this section we will focus on a data model called

building graph. In our system, on the one hand the

building graph is used as a structure the navigation

relies on. On the other hand the system provides a

module to construct such a structure for an arbitrary

building.

In general, this graph consists of nodes and di-

rected edges. The nodes represent speciﬁc logical ar-

eas in a building which comprises unique attributes,

e.g., the entrance hall. They are used as naviga-

tion points and can obviously be applied as starting

points or endpoints. We distinguish between way-

points which depict landmarks lying on the route and

points of interest which are relevant navigation end-

points and a subset of waypoints. Both are discussed

below.

The edges represent all possible routes between

the nodes in the speciﬁc building. They have to be

directed edges because some routes can be restricted

with respect to their direction of movement, e.g., an

escalator. Every edge contains a rough distance value

between the connected nodes such that the user can

estimate how long he has to walk to reach the next

waypoint and can easily compare sections. An esti-

mation of the distance is sufﬁcient since a person can

hardly estimate accurate distances in buildings (Cut-

ting and Vishton, 1995). During the construction of a

building-graph, we assign a geographic direction to

each edge to compute the angle between two con-

secutive sections. In dependency on the angular de-

gree, the user can be precisely navigated by adapt-

ing the textual instructions according to the computed

value, e.g., turn left, turn right, turn around, followthe

route. The direction patterns base on work presented

in (Chowaw-Liebman et al., 2010).

Precise implementation of the building graph in

our application and all required components are pre-

sented in detail in Section 5.1.

5 IMPLEMENTATION

The following section describes the implementation

of the building graph construction tool and integration

into the existing application.

5.1 Building Graph

Building graph is the central construct that acts as

the base for landmark-based indoor navigation. It

ToolSupportforBuildingGraphConstruction

267

consists of all essential elements to describe a route

through a building. Such a graph is identiﬁed by the

address of the respective building.

A building is subdivided in different areas, e.g.,

different ﬂoors or departments which distinguish in

a logical matter. An area can be a subarea of an-

other one, e.g., there can be several departments on

the same ﬂoor of a building. Partially intersecting ar-

eas are not considered in our implementation.

Another important entity is a landmark. It con-

tains pictures of the reference point from the user’s

point of view, an identifying name, GPS-coordinates,

and Wi-Fi ﬁngerprints. The two latter attributes are

necessary because a landmark just as a waypoint can

be suggested to identify the starting point in the navi-

gation process.

A waypoint is similar to a landmark and is de-

scribed by the same attributes. The difference is that

a waypoint contains a set of landmarks which are vis-

ible from this point and is not necessarily a discrete

object. In fact, waypoints represent the nodes of the

building-graph, i.e., the starting points and endpoints

of every section of a route which have to be reached

to continue with the next section, e.g., branches or

stairs to the next ﬂoor. A point of interest (POI) is a

special waypoint which can be the target of an entire

route. They additionally contain a textual description

of the target. A waypoint is associated to the area it

is located in and to other adjacent waypoints by seg-

ment objects. Segments correspond to the edges of

the building graph and have to fulﬁll the attributes dis-

cussed above. Especially, all landmarks positioned on

a section are included into such a segment.

Landmarks and POIs are collected in logical

categories: LandmarkCategory, POICategory, and

POISubcategory. Landmarks are categorized by in-

cluded attributes, e.g., stairs, corridors, pillars, lifts.

POICategory and POISubcategory are used to fa-

cilitate the manual choices of the endpoint.

To support the user by determining her starting

point, GPS-coordinates and Wi-Fi ﬁngerprints are as-

signed to landmarks and waypoints. With the appro-

priate infrastructure, the navigation system can sug-

gest possible starting points based on hardware posi-

tioning. A GPS object contains a latitude, a longi-

tude, and an accuracy value and refers to every land-

mark and waypoint with the same coordinates. The

accuracy value speciﬁes the accuracy during the mea-

surement to determine the coordinates of a point in a

building conditioned by shielding. The system sug-

gests only estimated points whose coordinates stay

within this value.

Wi-Fi ﬁngerprinting (Chan et al., 2009) uses the

signals of Wi-Fi access points for positioning. To de-

Segment

Waypoint

POI

GPS

POISubcategory

POICategory

Landmark

Address

Area

Landmark

Category

WiFi

Fingerprint

1 * * *

1 1

Figure 1: Building graph structure.

termine the user position, the measured ﬁngerprint of

the navigation system is compared to the ﬁngerprint

objects in the database. Hence, every Wi-Fi ﬁnger-

print object requiresthe creation date, the access point

identiﬁer, the signal strength, and the references to the

considered landmarks and waypoints.

The overall structure of the building gragh is de-

picted in Figure 1.

5.2 Design Principles for Workﬂow

Improvement

The initial version of graph construction tool used a

loosely coupled collection of views bound by a cen-

tral menu. Each of these views addressed a certain el-

ement of the building graph. Thus, these components

were logically linked, e.g., a segment always had a

starting and an endpoint. However, due to the nature

of this approach, there were no clear sequences of re-

quired steps and the user had to ﬁgure out the right

construction steps on his own (see Figure 3). Addi-

tionally, those views were not optimized for usability.

Views were not consistent and lacked simplicity and

clarity. The principles of reducing memory load and

offer informative feedback are also addressed in the

ﬂoor plan view described in the next section.

By redesigning the views according to the ”Eight

Golden Rules of Interface Design” by (Shneiderman,

2003) and adding a clear creation sequence of steps,

usability of the tool should be clearly improved. We

focused on principles of consistency within the views

and easy reversal of actions. To approach the princi-

ples of designing dialog to yield closure and to reduce

short-term memory load, we used the wizard design

WINSYS2013-InternationalConferenceonWirelessInformationNetworksandSystems

268

Start

Enter

Address

Stop

Create

Area

Create

Waypoint/POI

Create

Waypoint/POI

Create

Segment

Create

Landmark

Create

Landmark

Create

Landmark

Figure 2: Logical building graph construction steps (sequence for wizard design pattern).

pattern for graph creation steps (Tidwell, 2010). User

is locked in a certain sequence of views and therefore

is not able to do wrong actions, e.g, user has ﬁrst to

create two waypoints to be able to create one segment

(see Figure 2). He also does not need to remember

all prior actions and is requested only to ﬁll out infor-

mation that is required as a logical consequence to his

actions.

Start

Enter

Address

Create

Area

Create

Waypoint/POI

Create

Segment

Create

Landmark

Figure 3: Initial building graph construction.

5.3 Floor Plan View

To achieve a more usable approach for building graph

creation it was our aim to provide users with a famil-

iar presentation. While a menu-based view is simple

to implement, it requires a huge cognitive overhead

to imagine and remember the created building graph.

The depiction of that structure is abstract and user is

required to build a mental map of the building in his

head and synchronize it to the actual graph. Without

visual feedback and a lot of experience it is a very

difﬁcult task to accomplish.

We decided to include a view into the tool which

displays the ground plan of the target building. In

terms of navigation, users are more familiar with the

concept of a visual 2-dimensional presentation of the

terrain from above known from most outdoor naviga-

tion systems and indoor plans. We differ between two

kind of situations: either a ﬂoor plan does or does not

exist for the target building. In case of the missing

plan, we use a blank canvas, i.e., white background,

to let users build the graph as a visualization of their

mental maps (see Figure 4(a)). This way users may

express their view of the building and create a clear

structure. They may freely place waypoints in the

building and combine these with segments of a de-

ﬁned length. The length of the segments is estimated

by the creating user. These distances are only guide

values for the navigating user and thus do not need

to be precise. They serve to represent distance differ-

ences among the sections within a route. Finally, on

the one hand, by using this view, we are able to help

users while creating a building graph with a perma-

nent visual feedback. On the other hand, this view

helps during the navigation to visualize walking di-

rections and route length with reference to the passed

segments. However, in this view, we are missing the

constraining features of a building plan. Hallways and

clear separated rooms often help users orient them-

selves.

In case of an existing ﬂoor plan, users are able to

create an even more informative and hence support-

ing building plan. Here, users are able to add, delete,

and update graph elements like waypoints and seg-

ments directly on top of the ﬂoor plan image (see Fig-

ure 4(b)). Therefore, the building graph structure was

extended for each visible element (i.e., waypoints and

segments) to include reference points on the image to

map the components to the visualization. Using these

coordinates, we are able to create a navigation view

that displays the entire route through the building (see

Figure 5). Natural constrains of the building support

the navigation. For this reason the ﬂoor plan view

with a real plan of the building is always preferable to

the black canvas approach.

Both navigation views were extended to support

users. Calculated route within the building is marked

blue, the passed route is grayed out, users current

location is marked green and the target POI is de-

picted as a checked square. Using this clear color-

ing schema, users are always informed about their

progress.

6 EVALUATION

In this section we discuss the evaluation of the imple-

ToolSupportforBuildingGraphConstruction

269

>(,),+--9?$ )"#$ 2+6#$ '#(*#'&$ )"#$ ,0+6#$ &#-#7)#*$ ,($ )"#$ ;'#+$ 37'##($ ).$ +$ -+'6#$

A,6%'#$B8BD$;'#+$<,)".%)$2-+($

(a) Graph construction on blank canvas

(b) Graph construction on ﬂoor plan

Figure 4: Graph construction in ﬂoor plan views.

mented tool for building graph construction. Every

participant had to complete two test routes within a

building. On one test route he used the improved tool,

on the other one he built a proper data structure for

four areas in the building with the initial version of the

software. Afterwards, the different approaches were

compared based on time measurements and user feed-

back.

6.1 Test Procedure and Setup

Description

The depicted navigation system was implemented in

Objective-C on the Cocoa-Framework. During the

test procedure the software ran on an Apple iPad 2

Wi-Fi with iOS 5.0.1.

The main building of RWTH Aachen University

was chosen to conduct the user tests. This building

has a complex architecture. It is rambling and con-

sists of three upper ﬂoors, two basement ﬂoors, differ-

ent kinds of stairways, and elevators. Although many

university departments are located in the main build-

ing, most students are not acquainted with this build-

ing, thus, it can be assumed that the test results are

not distorted. We used very detailed ﬂoor plans of

the building as the background for the ﬂoor plan view

(see 5).

The test group consisted of 2 female and 10 male

students aged between 19 and 30. Self-evaluated

technical skills of this group were well-distributed

from poor to excellent. The users were given a ver-

bal brieﬁng before the ﬁrst test run. Two test routes of

the same length and complexity were selected for the

study. The participants were asked to create building

graphs for each of those two routes once with the ini-

tial tool and once with the altered version from this

work. The resulting graphs consisted of four con-

nected waypoints with appropriate landmarks and in-

cluded one change of ﬂoor. Users were instructed

where to mark a waypoint but had to estimate the dis-

tance of each connecting segment between two points

in the resulting graph.

The main part of the evaluation bases on an on-

line questionnaire. Users could evaluate both tool ver-

sions, write comments, and rate the usability of the

system. The scale for the ratings ranged between 1

for poor results and 6 for excellent outcomes.

6.2 Outcomes

We observed that using our improved approach we

could lower graph creation time signiﬁcantly. The

WINSYS2013-InternationalConferenceonWirelessInformationNetworksandSystems

270

!"# ()**+',-&./&0' 92%/&0'#%2 :' %&92"6"6#'02%/'%&0'260'0& .-&./&%&0)

Figure 5: Route navigation.

prior approach was very struggling for the user and

required a lot of additional instructions. Views and

the proper sequence of creation steps were not clear

and required a massive cognitive load from the user.

After each action he had to reorient himself and re-

think his next steps. Introducing the wizard approach

for the views, we were able to create a clear step-by-

step sequence of views. Users always knew the next

required action and were able to fulﬁll the task more

easily and more efﬁciently.

However, most users instantly switched to the

ﬂoor plan view (see Section 5.3). Visual feedback and

the overview of the created building graph must not

be ignored and were highly preferred by the users.

Views from step-by-step instructions required a cer-

tain abstraction level from users. Areas in building

had to be mapped to textual descriptions and the cog-

nitive map. This approach was often confusing and

required experience with the tool and good memory.

The ﬂoor plan view solves this problem by mapping

the abstract structure of the building graph to an im-

age of the building plan.

The questionnaire among test participants con-

ﬁrms that observation. Even though users declared

that the visual ﬂoor plan is not especially helpful to

ﬁnd requested locations (average rating 2,83), it still

seems to support the users subconsciously. Therefore,

our test group rated the usability of the improved tool

(average rating 4,42) signiﬁcantly better than from the

initial version (average rating 2,08). Supporting this

outcome, all 12 users were aware of the improvement

of the modiﬁed tool and most users were satisﬁes with

this version (average rating 4,33). Consequently, all

12 users stated that they prefer the altered tool than

the initial version of the software.

7 CONCLUSIONS & FUTURE

WORK

Building graph is the central structure of landmark-

based navigation and thus requires to be ﬂexible, re-

liable and precise to guarantee proper indoor naviga-

tion. In this work, we introduced an improved tool

to construct such a graph in a user friendly and con-

venient way. Existing views were improved and rear-

ranged by using well-established design patterns and

principles. An additional view, the ﬂoor plan view,

was introduced. Users are now able to map the ab-

stract structure of the building graph to a ground plan

of the building. Thus, they have a permanent visual

feedback over the progress and all created graph com-

ponents.

The evaluation revealed that the modiﬁed version

of the graph construction tool is clearly an improve-

ment over the prior version. User study participants

observed a noticeable enhancement of the approach

and clearly stated to prefer the new version of the tool.

However, further work is required to improve and

evaluate graph construction as well as the entire en-

hanced navigation approach. First, ﬂoor plan view

with a blank canvas needs a proper evaluation. While

creating a building graph on a depiction of the build-

ing works properly, construction of this data structure

on a white backgroundrequires a more abstract proce-

dure. Users must project their mental map of the sur-

roundings into a visual representation. An evaluation

needs to reveal user interaction with this approach.

Introduced ﬂoor plan view needs an additional

evaluation. We will test usability and proper pre-

sentation for the navigation application. One intro-

duced idea was to include the building graph visual-

ization as the background for the step-by-step naviga-

tion instructions. This approach would combine the

overview of the graph with the detailed descriptions

of the route. User would not need to switch between

views and therefore need less actions for navigation.

However, we need to evaluate whether the added in-

formation is required and does not excess the accept-

able amount of depicted information.

ToolSupportforBuildingGraphConstruction

271

REFERENCES

Baras, K., Moreira, A., and Meneses, F. (2010). Navigation

Based on Symbolic Space Models. In International

Conference on Indoor Positioning and Indoor Navi-

gation (IPIN 2010).

Chan, E., Baciu, G., and Mak, S. (2009). Using wi-

ﬁ signal strength to localize in wireless sensor net-

works. In Communications and Mobile Computing,

2009. CMC ’09. WRI International Conference on,

volume 1, pages 538 –542.

Chowaw-Liebman, O., Christoph, U., Krempels, K.-H., and

Terwelp, C. (2010). Evaluation of an indoor naviga-

tion approach based on approximate positions. In In-

ternational Conference on Wireless Information Net-

works and Systems (WINSYS 2010), pages 195–201.

Cutting, J. E. and Vishton, P. M. (1995). Perceiving lay-

out and knowing distances: the integration, relative

potency and contextual use of different information

about depth. In Handbook of perception and Cog-

nition., volume 5: Perception of Space and Motion,

pages 69–117.

Downs, R. M. and Stea, D. (1982). Kognitive Karten: Die

Welt in unseren K¨opfen. UTB f¨ur Wissenschaft.

Heiniz, P., Krempels, K.-H., Terwelp, C., and W¨uller, S.

(2012). Landmark-based navigation in complex build-

ings. In International Conference on Indoor Position-

ing and Indoor Navigation (IPIN 2012).

Jensen, C. S., Lu, H., and Yang, B. (2009). Graph Model

Based Indoor Tracking. Tenth International Confer-

ence on Mobile Data Management: Systems, Services

and Middleware (MDM 2009), pages 122–131.

Krempels, K.-H., Patzak, S., St¨ulpnagel, J., and Terwelp,

C. (2011). Device whispering: An approach for

directory-less wlan positioning. In e-Business and

Telecommunications, Communications in Computer

and Information Science, pages 346–357. Springer

Berlin Heidelberg.

Shneiderman, B. (2003). Designing the User Interface.

Pearson Education.

Tidwell, J. (2010). Designing Interfaces. O’Reilly Media,

2nd edition edition.

WINSYS2013-InternationalConferenceonWirelessInformationNetworksandSystems

272