Reflections on the long-term use of an experimental digital signage system.
ABSTRACT In this paper we reflect on our long-term experiences of developing, deploying and supporting an experimental digital signage system. Existing public display systems almost always feature a single point of control that is responsible for scheduling content for presentation on the network and provide sophisticated mechanisms for controlling play-out timing and relative ordering. Our experiences suggest that such complex feature-sets are unnecessary in many cases and may be counter productive in signage systems. We describe an alternative, simpler paradigm for encouraging widespread use of signage systems based on shared 'content channels' between content providers and display owners. Our system has been in continuous use for approximately 3 years. We reflect and draw lessons from how our user community has adopted and used the resulting public display network. We believe that these reflections will be of benefit to future developers of ubiquitous display networks.
Conference Paper: Yarely: a software player for open pervasive display networks[Show abstract] [Hide abstract]
ABSTRACT: This paper describes Yarely, a software player designed to support the next generation of pervasive display networks. We identify five design goals for future digital signage players: the ability to provide basic signage functionality (media scheduling and playback) and support for openness, extensibility, resilience and appropriation. The paper describes the design and implementation of Yarely in light of these design goals. We present an evaluation of the system in the form of deployment and usage data alongside a reflection on the degree to which Yarely successfully addresses the needs of future pervasive display systems.Proceedings of the 2nd ACM International Symposium on Pervasive Displays; 06/2013
- [Show abstract] [Hide abstract]
ABSTRACT: The proliferation of digital signage systems has prompted a wealth of research that attempts to use public displays for more than just advertisement or transport schedules, such as their use for supporting communities. However, deploying and maintaining display systems “in the wild” that can support communities is challenging. Based on the authors’ experiences in designing and fielding a diverse range of community-supporting public display deployments, we identify a large set of challenges and issues that researchers working in this area are likely to encounter. Grouping them into five distinct layers -- (1) hardware, (2) system architecture, (3) content, (4) system interaction, and (5) community interaction design -- we draw up the P-LAYERS framework to enable a more systematic appreciation of the diverse range of issues associated with the development, the deployment, and the maintenance of such systems. Using three of our own deployments as illustrative examples, we will describe both our experiences within each individual layer, as well as point out interactions between the layers. We believe our framework provides a valuable aid for researchers looking to work in this space, alerting them to the issues they are likely to encounter during their deployments, and help them plan accordingly.ACM Transactions on Computer-Human Interaction 07/2013; 20(3). · 1.18 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: In the age of online social networks, local communities still play an essential role in supporting social cohesion. In this paper we present a study that explores the design of "interacting places" -- networked public multimedia services that foster community awareness between local members -- in the context of a student community. In order to have interacting places "fit in" with the existing communication practices of the students, we performed and analyzed a set of semi-structured interviews with n=17 students regarding their use of email, social networking services, and instant messaging to stay in touch with others. A follow-up online survey (n=76) then explored how networked public multimedia services could complement these practices. Following a "communicative ecology" approach -- a conceptual model that represents the technical, social, and discursive contexts of communication -- we draw up guidelines to support the design of both content and channels (applications) for interacting places in student communities.Proceedings of the 11th International Conference on Mobile and Ubiquitous Multimedia; 12/2012
Reflections on the Long-term Use of an Experimental
Digital Signage System
Sarah Clinch1, Nigel Davies1, Adrian Friday1, and Christos Efstratiou2
1Lancaster University, Lancaster, UK
2University of Cambridge, Cambridge, UK
In this paper we reflect on our long-term experiences of de-
veloping, deploying and supporting an experimental digital
signage system. Existing public display systems almost al-
ways feature a single point of control that is responsible for
scheduling content for presentation on the network and pro-
vide sophisticated mechanisms for controlling play-out tim-
ing and relative ordering. Our experiences suggest that such
be counter productive in signage systems. We describe an al-
ternative, simpler paradigm for encouraging widespread use
of signage systems based on shared ‘content channels’ be-
tween content providers and display owners. Our system has
been in continuous use for approximately 3 years. We reflect
and draw lessons from how our user community has adopted
and used the resulting public display network. We believe
that these reflections will be of benefit to future developers
of ubiquitous display networks.
Public display, digital signage, long-term deployments, HCI
ACM Classification Keywords
H1.2 User / Machine Systems: Human information process-
Design, Experimentation, Human Factors
Digital public displays have long been an area of interest
in both commercial and academic settings. As display hard-
ware costs have fallen, digital displays have become increas-
ingly prevalent, appearing in many public and semi-public
spaces. In parallel there has been an increase in research in
the area, with a particular focus on user interaction with pub-
lic displays. Our own work is driven by the desire to eventu-
ally create open display networks in which viewers can ap-
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
UbiComp’11, September 17–21, 2011, Beijing, China.
Copyright 2011 ACM 978-1-4503-0630-0/11/09...$10.00.
propriate displays for their own purposes—shifting the focus
of display networks away from pushing advertising at unre-
ceptive audiences and towards a key role as part of our future
global computing and communications infrastructure.
In 2004 we began an intensive research programme in this
area—a key part of which was the creation of a large scale
testbed for public display research. In order to justify the
cost and disruption associated with its deployment and to
help reduce the novelty factor associated with the displays
we considered it important that the testbed would also serve
as a general digital signage solution for our institution—
while as researchers we have many exciting ideas for how to
use public display networks it is inescapable that for many
We initially experimented with a range of commercial sig-
nage offerings but soon found that the feature set offered by
these systems did not fit well with our user community nor
did they encourage an ‘open system’ with content coming
from the diverse small groups we aimed to support. As a re-
sult we designed, developed and deployed our own signage
system on top of an experimental systems infrastructure. At
the time of writing our system has been in operation for ap-
proximately three years and in this paper we reflect on our
own and our users’ long-term experiences. In particular, we
make the following contributions:
• We present a novel design for a simplified signage con-
trol system based on a separation of content producers and
• We provide detailed insight into the system’s continued
daily use for approximately 3 years.
• We provide lessons for designers of future ubiquitous dis-
We believe these contributions will be of significant value to
future developers of ubiquitous display systems.
In order to support conventional “signage” functionality in
our organisation we initially explored the use of two dis-
tinct methods of managing and scheduling content within
our campus-based display network.
Figure 1. Sketch of a possible UIs for sharing content across University campus. From left to right, exploring (a) daily content vs. overriding ‘events’,
(b) associating dates and times with content, (c) a calendar timeline view, and (d) presenting private and shared (public) channels.
A commercial signage system
Deploying and controlling large scale networks of public
ing and managing content is a key part of any modern public
display system (e.g. ). We deployed a common commer-
cial signage system that provides software for content cre-
ation and scheduling, content transfer, content play-out and
monitoring of software and hardware on play-out devices.
The system provides the ability to ingest content into the
display network. Content that is to be scheduled has to be ar-
ranged into multi-track timelines that are called “playlists”.
Finally the user can specify which “channels” to schedule
thecontentin. Channels, inthiscontext, representpre-defined
sets of play-out devices. We provided access to several key
user groups such as the marketing department, colleges and
A drop-box system
For more casual content providers we built a “drop-box” so-
lution. This system was a simple web-based workflow that
allowed content providers to request the publication of a
piece of content they uploaded. The system notified the Dis-
play Management Team when and on which displays that
piece of content should be displayed and allowed the team
to accept and reject content easily. In most cases, the team
then used a commercial signage system to add the content to
the displays’ playlists.
These two systems were deployed and used to support rou-
tine scheduling of content within our deployment.
The commercial signage system and the drop-box system
provided two very different ways of managing content on
a network of public displays. The commercial system of-
fered a highly configurable ‘TV schedule’ like environment
that allowed content creators to decide exactly when their
content should be displayed, in what sequence and on which
displays. The system assumes a model of the world in which
a small number of well trained operators are responsible for
scheduling content and the learning curve for the system is
The “drop-box” approach was much simpler for the content
providers, but relied on a centralised administrator to decide
the scheduling patterns of individual pieces of content. This
often leading to a perception of unresponsiveness.
After a period of 16 months during which people had the
opportunity to experiment with both mechanisms we per-
formed a set of interviews with selected major users of the
system in the public arts, colleges and a marketing depart-
ment. The aim of these interviews was to discuss their expe-
riences with the two systems and identify the requirements
for a new system of managing content distribution on the
public display network. The resulting observations were:
• Too much flexibility hampers usability. Only very few
ule content using the commercial system (mainly in the
public arts who required more flexibility for scheduling
their content). Most interviewees believed the system not
well suited to non-technical users. In fact, most people
used the “drop-box”, despite having access to the full sys-
• Users want to control their own displays. People affiliated
with colleges and departments expressed a need to assume
ownership of displays in their environment. Although it
was clear that the displays belong to a University wide
infrastructure, the fact that a display is located in a partic-
ular building raised the issue of control from the people
that are affiliated with that location.
campus. An obvious requirement for any public display
system—people from different groups across the Univer-
sity expressed the need to publish announcements or pub-
licity content on multiple displays. Combining this re-
quirement with the desire for display ownership requires
us to balance control and sharing of content on displays.
• Users want real-time responses to changes in their con-
would have to create or modify content and then wait
(sometimes days) before a member of the Display Man-
agement Team could respond to their request. In practice
we found that they often wanted to make changes to con-
tent at the last minute and wanted real-time responses to
During the interviews we had the opportunity to discuss al-
ternative interfaces for a replacement system (see prototype
designs in Figure 1). Designs that were discussed ranged
from those based on calendars and timelines, to interfaces
where playback of each content item could be fine-tuned.
The interviews gave us clear indications drawn from our
users’ past experience: people liked the simplicity of the
“drop-box” approach, but they required coarse-grained con-
trol over content scheduling. It was clear that too much de-
control over what was displayed was equally undesirable.
THE E-CHANNEL SYSTEM
Based on our early experiences our aim was to develop a
system that provided a simple way of enabling users to dis-
tribute content to their own and others’ displays and also to
control the content on their own display. The resulting sys-
was to maintain a separation between content providers and
display owners. To support this separation we enable users
to adopt the following roles—that of a content provider or
that of a display owner:
1. Content providers generate content in a wide range of for-
mats (images, videos, web pages, live video streams) and
organize their content in logical containers called “Chan-
nels and is able to add and remove content at any time,
schedule the times that a channel is active, suspend it from
use, or delete it. Content providers are not affiliated with
or when content is displayed. Instead, their channels can
ers subscribe to them.
2. Display owners are linked to a number of physical dis-
playsintheinfrastructure, typically, thoseina‘local’phys-
ical space. For example, College officers are considered
display owners for displays located within the College’s
premises and departmental administrators are owners of
the displays in their own department. A display owner
controls the content on their own displays by means of
channel subscriptions. They can subscribe or unsubscribe
be scheduled to be on or off.
The two roles are completely separate from one another al-
though certain users are both content providers and display
owners. For example, a college officer may be a display
owner controlling the displays in the college, and at the same
time a content provider to those displays creating content for
a ‘college news’ channel. In such cases users can create “pri-
vate channels” with content only available for displays under
The general architecture of the e-Channel system imposes
no restrictions on how content is published through chan-
nels and how display owners subscribe to them. A channel
provider may change the content of a channel at any time.
There is no way for a subscriber to preview the content—
they must rely on the channel description and the reputation
of the channel source to inform their subscription decisions.
The e-Channel system provides two user interfaces: a file
system interface that allows content providers to add and re-
move channel content, and a web interface that allows con-
ing parameters, subscriptions and display availability.
Content providers can use the web interface to create chan-
nels and access a dedicated folder on a networked file system
associated with the channel. Once a channel has been cre-
ated, content providers are able to add content to the chan-
nel by dragging and dropping media files into this shared
folder. The aim was to maintain the benefits of the “drop-
box” approach without the need to upload via a web browser.
Users find manipulating content using the file system sub-
stantially less cumbersome than a web-based approach, es-
pecially for large video files and operations involving many
pieces of content. Conceptually, the model is also very sim-
ple to grasp: users know that files in the folder correspond to
their channel’s content; they simply drag them out or delete
them to take them out of the system. Content providers are
able to limit the availability of their channels to certain pub-
lic displays, and to certain dates and times using the web
interface (see Figure 2).
Display owners subscribe their displays to one or more chan-
nels using the web interface. The system produces a sched-
ule for each display based on the set of content items from
(a) Controlling subscriptions on a display.(b) Controlling channel availability.
Figure 2. Screenshot showing the UI that display owners use to control their subscriptions (left), and content creators to control their channels (right).
all of the channels that the display is currently subscribed to.
Our scheduling policy treats all channels equally, although
we prioritise newer content and specific types of content
(such as emergency alerts and interactive applications1).
new / modified / deleted files
Figure 3. Overview of the e-Channel architecture.
Figure 3 shows an overview of the implementation of the
channels system. The web-based parts of the channels sys-
tem are built using PHP. Information about channels and
subscriptions to channels are stored in a MySQL database.
A Python process periodically scans the folder structure con-
taining the channel folders and records changes to the file
system in the MySQL database.
Each channel includes user-defined availability constraints
that are inherited by the content in the channel. These are
expressed in the form of repeating times slots such as “ev-
ery weekday from 8:00 to 17:00”, “every day from 12:00 to
15:00”. In the same manner, each display has similar time
constraints for its hours of operation. The e-Channel sched-
uler scans all of the subscriptions in the system and extracts
the intersection of these time constraints for every channel-
display combination. These form the constraints for when
each content item should be shown. The available content
is pooled together and ordered by the scheduler then dis-
played in a round-robin fashion. The duration of videos is
extracted from the media file, while static media items, such
as web pages and images, are shown for a fixed amount of
time (currently 10 seconds each).
1Discussion of these features are regrettably outside the scope of
experiences of the e-Channel system.
REFLECTIONS ON HOW CHANNELS ARE USED
The e-Channel system has been in open use since May 2008.
The first non-test content was added by a user on 14thMay
and the most recent (at the time of writing) on 11thApril
2011 (1,062 content days, or about 2.9 years duration). Dur-
ing this time the channels system has grown to 81 individ-
ual users in 33 groups who’ve created 3,700 content items
(1,796 unique items) in 102 channels. Figure 4 shows the
duration of each user’s involvement with the e-channels sys-
tem. The current live status of the system is 64 active chan-
nels containing 511 files. 40 of these channels are marked
private, so are not available to displays other than those con-
trolled by the owner. There are 24 public channels available
(188 content items) that any display owner can subscribe to.
proportions of shared and private channels.
Use and sharing of channels
There are considerable differences in the practices of the dif-
ferent user groups and how they use channels, especially
when it comes to sharing content for other displays (shared
channels). The modal number of active channels is 1, but the
maximum is 8 and half the groups have more than 1 channel
(median 2). Content providers typically have few suspended
channels (58% have 1 or more), but in one case (our heavi-
est channel user) there are 13 suspended and only one active
channel. Looking deeper we find that this content provider
has a remit of observing religious events and each channel
is named after a particular festival. All of these channels are
marked private rather than shared, so the channels are just
serving as a tool for organising and controlling the release
of their content.
Another content provider which represents the students in-
terests, has 2 active channels and 4 suspended ones, all their
channels are marked shared. In this case it appears that the
channels reflect the evolving organisational sub-units and
their remits: new channels have been created and the old
ones suspended, as the unit has rebranded itself. Rather
Days in use
0200 400 600 8001000
Figure 4. Breakdown of length of use of the channels system (from date
of first piece of content to most recent). Note more than half the active
users have used the system for 1 year or more, 8 or so over 2 years and
new users continuing to join (active in the last 100 days).
than create new channels, when an update is made content
is changed within the existing channels. This has the asso-
ciated implication that display owners aren’t burdened with
subscribing to get the new content. It also highlights that
once signed up to a channel, the content in that channel may
change—a display owner is very much trusting the source of
the content to continue to meet their expectations.
From Figure 5 we see the relative numbers of operations
on channels logged by the system (note that these are ac-
cumulated over different usage periods as outlined above).
We observe that significantly more channels are added than
deleted. Instead of removing the channels, users seem to
prefer ‘suspending’ them, which removes their content from
the schedule without discarding the channel and its contents.
Groups Chap and Arts make very active use of this mecha-
nism: from the detailed logs we see two other forms of be-
haviour exemplified. Firstly, channels are often suspended
for a few days while content is updated and then activated
again. Secondly, achanneliscreated, immediatelysuspended,
then ‘made live when ready’.
Adjusting the content displayed
Users who can both create channels and have control of a
display’s subscriptions also exhibit different patterns of be-
haviour, as illustrated in Figure 6. The chart represents each
user group’s activity in a horizontal strip with channel ad-
justments, display subscription adjustments and file system
activity (triangles, crosses and dots respectively). There are
several different patterns: the press office (PR) and college
residence (Res 5) make frequent adjustments to their con-
tent but rarely create channels for it and only infrequently
adjust their displays’ subscriptions. Theatre created a chan-
nel, very regularly reviewed their subscriptions and no fur-
Building 1 Building 2
Dep E Dep P
Faculty 1 Faculty 2
Res 1 Res 2Res 3 Res 4Res 5Res 6 Res 7Res 8Res 9
Number of events
(adding, deleting, suspending and activating from suspension).
Summary of channel related operations per user group
ther content appears for nearly 1 year: this was a change of
personnel and a restructuring (Theatre becomes Arts)—note
how Arts uses channels more actively than her predecessor.
Res 1 experimented with the system at the beginning (burst
of content and channel manipulation), then continues to ma-
nipulate channels periodically, rarely adjusts subscriptions
directly, but provides new content every 6 weeks or so via
the file system. Department (Dep E) displays a burst of ac-
tivity, adds content burstily for 1 year then use of the system
stops entirely (the person who learnt the system left, but did
not pass the role on).
Figure 6. Adjusting channels (triangles), adjusting display subscrip-
tions (crosses) and modifying content in channels (dots)
Overall, the amount of content added daily is typically small
(10s of files), but there are exceptions (Figure 7). Spikes,
mally slide decks saved as galleries of images or (as in this
case) a library of photographs added for an open day. Con-
tent either has a short lifetime, most commonly 7–10 days
or lasts for many weeks. There is a second peak at around
120 days, corresponding to the teaching cycles of the Uni-
versity. The third, lesser, peak is roughly bi-monthly. These
peaks closely match the content validity periods we identify
(see the following section), in which 12% of content images
have a validity of up to 1 week (a further 11% between 1
week and 1 month) and 7% a validity of 2–3 months.
2008-052008-062008-072008-08 2008-092008-10 2008-112008-12 2009-012009-02 2009-032009-04 2009-05 2009-062009-072009-082009-09 2009-102009-11 2009-12 2010-01 2010-02 2010-032010-04 2010-052010-06 2010-072010-08 2010-092010-10 2010-112010-12 2011-012011-022011-032011-04
(a) Profile of files and file types added per month
(b) Length of time content is in the system
Figure 7. Volume and lifetime of content
Gaming the system
One surprise for us when we analysed the content in the sys-
tem was that over the lifetime of the system we found sev-
eral duplicate files (affecting 11% of the content), not just
in separate channel folders, but within the same channel. In
exceptional cases we have found as many as 6 copies of the
same file. This appears to be users deliberately trying to ma-
nipulate the system to offer ‘implicit priorities’ so that their
content appears more often (which it will!). Even though
only a very small number of content providers seem to do
this systematically, our instinctive reaction was to plan mea-
sures to prevent this behaviour e.g. based on content finger-
printing, but we have since noticed that the most replicated
item was an important health message that needed more air
time—so our instinct would’ve been too hasty.
Usage lessons learned
• User engagement follows very varied patterns. We note
that even within our user community there are very differ-
ent patterns of engagement and usage. In particular, we
note that the frequency of content generation varies dra-
matically and that for many users the system needs to be
optimised for occasional use. This suggests that our ap-
proach of a very simple interface is likely to be of benefit
to many of our users.
• Users develop a sophisticated understanding of the sys-
gests that some users develop fairly sophisticated mental
models of how the system behaves and are able to ma-
nipulate the scheduling algorithms for their own benefit.
Addressing this may have unintentional consequences as
we discovered—but equally it is clear that signage sys-
tems that can be manipulated by sophisticated users will
• The perceived lifetime of content is bi-polar—very short
term or very long term. While our sampling of content
suggested a range of times for which content might be
valid the behaviour of users indicated content was either
users either create content for a very specific event or they
create content for an event in the future and essentially
forget about the content once it is in the system. Mech-
anisms for indicating the lifetime of content at creation
time would help to overcome this problem which mani-
fests itself as out of date content appearing on screens.
REFLECTIONS ON CONTENT IN E-CHANNELS
Reflecting on the types of content people add to the system,
we find that of the 1,796 unique content items (Figure 8),
there are 1,494 images, 128 videos, 73 web pages, 6 live
video streams and 95 unsupported files.
Images are clearly the most popular content type and this is
unchanging throughout the duration of the deployment (Fig-
ure 7). In fact, it seems that over time the diversity of content
types is reducing—perhaps due to the extra effort involved
in developing other forms of content—increasing the dom-
ination of images as the primary content format. We ex-
amined a random sample of 100 images for format, time and
location-sensitivity, intended audience, provider identity and
purpose. 95 of the sampled images appeared to have been in-
tentionally added to the channel. 5 of the items appeared to
Building 1Building 2
Dep E Dep P
Faculty 1Faculty 2
Res 1 Res 2 Res 3Res 4 Res 5Res 6
Building 1Building 2
Dep EDep P
Faculty 1Faculty 2
Res 1Res 2Res 3Res 4Res 5Res 6
Figure 8. (a) Content items added to the channel system by provider group and (b) ratio of content types by provider group.
have been accidentally introduced, these included ‘a button
image’ that had presumably been saved as a supporting im-
age for a webpage, without the realisation that it would be
scheduled as an independent content item.
Formatting. The files were largely in JPEG format (94%)
and were in a resolution and aspect ratio appropriate to the
the specific display in mind (80%). 76% of items were de-
signed with some common style or identity (although only
55% of content items contained a textual or iconic reference
explicitly telling the viewer who the identity belonged to).
Temporal and locative properties. 40% of content items had
no obvious time constraints, of the remaining items 17% had
a validity of one day (although these typically referred to an
event that might be repeated—an open day or theatre perfor-
mance, extending the validity), 12% had a validity of up to
week, 11% up to a month, 7% a validity of 2-3 months and
6% a validity of 1 year. The remaining 7% were valid for
the duration of a building project (typically up to one year
in length). 95% of the content appeared to be University
specific (the remaining external 5% was still targeted to a
University audience rather than, e.g. referring to local or na-
tional events or news, as commonly seen on digital signage).
The majority of items were not tailored to particular display
location (approximately 20% of the deliberately placed im-
ages would have made little sense outside of their intended
Purpose and audience. The majority of images had a sin-
gle clear purpose: only 3 of the 100 images appeared to
serve multiple purposes and the purpose of 6 further images
could not be identified. Purpose was closely tied with audi-
ence: 59% of the identified purposes concerned raising in-
Raise awareness of an event (future)
Raise awareness of an event (past)
Raise awareness of service
Raise awareness of departmental project
Raise awareness of building project
Information about student experience
Table 1. Breakdown of image purposes.
Microsoft Office/Open Office files
iTunes video (.m4v)
Windows shortcut files (.lnk)
Supporting files (.swf/.js)
Windows thumbnail cache (.db)
Table 2. Table of unsupported file types
ternal awareness of things within the university; the remain-
ing items were more broadly directed, containing items that
may be of interest to visitors as well as members of the Uni-
versity (Table 1). Some content items were directed at sub-
groups within the University but these were typically large
subgroups(e.g. 3toallteachingstaff, 8toallstudents)rather
than smaller ones (e.g. 4 to members of a particular college).
Of the 27 channel providers, 11 (41%) had submitted videos
to the channel system. Typically the number of videos per
provider was significantly smaller than the quantity of im-
ages, but for three providers (the theatre, student recruitment
and a volunteering unit) videos represented over 50% of the
content they produce for the system. In the case of the the-
atre this is clearly due to the utility of the format for showing
trailers and excerpts from upcoming/previous performances.
Student recruitment used video to produce student profiles.
The volunteering unit primarily used video to highlight their
rebranding: in this final case, the videos were also used to
extend the content beyond its normal length—animations of
approximately 15 seconds were looped for 1.5 minutes. The
median video length was 96 seconds in length (1stquartile:
43 seconds, 3rdquartile: 180 seconds) and the longest video
submitted was 13 minutes long. All of the longer videos
(over 5 minutes in length) were created either by the theatre
or from public (performing) arts—again demonstrating the
usefulness of video for performance-related purposes. 61%
of the files were in the appropriate aspect ratio for the dis-
plays (16:9) and 33% were in 4:3.
3 used this as the dominant format for their content) either in
the form of HTML files dropped directly into the file system
or (more commonly) as web URLs contained within .url or
.webloc files. A sample of 15 files was analysed. Web pages
were the only content format to include non-University spe-
cific items but these still only accounted for 3 files (20%)
and one of these was still of local interest (a location specific
weather report), the remaining two files were national news
and the web page of a catalogue shopping company. Of the
remaining pages 25% were locally relevant pages on popu-
lar social networking sites (e.g. a welcome video uploaded
to YouTube or departmental Twitter page) and the remain-
ing pages were typically those belonging to the department
or other entity owning the channel. One page produced a
404 error and we have also observed other HTTP errors dur-
ing the time the system has been deployed—a risk for this
kind of content that does not apply to other medias (particu-
larly risky when the web pages themselves are not under the
control of the content provider). Furthermore, whilst Web
content created for other purposes can easily be recycled as
a source of content for the e-Channel system this is obvi-
ous when the content appears on the display. Unlike the
images and videos, only 20% of the Web content was ob-
viously created with the displays in mind and two of these
were videos. Scheduling videos via the Web in this way
also creates additional problems: firstly, that Web content
is shown on the display for a default period of 10 seconds,
most likely clipping the video content within the page (un-
like videos dropped into the file share which are shown for
the duration of the video). Secondly, that many web pages
with embedded videos are not set to ‘autoplay’ when the
page is loaded, meaning that the video does not play at all:
this leaves viewers looking at the title and/or initial frame.
Live video streams
All 6 streams were submitted by the development team on
and submit this type of content. All were used for live events
(sports competitions, graduations and a ‘freshers fair’).
Unsupported content within the channel system can be di-
vided into two groups: deliberately placed unsupported files
for (85%), such as those created by office applications (in
these cases it is assumed that the user was unaware that the
file would not be supported) and invisible operating system
or application files (where the user was unaware that the
file was being transferred) (Table 2). Figure 7 suggests that
whilst unsupported content continues to be dropped into the
system, the quantity does seem to be gradually reducing as
content providers determine which content formats actually
appear on their displays.
we’ve been asked to remove ‘inappropriate’ content. Once
and another, where a college principal was concerned the au-
dience would mistakenly assume his college was advocating
a particular anti-religious message. More generally, as sys-
tem administrators we approve the addition of new users to
the system and rely on these users to assume responsibility
for moderating the content in their channels.
Content lessons learned
• Content is localised but rarely display specific. We found
that while most content created was specifically targeted
at sub-sections of the University population it was rare for
content to be specialised for a specific display.
• Web pages make for poor signage content. While it should
be obvious that general purpose web pages will not be
particularly compelling sources of content for a signage
system we found that many new user’s first idea of how
to create content for a signage system is to simply show
web pages—often from external sources. Over time they
appear to recognise that this is not appropriate and focus
on the use of images created specifically for the displays.
REFLECTIONS ON DEPLOYMENT & ACQUISITION
The deployment of the e-Channel system started in early
2008. This was a slow process that involved progressive ad-
dition of new user groups into the system over the period of
the next 12 months. Initially we carried out significant test-
ing and monitoring for each new user that we added but as
the system matured adding users became a simple and reli-
able task. We can use the channel system itself to manage
the user groups, often adding them in response to an email
or alongside a user through a web browser. We took the time
to prepare a detailed user guide that has proved worth its
weight in gold in saving time-consuming training (although
we still provide some to those less comfortable with web-
based systems and technology generally). Normally there is
a colleague to show them how to use the system, it is con-
ceptually simple, so there is rarely the need for extra support
and the user guide is often sufficient on its own.
The challenge for content providers, and where we often
have to advise, is finding a tool-chain that allows them to
generate content that they are happy with aesthetically and
in a format that the system supports. The most readily avail-
able tool both in terms of institutional licensing and support,
and familiarity for our users is Microsoft Office. It is unsur-
prising that posters created with Powerpoint are quite preva-
lent in the system. Keener users have purchased specialist
products to allow them to produce more sophisticated and
attention grabbing content, such as animations and videos.
New users often try putting office documents directly into
the system. We can quickly spot this and guide them to ex-
port appropriate formats. Initially we thought not offering
support for office formats placed an unnecessary burden on
content providers, but we now appreciate this forces them to
deliberately create content specifically designed and format-
ted for the screens rather than reusing existing documents.
Over time deploying the software became synonymous with
hardware acquisition as most new users were also installing
hardware at the same time. During this process we noticed a
particular focus on the cost of hardware purchase while to-
tally failing to adequately take into account the cost of con-
tent production. We went to some lengths to alert them to
this issue but it remains a concern with new deployments—
typically hardware is well budgeted for while content pro-
duction and the responsibility for producing content is a real
cost that is not well catered for.
Deployment lessons learned
• System deployment took place over an extended period of
time. While we did not expect the system to be widely de-
ployed and adopted overnight we did not anticipate a de-
ployment period lasting many months. However, the com-
plexity of hardware deployment and, more significantly,
role changes within organisational units can easily lead to
these sorts of delays. This is likely to be the case for many
similar Ubicomp deployments.
• Users fail to appreciate the significance and cost of con-
tent. We found that our users consistently underestimated
the time and expense required to create content. In prac-
tice many also lacked the appropriate skills and training in
this area and careful use of templates becomes a require-
ment to ensure a satisfactory level of presentation.
As display hardware costs have fallen large-scale advertising
networks have emerged [13, 15] and numerous signage sys-
tems to support digital signage installations (e.g. [16, 14]).
Research in public and semi-public displays began in the
1980s with the development and installation of long-lived
audio and video connections between physically separated
laboration and awareness continue to be significant themes
[6, 8, 1] and much of our content analysis suggests this is
true for our display network also—the provision of the e-
Channel system to easily support content from multiple de-
partments (and even multiple contributors within a depart-
ment) does make them a useful tool for raising awareness
between otherwise separated members of the university. A
trend towards multi-user systems is also seen in recent digi-
tal signage solutions .
InthecontextofpublicdisplaysinaUniversitysetting, M¨ uller
et al.  have deployed the iDisplay system. Motivated
by similar requirements that have led to the design of the
e-Channel system, the iDisplay system is designed to offer
explicit support for pre-defined content types, termed “ac-
tionables” (posters offering people to take a specific action).
In the design of the e-Campus system there are no explicit
tem, leaving such mechanisms to be defined by the content
providers and display owners. We believe that this approach
allows more flexibility, enabling users to appropriate the sys-
tem for a wider range of purposes.
One key issue for public and semi-public displays (particu-
larly commercially) is that of relevance and user attention:
Huang et al. observed 46 public displays in a variety of pub-
lic spaces and reported infrequent, brief glancing  whilst
M¨ uller et al.  studied user expectation and its impact
upon the attention users give to displays. They found that
people typically attended to displays only when they ex-
pected the content to be of interest to them (and that typi-
‘interesting’ content included local news, events and infor-
likely to expect interesting content in a university context
than within a commercial premises. Our work considers one
such university display network in which many of the previ-
ously suggested content types are scheduled for user view-
ing on a daily basis. We hope that systems such as ours can
help to reduce the barrier to entry for content creators help-
ing to increase the amount of interesting content in display
More recent research in displays focusses on novel display
technologies , interaction [9, 5] and mechanisms for im-
proving the content relevance [10, 3]. Whilst this work is in-
teresting and useful it does not reflect current everyday digi-
tal signage use. Correspondingly, research into the everyday
usage process for commercially deployed signage systems
in common use is understandingly lacking. In this paper we
attempt to address this gap by providing insight into the us-
age of an experimental digital signage system from the point
of view of the content providers and display owners.
Universities, like many large businesses, are made up of di-
verse stakeholders with differing needs to communicate with
many audiences. In this paper we have reflected on our expe-
riences of a highly successful, long-lasting Ubicomp system
to support digital signage on a University campus. The sys-
tem is open to content from a variety of sources whose needs
and uses have emerged and evolved over a 3 year period, and
continues to be in everyday use. We have reflected on how
the system is used in detail (content and use related prac-
tices) and provided a set of lessons for designers of future
Ubicomp public display systems.
Key to the system’s success is its parsimony and concep-
tual simplicity: it features a simple, flexible design based on
‘channels’ of content and a separation of concerns between
content providers and display owners. A web-based system
offers the stakeholders control, rather than the centralised
authority found in many conventional signage systems. The
simplicity of marrying the web interface with a file system
interface greatly simplifies everyday use of the system and
many content providers update and manage their content this
way. The ability to split between private and shared channels
has enabled them to choose which content is for their display
and which for a more general audience. Activation and sus-
pension of channels provides simple coarse-grained control
over content exposure—and content providers make exten-
sive use of this when they update and prepare their content.
One note on usability can be made here. In an environment
where non-technical people are asked to be involved in the
management of a content dissemination infrastructure “less
is more”. The limited control offered by the e-Channel sys-
tem makes it a lot easier for users to exploit the display net-
work. However, finding the golden ratio between usability
and control is always difficult. Certain content providers
require more control over the scheduling of content within
each channel and potentially even the priority of the content
itself, while at the same time a small number of them still
feel that the e-Channel system is somehow difficult to con-
trol. An observation based on experience is that it is always
pion simple elegant designs. We hope that lessons from our
work can be taken forward to help other Ubicomp system
designers achieve long lasting and usable systems.
Seventh Framework Programme (FP7/2007-2013), grant ref.
244011 and Lancaster University. Thanks to Oliver Storz
for his contribution to the e-Campus infrastructure, and our
anonymous reviewers for their constructive remarks.
1. BARDRAM, J., HANSEN, T., AND SOEGAARD, M.
Awaremedia - a shared interactive display supporting
social, temporal and spatial awareness in surgery. In
CSCW ’06 (2006), pp. 109–118.
2. BLY, S., HARRISON, S., AND IRWIN, S. Media
spaces: bringing people together in a video, audio, and
computing environment. Commun. ACM 36, 1 (1993),
3. BUZECK, M., AND M¨ ULLER, J. Twittersigns:
microblogging on the walls. In Proceedings of the
international conference on Multimedia (NY, USA,
2010), MM ’10, ACM, pp. 819–822.
4. CHANDLER, A., FINNEY, J., LEWIS, C., AND DIX,
A. Toward emergent technology for blended public
displays. In Proceedings of the 11th international
conference on Ubiquitous computing (2009), Ubicomp
’09, pp. 101–104.
5. ERBAD, A., BLACKSTOCK, M., FRIDAY, A., LEA,
R., AND AL-MUHTADI, J. Magic broker: A
middleware toolkit for interactive public displays. In
Proceedings of the 2008 Sixth Annual IEEE
International Conference on Pervasive Computing and
Communications (Washington, DC, USA, 2008), IEEE
Computer Society, pp. 509–514.
6. GREENBERG, S., AND ROUNDING, M. The
notification collage: Posting information to public and
personal displays. In CHI 2001 (2001).
7. HUANG, E., KOSTER, A., AND BORCHERS, J.
Overcoming assumptions and uncovering practices:
When does the public really look at public displays? In
Pervasive 2008 (2008).
8. HUANG, E., TULLIO, J., COSTA, T., AND
MCCARTHY, J. Promoting awareness of work activities
through peripheral displays. In CHI 2002 (Extended
Abstracts) (NY, USA, 2002), ACM, pp. 648–649.
9. M¨ ULLER, J., ALT, F., MICHELIS, D., AND SCHMIDT,
A. Requirements and design space for interactive
public displays. In Proceedings of the international
conference on Multimedia (NY, USA, 2010), MM ’10,
ACM, pp. 1285–1294.
10. M¨ ULLER, J., EXELER, J., BUZECK, M., AND
KR¨ UGER, A. Reflectivesigns: Digital signs that adapt
to audience attention. In Pervasive ’09 (2009), Springer.
11. M¨ ULLER, J., PACZKOWSKI, O., AND KR¨ UGER, A.
Situated public news and reminder displays. In
Proceedings of the 2007 European conference on
Ambient intelligence (Berlin, Heidelberg, 2007),
AmI’07, Springer-Verlag, pp. 248–265.
12. M¨ ULLER, J., WILMSMANN, D., EXELER, J.,
BUZECK, M., SCHMIDT, A., JAY, T., AND KR¨ UGER,
A. Display blindness: The effect of expectations on
attention towards digital signage. In Pervasive ’09
(Berlin, Heidelberg, 2009), Springer-Verlag.
13. NEO BOOKING SA. Bookingdooh-
digital-out-of-home online planning and booking
14. REMOTE MEDIA LIMITED. digital signage software —
signagelive delivering media to your screen.
15. SEESAW NETWORKS INC. Place-based digital video
16. SONY CORPORATION. Ziris digital signage.