Content uploaded by Mads Haahr
Author content
All content in this area was uploaded by Mads Haahr on May 20, 2015
Content may be subject to copyright.
A Narrative Architecture for Story-Driven
Location-Based Mobile Games
Katsiaryna Naliuka, Tara Carrigy, Natasa Paterson, and Mads Haahr
School of Computer Science and Statistics, Trinity College Dublin, Ireland
Abstract. Dramatic improvements in smartphones over the last few
years have positioned them as a major platform for interactive media
content. In addition to being much more portable than laptop computers,
smartphones also support a sophisticated combination of GPS, sensors
and communications interfaces that allow extracting context information
related to their environment, such as location, orientation and weather
data. This combination of mobility and context-sensitivity opens up in-
teresting possibilities in relation to interactive narrative, and for example
allows audience immersion into an interactive story to be improved by
placing the content in physical locations that are of direct relevance to the
story. In this paper, we present a general-purpose narrative architecture
that allows a considerable range of story-based game and guide content
to be expressed in location-aware manner. We also present a case study
of an actual location-aware augmented reality game, which demonstrates
the architecture in a commercial setting and shows that it is sufficiently
lightweight to run on the current generation of smartphones.
Keywords: Narrative Architecture, Mobile Gaming and Storytelling
1 Introduction
Smartphones are emerging as a personal computational platform of significant
popularity. The latest generation of handsets are equipped with a sophisticated
combination of camera, GPS, orientation sensors and excellent data transmission
capabilities. For the purposes of games, these features allow a much greater
degree of sophistication than past generations of mobile games. In particular,
they facilitate creation of a new genre: location-based mobile games (LBMGs).
One of the advantages of location-awareness is that it lends itself well to
presenting the historical information by placing the player in a relevant context.
This is why location-aware games have significant educational potential and can
be used as interactive mobile guides. However, a significant challenge remains
with regard to structuring the story content in a fashion that allows it to be
presented adaptively in response to changes in user location and other types of
context. To address this challenge, we have developed a flexible narrative archi-
tecture that allows non-linear stories with multiple branches to be delivered in a
location-aware manner. In addition, the architecture supports efficient resource
management, which is critical for mobile applications.
In this paper we present the architecture and discuss our preliminary expe-
riences with it in the context of Viking Ghost Hunt (VGH), a location-based,
story-driven, augmented reality game that runs on Android smartphones. The
game is a hybrid of a mystery game and Gothic ghost story that incorporates
historical information about the Viking presence in Dublin (800–1169 AD). The
locations, in which the game is played, are selected because of the association
with the Irish Viking history and are of direct historical significance to the story.
The rest of the paper is structured as follows. In Sec. 2 we briefly review
the related work. Sec. 3 describes Viking Ghost Hunt and the requirements it
sets for the narrative architecture. Sec. 4 presents the architecture to support
interactive context-dependent narrative, which is the main contribution of the
paper. In Sec. 5 we describe how the architecture is applied in VGH. Sec. 6
contains discussion of the approach. Concluding remarks in Sec. 7 end the paper.
2 Related work
The capabilities of GPS enabled devices have motivated the development of
numerous applications that use the information about the physical location of
the user. Location-based mobile games (LBMGs) and tourist guides are popular
applications for such location-aware technology. During the last few years a broad
range of such applications was developed, ranging from virtual tourist guides [17]
to location-aware multimedia storytelling [15]. The settings, in which location-
aware applications are employed, range from the user strolling freely in the urban
or countryside environment [16] to the user being a passenger on a journey, upon
which he or she exercises no control [5].
Generally speaking location-aware applications can be more or less interac-
tive, ranging from more traditional mobile guides with limited user interaction,
such as the Lancaster GUIDE project [6] to, at the other extreme, highly inter-
active LBMGs [3]. Many of the latter operate in an open-ended game world with
set up rules but with no underlying narrative, such as the Can you see me now?
game where physical players chase the online users through the city streets. More
sophisticated game world is created in the Pirates! [4] game, where players can
undertake predefined missions, however the core of the game is interaction with
other players and with NPCs.
The interaction in the game does not need to be limited to the explicit actions
of the player. The player’s behavior, such as walking in a particular direction or
staying for some time in the particular place, can also be used to affect narrative.
For instance, in the Geist project [12] the player walks through a historical
site and is presented with location related stories. However, in contrast with
conventional mobile guides where the user is always presented with the same
content at the same location, in Geist the story is automatically adapted to the
user experience. The player gets a different story depending on the order in which
the locations are visited and on the amount of time spent in each location. Yet
the application aims to bind these stories together in a consistent experience. It
is worth noting that the boundary between LBMGs and interactive guides seems
to be blurring, since factually correct information is incorporated in some games,
and mobile guides aim to present the historical information to the user in more
entertaining way. For example, the Who Killed Hanne Holmgaard? project [16]
engages the player in a detective game, while at the same time familiarizing
him or her with the history of Denmark under the Nazi occupation. From the
other hand, in the Voices of Oakland project [7] the guided tour is shaped as
engaging encounters with ghosts. In the classification of interactive mobile guides
by Kjeldskov and Paay [10] LBMGs are included in this definition.
In our research we were concerned with developing an architecture for han-
dling narrative suitable for a wide range of location-based and, more broadly,
context-aware games. The beat sequencer from Fa¸cade is the closest example of
similar architecture [13]. In Fa¸cade the story is told through beats, i.e., pieces of
narrative united by a common topic. Beats are sequenced in response to player
interaction. When one beat is over, the beat sequencer is responsible for selecting
the next. The choice is guided by metadata associated with the beats, and each
beat is characterized by its set of preconditions, priority and story effects. The
goal of the beat sequencer is to select an available beat with the story tension
effects matching an author-specified story tension arc.
3 Viking Ghost Hunt
The game that we constructed, Viking Ghost Hunt (VGH), is a location-based
adventure game, driven by a Gothic ghost story set in Viking Dublin (800–1169
A.D.). In this game the player assumes the role of a paranormal investigator
and moves around the city hunting for ghosts, collecting evidence and solving
the mysteries of haunted Viking Dublin. The game is narrative-driven in the
sense that, in order to progress the game, the player must unlock a sequence
of distributed, location-specific, narrative fragments, which together make up a
meaningful story arc. Hence, the VGH narrative, which is mapped onto actions
and space, is revealed through the players’ activities as they move through the
locations and, in order to preserve the aesthetics of the role-play, these activities
are presented in the context of paranormal investigation. In this form of sto-
rytelling, the player is not an external observer, but becomes an active player,
who advances in the game and progresses the story by interacting with narrative
elements, such as location, character and plotline.
As is typical of the quest structure, and common in the adventure game
genre, the player’s actions are crucial to accessing the narrative and the chal-
lenges faced by the player integrate the story elements into the game play [1,
8, 9]. The purpose of the quest, therefore, is two-fold; it is both symbolic and
structural. By linking the narrative to the player’s activities, the quest provides
the player with meaningful goals while simultaneously functioning as a structure
for achieving these goals through specific activities [19]. The quest provides a
bridge between the game fiction and game rules and for this reason it is a useful
technique for creating the illusion of agency, which Janet Murray describes as
“the satisfying power to take meaningful action and see the results of our de-
cisions and choices” [14, p.126]. Supported by storytelling it can be a powerful
medium to convey the historical information to the player in entertaining form.
The characteristic of the quest structure that contributes most significantly
to the player’s sense of agency is the balance between embedded and emergent
narrative [11, 18]. Embedded narrative is pre-generated prior to player’s interac-
tion and tends to be more linear in nature. It contains the over-arching story arc
and gives structure to the order in which the narrative is conveyed, thus preserv-
ing a coherent expression of the plotline. Emergent narrative is more open-ended
and less predictable but is more responsive to the player’s interaction and there-
fore facilitates a greater sense of agency. While a linear narrative structure fits
easily with the classic Hollywood 3-act structure, which is popular for building
dramatic tension and emotional impact, a non-linear, branching structure affords
the player increased agency and the potential of re-playing the game.
With these considerations in mind, we were motivated to create a quest
structure that could balance both embedded and emergent narrative. Multiple
branching options were introduced at key points of an essentially linear structure,
thus forming parallel story paths. In order to contain the potential unwieldiness
of these optional branches, a set of inevitable event points to which all player
must return, was introduced. This fold back structure [2], is especially useful in
narrative-driven location-based gaming where game designers may also want to
guide people through both essential narrative plot-points and specific locations.
The main success criteria for the architecture to support the described type
of narrative can therefore be summarized as follows:
–it should be capable of supporting location-specific narrative;
–it should have ability to handle non-linear branching stories;
–as smartphones remain a resource-constrained platform it should facilitate
efficient resource management.
4 Narrative architecture
The main concept of our infrastructure is the concept of the challenge. Chal-
lenge refers to an atomic part of the story, i.e, to the task that the player needs
to accomplish to advance the game. The challenge is responsible for manipu-
lating (creating, destroying or changing) the passive entities, through which the
narrative is delivered to the player. In VGH, these entities include map over-
lays, ghosts (which are defined by their video and/or audio manifestations) and
atmospheric background sound effects.
The lifecycle of the challenge is regulated by a set of controlling conditions,
which consists of the precondition, the abort condition and the completion con-
dition. Initially, each challenge is in the passive state. The passive challenge can
never be completed unless it becomes active. A challenge is activated when its
precondition is satisfied. An active challenge may become completed if a new
event satisfies its completion condition. If the abort condition is satisfied it be-
comes passive again (see the state diagram in Fig. 1(a)). A challenge might
change between active and passive states an arbitrary number of times, but it
can only be completed once. A completed challenge can still be activated and
deactivated and can manipulate its passive entities. This can be useful if the chal-
lenge controls a permanent part of the game world. In this case, completion of
the challenge triggers the progress in the game but passive entities controlled by
the challenge remain in the game and can enhance the further game experience.
(a) Challenge state diagram (b) Mission infrastructure
Fig. 1. Main components of the narrative architecture
Each change in the state of the challenge triggers the corresponding action.
When the challenge is activated, its start action is triggered, deactivation triggers
the stop action and completion triggers complete. These actions can be specified
for each challenge separately and allow the challenge to adopt to its new state,
for instance, by manipulating the managed passive entities. The change in the
internal state of the challenge is not perceivable to the player but the change in
passive entities can be perceived through the corresponding UI elements.
The challenges are notified about changes in the game state through events.
Events are small objects that encapsulate relevant information about changes.
Examples of events include receiving a new update about the relevant context
changes (e.g., change of the location of the player), seeing or hearing the ghost
manifestation or the completion of one of the challenges. This last type of events
makes it possible to build a hierarchy of challenges by using completion of one
of the challenges in the controlling conditions of another. For instance, in this
way it is possible to chain the challenges, activating the next when the previous
is completed. Another way to create a more complex structure is to define a
master challenge that manages several smaller challenges and is completed when
a certain number of these “child” challenges is completed. Some examples of
such composite challenges are described in Section 5.
To bind all challenges together and to ensure delivery of events to each of
them the challenges are encapsulated inside the mission component. This com-
ponent handles all challenges that relate to the particular story. It is responsible
for accumulating game events and delivering them to the challenges according
to their internal state. When a new event is received the mission checks pre-
conditions of passive challenges and activates those for which preconditions are
satisfied. For active challenges the mission checks abort and completion con-
ditions and changes the state of the challenges accordingly. Figure 1(b) shows
the overall picture of the mission infrastructure. The arrows show how events
are propagated through the mission to challenges, and how the challenges in
response modify passive entities managed by mission.
In practice, there is no need to load all possible challenges into the infras-
tructure from the very start of the game. Instead we allow challenges to in-
stantiate other challenges at the appropriate times. Therefore only a limited set
of the initial challenges is instantiated first. Challenges are also allowed to un-
load themselves from the mission. After unloading they will no longer be notified
about the new events, and the resources that they hold at this point can be deal-
located. This strategy is particularly advantageous in the resource-constrained
environment of the mobile device.
5 Case study
The mission that we used as a proof-of-concept for our narrative architecture is
set in the surroundings of the medieval church of St. Audoen. The church lies
within the old Viking city walls in Dublin and is surrounded by public parkland.
There is a number of local stories associated with the place. In particular, the
church has in its keeping the so-called ‘Lucky Stone’, a medieval tombstone,
which is believed to possess magical qualities. Several ghosts from different times
are believed to haunt the area including the Green Lady doomed to look eternally
for her lost baby and the local gang member murdered in vicious street fights
during the 18th century. Including these ghosts in the game allowed us to expand
the focus of the mission from the Dublin Viking history to other parts of the
city’s past without breaking consistency of the story.
The mission unfolds as follows. When players launch the game, the icon
on the map notifies them that the surroundings of St. Audoen’s church are
haunted. Once the player reaches the site they begin undertaking an investigation
by searching for and locating evidence of paranormal activity. In this part of
the game, spots of paranormal activity are randomly located around the site.
The evidence that the player gathers at this point include pictures of Viking
artifacts and ‘ghostly’ sounds. This gives the player a chance to get used to the
game interfaces while at the same time getting immersed in the atmosphere of
historical/paranormal investigation. Once the player gathers enough evidence of
paranormal activity the new ghost appears. This is Olaf, a trapped Viking ghost
who solicits the player to help him escape and thus provides the player with a
motivating goal for the rest of the mission. At this point, Olaf also delivers the
player’s first explicit challenge commanding him to find the Lucky Stone. After
finding the ghostly remnant of the Stone at the spot where it was originally
located the player returns to Olaf to receive further instructions.
At this point the player is confronted with a narrative fork that contains
a dramatic choice. By way of completing the final challenge, the player must
choose between two routes and two approaches: take the easiest and quickest
Southern route but at great risk to Olaf or take the longer and more dangerous
Northern route and face impending doom. On both routes the player encounters
other St. Audoen’s ghosts, and Olaf tells their stories to the player. Both routes
eventually lead the player to the same target: the garden where Olaf wants to
return. Once the player reaches the garden he or she receives Olaf’s thanks as
well as a medieval coin as a token of Olaf’s appreciation.
We implemented the described mission using the narrative architecture pre-
sented in Sec. 4. To capture the story we use 30 instances of 11 different chal-
lenges. Most of these challenges havea fairly simple structure. For example, there
is a challenge “Play narrative once near ghost”, which is activated when the
player sees the visible ghost. The challenge then starts playing an audio file as-
sociated with that ghost and is complete when the playback is finished. Seven
instances of this challenge are used in the game, each one parametrized with the
particular ghost and the audio file. Two more interesting challenges, which we
would like to draw attention to, are the transition between the atomic challenges
part and the encounter with Olaf, and the branching point in the mission.
The activation condition for the ghost of Olaf is that the player found 3 out
of 10 pieces of evidence of paranormal activity located in the region. Finding
each piece of evidence constitutes a simple atomic challenge: its precondition
is that the player is in the region and the completion condition is that the
player captures the corresponding piece of evidence. In order to keep track of
how many atomic challenges have been accomplished, a more complex challenge
is created (see Fig. 2(a)). This challenge maintains an internal counter, which
is increased each time when one of the atomic challenges is completed. The
completion condition for this challenge is that this internal counter reaches the
number of 3. When the challenge is complete it activates the new ‘find Olaf’
challenge. Also, the completion of this challenge is a signal for remaining atomic
challenges to abort and to release their resources.
(a) (b)
Fig. 2. Structure of some sample challenges
A more complicated challenge is used to manage the two branches of the
game. The principal scheme of this challenge is presented in Fig. 2(b). Sub-
challenges that are managed by this challenge are represented as rectangular
shapes while oval shapes mark the actions by which the superchallenge reacts
in response to completion of the subchallenges. Subchallenges in this case are
responsible for triggering pieces of narrative either at certain locations or when
the player sees the visible ghost.
At the branching point the progression of the game is determined by which of
the two challenges gets completed first. If the player follows the Northern route
they will first encounter the Lane of Hell while on the Southern route the Green
Lady will be met. Depending on which of the two the player reaches first, the
remaining challenges of the Northern or the Southern route are activated while
the challenge associated with the alternative branch is removed. Some challenges
placed on the Northern route are optional. For example, passing the Lane of Hell
the player can encounter some of the ghosts haunting the lane (the Leper and
the Hangman) and learn their stories. However, if the player does not find them
it is still possible to proceed to the subsequent parts of the game. Other pieces of
narrative are critical for the consistence of the story and therefore mandatory.
6 Discussion
In Sec. 3 we stated the key requirements for the narrative supporting architecture
for Viking Ghost Hunt. In this section we review our proposed architecture to
determine the extent to which it meets these requirements.
Delivering location-specific (and, generally, context-specific) narrative is en-
sured by the ability of challenges to react to changes in the context through
events. As challenges are capable of changing their state in response to the
events, the delivered narrative can be context-specific. Non-linearity of the story
is achieved by the fact that multiple challenges are allowed to be active simulta-
neously. In this case, the player’s actions define which of the challenges will be
accomplished and therefore which branches of the story will be unlocked next.
The resource management is simplified by the fact that the resources are man-
aged by each challenge autonomously. Together with “on-demand” loading of
challenges, this ensures that only the necessary resources are allocated, and that
the resources are deallocated as soon as they are no longer required.
It is not an easy task to evaluate the architecture in isolation from the game
it was used in. To illustrate its applicability for a broader range of LBMGs, we
examine how it can be used for supporting the five types of LBMGs identified in
the survey undertaken by Kjeldskov and Paay [10]. The first type of games they
mention is treasure hunt games with linear story. They can be represented as a
sequence of challenges creating and loading the next one when completed. Jig-
saw puzzles is the type of the games to which Viking Ghost Hunt most closely
relates. Its defining characteristic is that the player collects seemingly random
fragments of the narrative, which gradually combine into the bigger story. This
type of narrative can be modeled by creating “branches” of challenges that all
ultimately meet at one final point. As challenges from different branches are
active simultaneously the user will always have the possibility to give up a cer-
tain branch to follow the other one (possibly even without noticing the fact).
Fig. 3. Playing Viking Ghost Hunt: the ghost is seen on the screen
Domino games are similar to treasure hunt games except that the context in
which the current narrative fragment is delivered suggests which of the several
possible next fragments should be selected. This type of story can be supported
by our framework since active challenges are notified about all relevant changes
in the context. The context information will determine, which challenge is to be
created and loaded next. Scrabble games offer to the player a highly interactive
experience adopting the story to many events that are happening around the
player. The example the authors give uses information from the Bluetooth pro-
files of the surrounding phones to modify the story. To create this kind of game
with our framework it is necessary to introduce new types of game events to
capture relevant information from the surroundings. Once this is accomplished
it becomes possible to adapt the behavior of the challenges to the contextual in-
formation.In collecting butterflies type of games unrelated self-contained stories
are linked to the locations. To support this type of games multiple challenges
can be created, each using the player location in its controlling conditions.
7 Conclusion
In this paper we have presented an architecture to support location-based and,
more broadly, context-sensitive interactive narrative. The architecture has been
successfully applied in practice to develop a location-aware mobile game Viking
Ghost Hunt for Android handsets. We have also discussed the capability of the
architecture to support a wide range of LBMGs and interactive mobile guides.
The architecture is specifically designed to meet the requirements of the
resource-constrained environment of the smartphone and can be applied to sim-
plify the development of future mobile applications. More extensive evaluation
including field user trials is reserved for future work.
8 Acknowledgements
We are grateful to the National Digital Research Center (NDRC) for funding
the work described in this paper.
References
1. Aarseth, E.: Quest games as post-narrative discourse. In: Narrative across media:
The languages of storytelling. pp. 361–376. U. of Nebraska Press (2004)
2. Adams, E., Rollings, A.: Fundamentals of Game Design. Prentice Hall, Upper
Saddle River, NJ (2006)
3. Benford, S., Crabtree, A., Flintham, M., Drozd, A., Anastasi, R., Paxton, M.,
Tandavanitj, N., Adams, M., Row-Farr, J.: Can you see me now? ACM Trans.
Comput.-Hum. Interact. 13(1), 100–133 (2006)
4. Bj¨ork, S., Falk, J., Hansson, R., Ljungstr, P.: Pirates! Using the physical world as
a game board. In: Interact 2001:8th IFIP TC.13 Conf. on HCI. pp. 9–13 (2001)
5. Brunnberg, L., Juhlin, O., Gustafsson, A.: Games for passengers: accounting for
motion in location-based applications. In: FDG ’09: Proc. of the 4th Int. Conf. on
Foundations of Digital Games. pp. 26–33 (2009)
6. Cheverst, K., Davies, N., Mitchell, K., Friday, A., Efstratiou, C.: Developing a
context-aware electronic tourist guide: some issues and experiences. In: CHI ’00:
Proc. of the SIGCHI Conf. on Human Factors in Comp. Systems. pp. 17–24. ACM,
New York, NY, USA (2000)
7. Dow, S., Lee, J., Oezbek, C., MacIntyre, B., Bolter, J.D., Gandy, M.: Exploring
spatial narratives and mixed reality experiences in Oakland Cemetery. In: Advances
in Computer Entertainment Technology. pp. 51–60. ACM (2005)
8. Howard, J.: Quests : Design, theory, and history in games and narratives. A.K.
Peters, Wellesley, Massachusetts (2008)
9. Juul, J.: Half-Real: Video Games Between Real Rules and Fictional Worlds. MIT
Press, Cambridge, Massachusetts (2005)
10. Kjeldskov, J., Paay, J.: Augmenting the city with fiction: Fictional requirements
for mobile guides. In: Mobile Interaction with the Real World 2007/5th Workshop
on HCI in Mobile Guides Singapore. pp. 41–45 (2007)
11. LeBlanc, M.: Feedback systems and the dramatic structure of competition. In:
Game Developers Conference (1999)
12. Malaka, R., Schneider, K., Kretschmer, U.: Stage-based augmented edutainment.
In: Smart Graphics: 4th International Symposium. pp. 54–65 (2004)
13. Mateas, M., Stern, A.: Structuring content in the facade interactive drama ar-
chitecture. In: Proc. of the 1st Art. Intelligence and Interact. Dig. Entertainment
Conf. pp. 93–98 (2005)
14. Murray, J.H.: Hamlet on the holodeck : The future of narrative in cyberspace. Free
Press, New York (1997)
15. Nisi, V., Oakley, I., Haahr, M.: Location-aware multimedia stories: Bringing to-
gether real and virtual spaces. In: ARTECH Conf. on Dig. Arts. pp. 72–81 (2008)
16. Paay, J., Kjeldskov, J., Christensen, A., Ibsen, A., Jensen, D., Nielsen, G., Vutborg,
R.: Location-based storytelling in the urban environment. In: OZCHI ’08: Proc. of
the 20th Australasian Conf. on CHI. pp. 122–129. ACM (2008)
17. Roberts, D.L., Cantino, A.S., Jr., C.L.I.: Player autonomy versus designer intent:
A case study of interactive tour guides. In: AIIDE. pp. 95–97 (2007)
18. Salen, K., Zimmerman, E.: Rules of play: game design fundamentals. MIT Press,
Cambridge, Massachusetts (2004)
19. Tosca, S.: The quest problem in computer games. In: Proc. of The Technologies for
Interact. Dig. Storytelling and Entertainment Conf. (TIDSE). Darmstadt (2003)
