AIRSF: A New Entertainment Adaptive Framework for Stress Free Air Travel

Hao Liu, Jun Hu,  Matthias Rauterberg

Technische Universiteit Eindhoven
P.O. Box 513
5600 MB Eindhoven, Nederlands

{hao.liu, j.hu, g.w.m. rauterberg@tue.nl

ABSTRACT

Travel by air, especially long distance, the combination of long flight duration, limited space and an unusual cabin environment causes physical and psychological discomfort and even stress for a large group of passengers. In-flight entertainment systems are commonly installed on the long haul flights to increase the passenger’s comfort level.  However, the current installed and commercially available in-flight entertainment systems do not explore how the entertainment services can be adapted to reduce the passenger’s stress level systematically and intelligently. Also, these systems are designed and implemented based on a pre-set concept of what customer likes and requires as a homogeneous passenger group that has similar tastes and desires. In this paper, we present a new entertainment adaptive framework AIRSF for stress free air travel. Compared to the current in-flight entertainment framework, it can regulate the passenger’s physical and psychological states at comfort physical and psychological states with context-aware and personalized stress reduction entertainment service provision intelligently; What is more, based on the passenger’s bio and explicit feedback, it can automatically track, learn and adapt to the passenger’s preferences.

Categories and Subject Descriptors

H.1.2 [Models and Principles]: User/Machine Systems – Human factors, Human information processing.

General Terms

Design, human Factors, languages.

Keywords

Adaptive framework, context-awareness, in-flight entertainment.

1.     Introduction

Travel by air, especially long distance, is not a natural activity for humans.  Many people experience some degree of physiological and psychological discomfort and even stress when flying.  Excessive stresses may cause the passenger to become aggressive, over-reactive and even endanger the passenger’s health (Sophia 1998; WHO 2005). Many airlines have realized the potential of the on-board entertainment in improvement of customers’ satisfaction level. However, the current installed and commercially available in-flight entertainment systems do not explore how the entertainment services can be used to reduce the passenger’s negative stress systematically and intelligently to enable him/her a stress free trip. The entertainment services are simply delivered in forms of high-speed communication tools and state of art entertainment, which include audio/video on demand, games, in-flight email, internet access and ever-increasing digital entertainment options. This provides mental distraction and might lead to reduction or increase of psychological stresses; Also, these entertainment systems are built based on preset concept what customer likes and requires a homogeneous passenger group that has similar tastes and desires (Liu 2006). They present the same user entertainment interface and contents to each passenger. If the user wants to get personalized entertainment services, he/she needs to interact with systems to get desired entertainment services from the provided options. Regularly if the available choices are many and the interaction design is poor, the passenger tends to get disoriented and not manage to find the most appealing entertainment services.

   In this paper, we present a new entertainment adaptive framework AIRSF for stress free air travel. It integrates the concepts of context adaptive systems, user profiling, and methods of using entertainment services to reduce the user’s negative stresses into a linear feedback control system. In case of the linear feedback system, a control loop, including sensors for the passenger’s physical and psychological, context of use signal acquiring and modeling, adaptive control unit for entertainment service adaptation strategies, entertainment service provision, etc. components, is arranged in such a fashion as to try to regulate the passenger’s physical and psychological states at comfort states with context-aware and personalized stress reduction entertainment services. What is more, based on the passenger’s bio and explicit feedback, it can track, learn and adapt to the passenger’s preferences. The more the passenger uses the in-flight entertainment system, the better user preference can be built based on the mining of the passenger’s past behaviors. Thus the more personalized services can be provided by the system.

   This paper is organized as follows: In section 2 the new entertainment adaptive framework for stress free air travels AIRSF is presented. Following section 2, a case study is given in section 3. Finally, in section 4 the main conclusion is drawn and the future work is discussed.

2.     Related works

2.1     An overview of the current in-flight entertainment systems

After World War II commercial aviation flights became a daily event in which entertainment was requested by passengers to help the time pass. This was delivered with a projector movie during lengthy flights. The in-flight entertainment systems were upgraded to the CRT-based systems in the late 1980s and early 1990s. In the mid 1990s, the first in-seat video systems began to appear, and LCD technology started to replace CRT technology as the display technology of choice for overhead video. In the late 1990s and early 2000s, the first in-seat audio/video on-demand systems began to appear (Wiki 2008). Today, with the technology development, except for audio/video on-demand services, the entertainment services are also delivered in forms games, in-flight email, internet access and ever-increasing digital entertainment options.   

Figure 1. In-seat LCD-based in-flight entertainment systems (n.d.)

  In this subsection section, firstly, the current in-flight entertainment systems in the aircrafts of major airlines in the world are investigated. After that, the latest commercially available in-flight entertainment systems which are provided by major players in this field are investigated. Finally, how the current in-flight entertainment systems are designed and implemented to reduce the passenger’s negative physical and psychological stresses are analyzed and their limitations are discussed.

2.1.1 Current installed in-flight entertainment systems

To allow each airline the freedom to configure its aircrafts according to its budges and market demands, both airplane producers (Boeing and Airbus) and major in-flight entertainment system providers provide customized in-flight entertainment system to their customers. Liu (2006) investigated the current installed in-flight entertainment systems in the aircrafts of Airlines of Lufthansa, Air France, British Airways, American Airlines, Delta Airlines, and Japan Airlines which are top airlines in Europe, North America and Asia from Total Scheduled Passengers point of view (WATS, 2006).

   Generally, the in-flight entertainment services provided by these airlines might be divided into two categories: Passive- the user-system interaction levels are very low and the passenger simply enjoys a chosen form of entertainment that is presented to them in an organized and packaged form. Examples of passive entertainment services are audio and video on demand, audio and video broadcasting, e-book, moving up systems, etc. Active - the user spends time to actively interact with the entertainment system where the following entertainment service content is determined by the interaction between the user and the system. Examples of this type of entertainment are: gaming, etc. The exact entertainment services provided by an airline during air travel depend on a number of factors such as the aircraft type, business model of the airline, class seats (first class, business class, and economy class), etc.

   All the in-flight entertainment systems installed in the investigated airline’s aircrafts present the same interface and entertainment contents to each passenger no matter those passengers come from highly heterogeneous pools (such as age, gender, ethnicity, etc.), have different individual entertainment preferences. If the user wants to get desired entertainment services during air travel for recreation, he/she needs to interact with the in-flight entertainment system by means of touch screen, in-seat controller (refer to figure 2) to browse and select the desired entertainment services from the provided options. If the user selects a game to play, he/she can use the remote controller to interact with the system to play the game. In this situation, on the one hand if the available choices are many and/or the passenger himself/herself is not familiar with the service category structure, and/or the interaction design is poor (for example, there are more than twenty keys on the Japan Airline’s remote controller for the game control, audio and video on command control), the passenger tends to get disoriented and not manage to find the most appealing entertainment services; on the other hand if the available choices are limited (for example, most airlines investigated only provide several movies during air travel), the chance for the passenger to find desired entertainment services is slim. Under these circumstances, the in-flight entertainment system does not contribute to improve the passenger’s comfort level, on the contrary, to some extends, it exacerbates the situation.  

              

Figure 2.  Interactions between the passenger and the in-flight entertainment system (n.d)

2.1.2 Commercially available in-flight entertainment systems

In (Liu 2006), Liu investigated the latest commercially available in-flight entertainment systems which are provided by major in-flight entertainment system producers Panasonic Matsushita, Thales and Rockwell Collins. Panasonic Matsushita (n.d.) X-series in-flight entertainment system is the first in-flight entertainment system to be based on the research of passenger preferences and consumer trends worldwide to optimize the digital entertainment options for passengers. Rockwell Collins (n.d.) provides several TES series In-flight entertainment systems. Among them eTES collects usage statistics to assist airlines in determining an optimal content mix, thereby minimizing content costs and maximizing passenger satisfaction.

2.1.3 Discussions and conclusions

In this section, firstly, seven major airlines’ current installed in-flight entertainment systems are investigated. The following conclusions are drawn on these systems:

1.        All the airlines investigated did not explore how the entertainment services can be used to reduce the passenger’s negative physical and psychological stresses systematically and intelligently. The investigated in-flight entertainment systems contribute to the passenger’s psychological stress reduction by relaxing or distracting him/her with interesting reading materials, movies, music or games available on the aircraft’s in-flight entertainment system; For the passenger’s physical stress reduction, considering the limited space and safety constraints, the airlines usually provide some on chair physical exercise tips to the passenger either in paper flyers in front of the passenger’s seat or in electronic texts in the entertainment systems (n.d.). However, according to our investigation in most cases passengers tend to ignore these exercise tips. Therefore a more engaging solution is necessary.

2.        All these airlines provide video/audio on demand systems. The passenger must interact with the system to browser the menu and select the desired audio/video programs from the provided options. There is no context-aware and personalized entertainment service recommendation.

     The latest commercially available in-flight entertainment systems provided by major players Matsushita, Rockwell Collins and Thales are trying to assist  airlines to provide optimized digital entertainment options to meet passengers’ entertainment needs by collecting usage statistics, research on consumer trends etc. However, as shown in figure 3, these systems still did not explore the passenger’s implicit entertainment requirements implied by his/her personal profile, bio signal and the aircraft flight situations, etc. to recommend the passenger situational aware personalized entertainment services. Also these systems did not explore how the entertainment services can be used to reduce the passenger’s negative stresses systematically.

Figure 3. The adaptive relation between In-flight entertainment (IFE) system producer, Airlines, Passenger and IFE system

2.2     Related works of stress reduction with entertainment provision

In this paper, we concentrate on related works of using music and gaming to reduce the passenger’s negative physical and psychological stresses.

2.2.1     Using music to reduce a user’s psychological stress

There is a long literature involving the use of music for reducing the user’s psychological stress. Miluk-Kolasa et al. (1996) showed that music was one of the relaxing adjuncts in modulating the ascent of autonomic responses to negative stress. Knight and Rickard (2001) revealed that relaxing music attenuated blood pressure and heart rate after a stressful task; moreover, the level of subjective anxiety was reduced after the presentation of relaxing music. The tempo of the music being listened to appears to be an important parameter here. Steelman (1991) looked at a number of studies of music's effect on relaxation where tempo was varied and concluded that tempos of 60 to 80 beats per minute reduce the stress response and induce relaxation, while tempos between 100 and 120 beats per minute stimulate the sympathetic nervous system. White and Shaw (1991) reported similar results and argued that tempos slower than the average human's heart rate (40 to 60 beats per minute) induce suspense, while tempos of 60 beats per minute with a low pitch are most soothing. Stratton and Zalanowski (1984) conducted experiments and found that preference, familiarity or past experiences with the music have an overriding effect on positive behavior change than other types of music.

2.2.2     Using game to reduce a user’s physical stress

Prolonged immobility by sitting in a chair during long haul flights can lead to pooling of blood in the legs. It is known that immobility causes physical discomfort in general and formation of blood clots in the body and especially in deep veins (Deep Vein Thrombosis, or DVT in short). To reduce physical stress some airlines recommend passengers in-flight exercises (n.d.). The main goal of these exercises is to stimulate the blood flow and prevent health related issues like deep vein thrombosis (DVT), stiffness and fatigue. The problem with these recommendations is that it depends on the passenger whether the exercises are executed or not and can easily be ignored. Therefore a more engaging solution is necessary. In (Westelaken 2008), Westelaken presents a new interactive in-flight game chair. If the passenger has been in immobility state for a long time, the game chair would recommend him/her to play games. If the passenger accepts the recommendation to play games, he must move with the pre-defined in-flight exercise patterns to play them.

3.     AIRSF: A new entertainment adaptive framework for stress free air travel

Figure 4 present our new in-flight entertainment adaptive framework AIRSF. In the figure, the framework starts by setting the passenger’s physical and psychological target comfortable states. Then, the system begins observing the passenger’s current physical and psychological states (modeled on the passenger’s bio feedback signals) that it wishes to control. This step of perception creates an internal representation of the passenger’s physical and psychological situation. After that, depending on the difference between the target and the current real physical or psychological state, the adaptive inference component in the framework must determine (1) whether the passenger is in the target state or not; and (2) if the passenger is not in the target state then optimized entertainment services are recommended based on user preference and available entertainment content etc. information to transfer the passenger from the current state to the target state to improve his/her comfort level.  The passenger himself/herself is an adaptive system; his/her perception creates an internal representation of the entertainment service. This perception affects the passenger’s physical and psychological states. During this process, the passenger’s physical and psychological states may also influenced by the set of variables which in the control system called disturbances (Ashby 1956).  The change in the passenger’s physical and psychological states is again perceived by the system, and this again triggers the adaptation process we have described, thus closing the control loop. User’s entertainment preference depends on context of use which may include the passenger’s physical and psychological states, the activity.   In figure 4, if the system recommends entertainment services that the passenger does not like, he may reject the recommended services and select desired entertainment himself/herself or just shut down the system. By mining on the context of use, selected entertainment services by the passenger, the passenger’s explicit and implicit feedbacks on the recommendation, the system can automatically learn and adapt to the passenger’s preferences, thus the more the passenger uses the in-flight entertainment system, the more personalized entertainment services can be recommended to the passenger intelligently.    

Figure 4. AIRSF: a new entertainment adaptive framework for stress free air travel

3.1     Describing entertainment contents

An entertainment service can be described by a set of attribute/value pairs. It can be expressed formally by an ordered vector   where is the  attribute/value pair.  In this paper, for simplicity reasons, sometimes we represent   as  since   is an ordered vector. For example, each piece of music can be described by a set of attribute /value. Each attribute/value pair describes one aspect of the music. Some attributes contain information mainly for user preferred music recommendation/selection (e.g. title, artist, recording company). Some attributes address the nature of the piece of music (e.g., tempo, rhythm); Games can also be described as a collection of attribute-value pairs just as music description. These attributes provide a general overview, or flavour, of the game. They explain the goal of the game, the basic gist of the story, and some general ideas of the kinds of activities in the game. For example, The ESRB (n.d.), which is set up by the entertainment software industry’s trade association, maintains a two-part rating system for video and computer games: the rating symbol, such as E or M, which suggests the game’s age appropriateness; and content descriptors, like Blood and Gore, which point out specific elements of the game that have caused the rating and that may be of concern.

3.2     Describing context of use

Context of use is a categorization of the actual situation under which the service is delivered by the system. As shown by figure 1.3, one category of the actual situation may be aggregated by several sensors’ datum.  Context of use can be expressed formally as an ordered vector  where  is  category of the actual situation under which the service is delivered by the system.

                                                    

Figure 5 Context of use

3.3     Describing user preference

The information of a user which can reflect his/her NRDs (Needs Requirements and Desires) on the preferred system behaviors explicitly is called a user preference model (Salem, Rauterberg 2004). It is usually be integrated into the system to impart the user preference knowledge to the system to enable automatic personalized system behavior adaptations and avoid “unnecessary” dialogues between the system and the user. For the context-adaptive systems where the context of use is also considered for system behavior adaptation, the main contents of the user profile is a subset of the intersection collection among the real world user model, the available system behaviors and the context of uses (see figure 6). The information items in this sub set can reflect the user’s NRDs on the preferred system behaviors under contexts of use explicitly. 

Figure6. User preference for context-adaptive systems

3.3.1 User preference modeling

   A detailed user preference model has been formally defined for the new entertainment adaptive framework in (Liu 2008). In this model, user preference is modeled by a two-layer tree with dynamic changeable structures. The top layer of the tree is used for modeling user’s long term entertainment service preference. Each node represents user’s long term evolving commitment to certain categories of entertainment service. The lower layer of the tree is used for modeling user spontaneous entertainment service requirement which depends on context of use. Each node relates one context of use to one or more desired entertainment service requirements. The tree is dynamically constructed by the formal “subordination” or “refinement” or “composition” relation definitions among nodes. The advantage of this structure is three folds: (1) it can not only model the user’s long term but also spontaneous preference items. More over, the relations between all the preference items are formally defined; (2) if the number of preference items is many, it is more efficient and easier to find the right preference items; (3) if the user desired service has been removed, the system can utilize the personalized hierarchy service structure of the preference tree to calculate and recommend similar services.

   For simplicity reasons, in this paper we only introduce a piece of user dynamic preference item

 

definition:  

 is defined as  where  is defined as

, 

 is the traditional VSM (Vector space model) to describe the attributes with different weighting schemes where

.

3.3.1 User preference learning

The user’s feedback to the recommended/self selected entertainment services needs to be logged for user entertainment preference learning. In this sub section, we first define a data structure for user feedback logging, and then give a strategy for user feedback information logging.

   A piece of user feedback information is formally expressed by an ordered vector 

 w

here is the context of use, is the entertainment service which is recommended by the system or self selected by the passenger,  is the passenger’s attitude towards the recommended entertainment service, if the user declined the recommendation, its value is 0, otherwise its value is 1,  is the time mark.

   Each time there is a value change of,, or ,  one piece of user feedback information is produced and stored in a database file.

   Based on the logging information, the user preference learning component can learn pieces of user dynamic preference items and build a user entertainment preference tree (Liu 2008). Also, the user preference learning component needs to mine on the user feedback log file to get entertainment service effects on the passenger’s internal physical and psychological state change, thus enabling optimization of entertainment service recommendation.

   A piece of entertainment service effect information is formally expressed by an ordered vector 

 

where is the context of use, is one category of  entertainment services which share the same effect of transferring the passenger from one state to another,  depicts the passenger’s physical or psychological state transfer from one to another, it can be further refined with an ordered vector   where is the passenger’s state before he/she enjoys the entertainment services,   is the passenger’s transferred state after he/she enjoyed the entertainment services and   is not equal to ,  is the cost of transferring the passenger state from  to  .  It may be the time cost of transferring the passenger state from one to another.

3.4     Adaptive Inference

Adaptive inference is the central part of the new entertainment adaptive framework. Depending on the difference between the target and the current real physical or psychological state of the passenger, the adaptive inference  determines (1) whether the passenger is in the target state or not; and (2) if the passenger is not in the target state then optimized entertainment services are recommended based on user preference, entertainment service effect and available entertainment content information to transfer the passenger from the current state to the target state to improve his/her comfort level with the minimum time cost.   

   Markov decision processes (n.d.) provide a mathematical framework for modeling decision-making in situations where outcomes are partly random and partly under the control of the decision maker. So Markov decision processes are a perfect substitute of traditional PID controller for a bio feedback control systems which subjective are users.

   First, a few definitions need to be defined to use Markov decision process to optimize entertainment service recommendation to transfer the passenger from the current state to the target state with minimum time cost.

Definition 1: A passenger’s physical or psychological state space can be formally expressed as 

 

where  and  is one of his/her possible physical and psychological states.

Definition 2: Adaptive inference action space is formally expressed as   

where  and  is one category of its recommended entertainment services (actions) which could enable the passenger’s   physical or psychological state transition.

Definition 3: A transition probability function  is defined as the probability that action in state  at time  will lead to state  at time. It can be formally represented as.

Definition 4:  is defined as the time reward received after transition to state  from state  with transition probability.  equals  where is the time cost of transferring the passenger state from  to  defined in sub section 3.3.1.

Definition 5:   A transition probability matrix  is defined as

     where .

    Based on the above definitions, the optimized entertainment services recommendation with the minimum time cost to transferring the passenger state from current  to the target  could be solved with the Bellman equation:

 

where is the discount rate and satisfies  If we want the passenger to transfer from current state to the target state  with minimum intermediate states,  could be tuned smaller.  Otherwise  could be tuned towards 1. By this Bellman equation, the adaptive inference can compute an optimized action (entertainment service) list to transfer the passenger from the current state to the target state with minimum time cost.

4.     Implementation architecture

Figure 7 shows the main components that make up the new interactive IFE implementation architecture. It implements the framework’s concept presented in section 3. The whole architecture is divided into five abstraction levels from functionality point of view. The lowest level is the resource level which contains entertainment contents such as music, the sensors such as bio sensors and user static profile information such as demographic information. The second layer is the resource manager layer which includes entertainment service manager, context of use manager and user profile manager. The entertainment service manager is responsible for the in-flight entertainment service (such as music and game) registration, categorization, un-registration, etc service management functions. The context of use managercollects and models signals from sensors and updates context of use information in database. It first acquires bio signals from sensors, and then based on these signals to model the passenger’s physical and psychological states and store this information to the database. The user profile manager collects, and updates the user’s static information such as demographic information.  The third layer is the database layer which constitutes by a database. It acts not only as a data repository, but also enables the layers and the components in the layers loosely coupled. This increases the flexibility of the whole architecture. For example, replacing or updating components in the resource manager layer does not affect the architecture performance unless data structures they store in the database changed. The fourth layer is the adaptive control unit layer which includes user feedback log, adaptive inference and user preference learning components.  It is used to mediate between the user profile, context of use and available entertainment contents to provide the passenger personalized reduction entertainment to transfer the passenger from the current physical and psychological state to the target comfortable state. The user feedback log component is responsible for logging the user’s feedback to the recommended entertainment services, the effects of the recommended entertainment services (the time cost of transferring a passenger’s psychological state from one to another), and context of use information for user entertainment preference learning. The user preference learning manger is responsible for user preference learning based on user’s past interactions with the recommended or self selected entertainment services. It forwards learned results to the database for storage. The adaptive inference is the core component of the whole architecture. It is used to mediate between the available entertainment contents, context of use and the user profile etc. information according to Markov decision processes to: (1) recommend the minimum cost (e.g. time cost) personalized stress reduction entertainment services to the passenger to transform him/her from the current physical or  psychological state to the target comfortable state; (2) present personalized stress reduction entertainment service contents according to the passenger’s profile, context of use, expert knowledge, etc. information if the passenger wants to select entertainment services themselves. The fifth layer is the interface layer.  The passenger interacts with the adaptive in-flight entertainment system interface to get personalized entertainment services. 

Figure 7 the implementation architecture of AIRSF

5.     Case study

In the following use case study, the passenger’s heart rate data is used as an indication of his/her psychological stress during air travel. For a child (age 6-15), his/her normal heart rate at rest is 70-100 beats per minutes. For an adult (age 18 and over), his/her normal heart rate at rest is 60-100 beats per minutes. Then for each age group, there are three psychological states: high (100-220), normal (60-100 or 70-100) and low (60-0 or 70-0). The target state is the normal state. We assume that there are seven categories of musicact as actions (music).  Each of them can transfer the passenger from one state to another (see figure 8).  

 

           Figure 8 Transition graph for the use case

   For a passenger, the system starts by recommending stress reduction music according to expert knowledge to the passenger if he/she is in stress. During this process, the system logs the passenger’s feedback on the recommended/self selected music. Then, after the passenger has used the in-flight entertainment system for several times, the system can learn and build the passenger’s preference tree, the music effects on the passenger (rewards), the possibility of categories of music transferring the passenger from one state to another. Thus completes the passenger’s personalized parameters in figure 8.

   During the flying, if the passenger feels stressed and his/her heart rate is high, then the system decides that the best action to get the maximum reward at the current context of use is  to transfer Mr. John’s state from high to low with the minimum time cost where  is a sub categories of music of according to the Bellman equation introduced in section 3.  The system acquires  by searching the user preference tree with the current context of use which is the passenger is at rest and state is high. Then, we assume that after some time, Mr. John’s heart rate back to normal, then according to Bellman equation, the system decides that the best action to get the maximum reward at the current context of use is  where  is a sub categories of music of. The system acquires by searching the user preference tree with the current context of use which is the passenger is at rest and state is high.

The more time a passenger spent on board of an airline’s craft, the better user preference can be built based on the mining of his/her past behaviors. Thus the more personalized services can be recommended by the system and the more “unnecessary” dialogues between the system and the user can be avoided.

6.     Conclusion

In-flight entertainment systems play an important role in improving passengers’ comfort level during air travel.   Today, the current in-flight entertainment systems have made significant progress in providing user preferred entertainment services with user friendly interface, interaction mode design and ever increasing entertainment options, etc.  However, despite all these advances, the current installed and commercially available in-flight entertainment systems surveyed in this paper still has much room to improve to make it to enable stress free air travels with the passenger context-aware and personalized stress reduction entertainment provision. In this paper, we present a new entertainment adaptive framework AIRSF for stress free air travel. Compared to the current in-flight entertainment framework, it can regulate the passenger’s physical and psychological states at comfort physical and psychological states with context-aware and personalized stress reduction entertainment service provision; What is more, based on the passenger’s bio and explicit feedback, it can automatically track, learn and adapt to the passenger’s preferences. After the introduction of the framework and its implementation architecture, we use a case study to showcase our framework’s ideas and validate its claimed advantages.

    In the future, we planned to do the real world test to validate and improve our framework.

7.     ACKNOWLEDGMENTS

This project is sponsored by the European Commission DG H.3 Research, Aeronautics Unit under the 6^th Framework Programme, under contract Number: AST5-CT-2006-030958.

REFERENCES

ArtsyKen (n.d.).  Interactions between the passenger and the in-flight entertainment system. Retrieved March 1, 2008 from ArtsyKen’s Web site:     http://artsyken.com/2003_12_01_archive.php.

Ashby W.R. (1956). Introduction to Cybernetics, Chapman & Hall, London, 1956.

Cleveland Clinic Health System (n.d.). Your Pulse and Your Target Heart Rate. Retrieved June 1, 2008 from Cleveland Clinic Health System’s Web site: http://www.cchs.net/health/healthinfo/docs/0900/0984.asp?index=5508

CNET(n.d.). Photo of in-seat LCD-based In-flight entertainment systems.  Retrieved March 27, 2008 from CNET’s Web site:  http://www.cnet.com.au/dvdpvr/0,239035801,339283273-8s,00.

Continental Airlines (n.d.). In-flight exercises. Retrieved April 8, 2007 from Continental Airlines’ Web site: http://www.continental.com/web/en-US/content/travel/inflight/health.aspx.

ESRB (n.d.). Game ratings. www.esrb.org/ratings/index.jsp

Knight WEJ. and Rickard NS. (2001). Relaxing Music Prevents Stress-induced Increases in Subjective Anxiety, Systolic Blood Pressure, and Heart Rate in Healthy, Males and Females’, Journal of Music Therapy 38(4), pp. 254–72.

Liu H. (2006). State of Art of In-flight Entertainment Systems and Office Work Infra structure. Deliverable 4.1 of European project Smart tEchnologies for stress free Air Travel, Technical university of Eindhoven, 2006.

Liu H., Salem B., Rauterberg M. (2008). Adaptive User Preference Modeling and Its Application to In-flight Entertainment. Submitted to SMAP2008.

Miluk-Kolasa B., Matejek M. and Stupnicki R. (1996). The Effects of Music Listening on Changes in Selected Physiological Parameters in Adult Pre-surgical Patients, Journal of Music Therapy 33, pp 208–218.

Panasonic Matsushita (n.d.). In-flight entertainment systems. Retrieved November 25, 2006 from Matsushita’s   Web site:  http://www.mascorp.com/products.html.

Rockwell Collins (n.d.). In-flight entertainment product catalog. Retrieved November 25, 2006 from Rockwell Collins’s   Web site:  http://www.rockwellcollins.com/ecat/at/ xxProductList.html?smenu=3.

Salem B., Rauterberg M. (2004). Multiple User Profile Merging (MUPE): Key Challenges for Environment Awareness. Lecture Notes in Computer Science, vol. 4161, pp. 103-116.

Sophia K. (1998). Sky rage. Flight safety Australia, July 1998,  pp. 36-37.

Steelman VM. (1991). Relaxing to the beat: music therapy in perioperative nursing, Today’s OR Nurse, Vol. 13, pp.18-22.

Stratton VN. and Zalanowski AH. (1984) ‘The Relationship between Music, Degree of Liking, and Self-Reported Relaxation’, Journal of Music Therapy 21(4): 184–92.

WATS (2006), World Air Transport Statistics Special 50th Edition, International Air Transport Association.

Westelaken R., Hu J., Liu H., and Rauterberg M. (2008). Integrating gesture recognition in airplane seats.. Proceedings of the Edutainment Conference, Nanjing, LNCS, page to appear, 2008. 

White J. and Shaw C. (1991). Music therapy: a means of reducing anxiety in the myocardial infarction patient, Wisconsin Medical Journal, pp. 434-437.

Wikipedia, Markov decision process, retrieved June 1, 2008 from Wikipedia’s Web site: http://en.wikipedia.org/wiki/Markov_decision_process

Wikipedia,  In-flight entertainment, Retrieved March 1, 2008 from Wikipedia’s Web site: http://en.wikipedia.org/wiki/In-flight_entertainment

Wired blog network (n.d.). Interaction between the passenger and the in-flight entertainment system. Retrieved March 1, 2007 from Wired blog network’s Web site:

 http://blog.wired.com/gadgets/2007/08/virgin-america-.html.

World Health Organization (2005). Travel by air: health considerations. Retrieved March 1, 2008 from World Health Organization’s Web site: http://whqlibdoc.who.int/publications/2005/9241580364_chap2.pdf.