论文部分内容阅读
Abstract
In recent years， high?resolution video has developed rapidly and widescreen smart devices have become popular. We present an Android application called WeWatch that enables high?resolution video to be shared across two mobile devices when they are close to each other. This concept has its inspiration in machine?to?machine connections in the Internet of Things （loT）. We ensure that the two parts of the video are the same size over both screens and are synchronous. Further， a user can play， pause， or stop the video by moving one device a certain distance from the other. We decide on appropriate distances through experimentation. We implemented WeWatch on Android operating system and then optimize Watch so battery consumption is reduced. The user experience provided by WeWatch was evaluated by students through a questionnaire， and the reviews indicated that WeWatch does improve the viewing experience.
Keywords
together watching experience； screen adaptation； internet of things； distance estimation； energy efficiency
W1 Introduction
ith the rapid development of the mobile industry and growth in the number of high?resolution videos， screen size has become an important factor in product design. Widescreen mobile devices have become more and more popular， but this has led to many problems； for example， a large screen makes one?handed operation more difficult [1]. Nowadays， researchers are trying to improve the viewing experience on smart phones and tablets. The following example highlights the goals we are trying to achieve：
Henry has an old smart phone with a small， low?resolution screen， so he buys a smart phone with a bigger screen. He is watching a movie when：
·He puts the old phone and new phone side by side to get a bigger screen， and the movie plays across the two screens. In this way， he can enjoy the movie on a larger overall screen.
·Someone knocks at the door. He moves one phone away from the other and the video pauses. When he comes back， he only needs to put the two phones together again to pick up where the movie left off.
In this paper we propose WeWatch， an application for viewing videos across two devices with different screen sizes and resolutions （Fig. 1）.
WeWatch provides a better viewing experience than a video player on a single device. It encourages people to use idle devices and share videos. WeWatch has two unique features： screen size adaptation and the use of distance to control video playback. Previous studies on video adaptation mainly focus on optimally matching the resources and capabilities of diverse client devices [2]. However， in order to aggregate resources， these client devices need collaboration and synchronization. This is difficult because different devices have different screen sizes and resolutions， and it is also difficult to transform resources so that they match different devices. To solve this problem， we propose an adaptation algorithm that only matches two devices instead of matching both resources and devices. Our adaptation algorithm can match the size of video windows on both devices so that the video only needs be divided into two parts of the same size. When WeWatch is launched on one device， the device transmits its screen size and resolution to the other device， which automatically adjusts itself.
We have devised a special way of using our application： changing the distance between devices. With WeWatch， a video can be paused， played or stopped by moving the devices certain distances from each other. The received signal strength indicator （RSSI） of Wi?Fi is used to measure these distances [3]. Wi?Fi signals are usually used for location?based services [4]， [5] but are rarely used for managing applications.
A demo of WeWatch is available online [6]. WeWatch works on two smart devices running Android [7]， [8]. We choose this platform because it is popular and open?source.
We have also tested the power consumption of WeWatch and found that it consumes about 40% less power than a normal video player， both in single?screen mode and dual?screen mode. Different parameters have obvious effects on battery consumption. To find the empirical Wi?Fi RSSI values of distances， we determine the relationship between them through real?life simulations. The results help us improve the implementation of the application.
The contributions of this paper are：
·It describes one of the first attempts to combine screens for a better viewing experience. Our scheme does not require any special hardware； it can be implemented on existing mobile devices.
·It describes a new adaptation algorithm that swiftly matches the screen size and resolution of different mobile devices.
·It describes a positioning method， which is used for determining suitable distances for playback control， and a synchronization method for smoother viewing experience.
·It describes a prototype of the screen?sharing system that has been evaluated for energy efficiency and user experience. 2 Related Work
The Internet of Things （IoT） comprises interconnected， uniquely identifiable computing devices embedded within the existing Internet infrastructure. Communication will expand to human?to?thing and thing?to?thing communication. Combining devices is a basic concept in the loT. Much attention has been paid to the IoT in recent years [9]-[11]； however， IoT applications are largely limited to intellectual storage management， public security， and automation [12]， [13]. Not enough attention has been paid to using this new technology in daily life. WeWatch draws on the idea of the loT to combine two mobile devices and give people a better viewing experience.
The loT includes RFID technology， sensor network， detection technology， intelligence technology， and nanotechnology [14]， [15]. Machine?to?machine connections are an important part of the loT， and WeWatch combines two smart devices using Wi?Fi ad hoc technology [16]， [17].
There are many existing systems and toolkits that enhance user experience on mobile devices [18]， [19]. However， most of these are mainly designed for one?way connections， i.e.， resources connecting to devices. WeWatch provides a higher level of communication by grouping entities together， i.e.， devices connecting to other devices. WeWatch can be used directly between portable devices and does not require Internet access.
Some toolkits are used to develop collaborative mobile applications [20]， collaborative and synchronous video annotation platforms [21]， and even impromptu assemblies of a logical computer from the best set of wireless components available nearby [22]. We have the same starting point， i.e.， to provide convenience through collaborative devices.
A proximeter [23] enables mobile proximity?based content sharing on portable devices. This sharing occurs when consumers of content move within the proximity of the owner of the content. A proximiter also creates opportunities for users to chat with their neighbors through their phones. The proximeter prototype is based on Meamo v4.0， which is an operating system in its infancy. Meamo is only supported by Nokia， and there is a lack of applications for the operating system.
Shu Liu and Yingxin Jiang [24] focused on face?to?face proximity estimation using smartphones and Bluetooth. They explored the relationship between Bluetooth RSSI and distances through a series of experiments. They compared the accuracy of Bluetooth， Wi?Fi and GPS using real?world measurements. However， it is impractical to use Bluetooth to measure the distance between devices. MobiUS [25] has developed a video application for displaying a high?resolution video across two mobile devices. The application is based on an efficient， collaborative half?frame decoding scheme. Because the collaborative system architecture is tightly coupled， the application only works on devices with the same screen size and resolution.
Content adaptation for mobile devices is usually directed towards online multimedia [26] or pictures [27]. With content adaptation， content of different sizes and resolutions can be adapted to different screens. However， researchers have rarely considered the implications of combining screens.
3 System Design
We develop WeWatch from both the client side and server side. Our application can decide the role of each device automatically when launched. Fig. 2 shows the WeWatch architecture.
The client and server are linked by Wi?Fi ad hoc networking technologies， which enable direct device?to?device communication. This helps reduce the traffic load in the wireless infrastructure and improve user experience. The role of devices can be easily decided because the hotspot and Wi?Fi cannot be turned on at the same time [28]， [29]. Some researchers have examined the trade?off associated with using Wi?Fi ad hoc and infrastructure modes. Their experiments show that mobile devices are stable when communicating directly with each other [30]. With our WeWatch， a user can choose a device to be a hotspot， with other connecting to that hotspot. We separate the ad?hoc function from our application so that a user can exploit it to make new functions.
The client and server sides can send requests to share videos， and these requests are executed after being authenticated by the valid distance from empirical RSSI values.
WeWatch comprises three modules： positioning， adaptation， and synchronization. The positioning module at the client side records the RSSI values and uses these to estimate the distances. Then， it changes the video mode and notices other modules. The adaptation module sends the screen size and resolution between the two devices so that the display across the two screens is the same size. The synchronization module ensures both screens are playing the same video frames at the same time.
3.1 Positioning Method
The positioning method used in WeWatch frees our hands from having to click buttons. Specific functions are initiated by dragging one device away from the other， changing the distance between them. Different distances are estimated according to RSSI values. For example， the video is played when the distance between devices is within 10 cm， the video and the video is paused when the distance is more than 10 cm. To quit dual?viewing mode， one of the devices device has to be moved more than 30 cm away. All these operations are reversible. Device?to?device Wi?Fi ad hoc communication enables the client side to be directly connected to the server side. The client side records the signal values sent from the server side. These values are reliable enough to estimate the distance between them. Wi?Fi triangulation is not accurate enough to detect exact position within a very close range [31]； however， we can use it to detect approximate position within a range， e.g.， within 10 cm， 10 cm-30 cm， and over 30 cm. These ranges are enough for distinguishing our functions in a video player.
3.2 Adaptation Algorithm
Most studies on phone screen interaction are focused on particular equipment— there are no comparative studies of mobile devices with different screen sizes. Screen sizes and resolution are not necessarily related. Some devices have a small， high?resolution screen. For example， the HTC DNA smartphone has a 5?inch， 1920 × 1080 pixel screen， which is higher than some computers. The high resolution in a small screen means it has more pixel dots per inch （dpi）. Therefore， we need to analyze them.
To provide a good viewing experience over two screens， we design a novel sharing algorithm that involves two groups of parameters： one from the target screen and the other from the current screen. With Android， the real screen size cannot be read directly and must be calculated using the pixel and dpi values of screen. To facilitate adaptation， the pixels and dpi of the target screen’s width and height are sent to the current device. Current device use these received data and corresponding parameters of itself to finish adaptation.
Because the dpi of the width and height are different， they need to be calculated separately. [Wratio] is the ratio of the dpi of the width of the target screen T to that of the current screen C：
[Wratio=Twdpi/Cwdpi] （1）
[Hratio] is the ratio of the dpi of the height of the target screen to that of the current screen：
[Hratio=Thdpi/Chdpi] （2）
The width and height pixels shown of the current screen are estimated according to the ratios in （1） and （2）， respectively， and pixels from the target screen. The adaptation models are：
[Cwidth=Wratio×Twidth] （3）
[Cheight=Wratio×Theight] （4）
where [Cwidth] is the width of the current screen， and [Cheight] is the height of the current screen. The window of the video player is modified according to these new pixel values. In this way， the two parts of a video are the same size. 3.3 Synchronization Method
We synchronize the two devices by immediately sharing the current time of the video between the devices. In this way， the video frames are accurately coordinated across the two screens. The current time of the video is sent from one device to the other， which adjusts itself to the same time. Although the data is transmitted very quickly， the timing of the video is slightly asynchronous， and delay in the video may be obvious. WeWatch pauses the video at both sides until the time is synchronized.
Synchronization is controlled by signals from the positioning module. We synchronize according to the three distance ranges mentioned in subsection 3.1. There are three video modes： combined， pause， and full screen. When the devices are within 10 cm of each other， each device plays half the frame of the video， and data is transmitted. When the devices are 10 cm-30 cm from each other， the video is paused and data is not transmitted， but the video is still split between the two screens. When the devices are more than 30 cm from each other， the video is shown in full?screen mode on both devices， and the two devices are disconnected from each other.
4 Performance Evaluation
We ran WeWatch on two tablets： Samsung GT?N5110 and Samsung SM?P601. The former had Android 4.1 and an 8?inch 1280 × 800 pixel WXGA screen. The latter had Android 4.3 and a 10.1?inch 2560 × 1600 pixel screen.
We had three primary expectations： real?time synchronous playback， screen size adaptation， and video controlled by distance. Because WeWatch need to open Wi?Fi and the screen should be light all the time， we need to consider the energy efficiency in the real world.
4.1 RSSI vs. Distance
Most works on Wi?Fi ranging through RSSI focuses on routing issues and are based on simulations [32]. The performance of peer?to?peer communication using real?world settings has not been adequately studied. Therefore， we measure the relationship between RSSI values and distances through experimentation.
We run our experiments on an office desk， a common place where people watch videos. Fig. 3 shows Wi?Fi RSSI for distances 0 cm-30 cm. We calculate the average RSSI from 300 values for each distance. Overall， the RSSI decreases as distance increases. The RSSI does not conform exactly to the theoretical values we had predicted， which decrease smoothly. There is an anomalous increase in the mean RSSI over 5 cm-10 cm. This anomaly occurs again over 25 cm-30 cm. This may be because the Wi?Fi signals are disturbed or shielded in such a closed environment. From our experiments， we find that 10 cm and 30 cm are two critical distances， and it is appropriate to use their RSSIs to divide distance. In this way， the video player can be controlled by distance. 4.2 Power Comparison
In mobile devices， battery power is very limited， and this is an important consideration when developing mobile applications. Before we determine how often the Wi?Fi signals should be collected， we compare the normal play function in WeWatch with that of the tablet’s own video player. When no data is being transmitted in WeWatch， the player lasts more than three hours longer than the Samsung player （Fig. 4a）. A possible reason for this is that WeWatch only has common functions， and we do not open any extra background programs. Then we evaluate the video?sharing function. A data connection consumes a lot of power， so the device last two hours less than before. We change the frequency of signal collection from 200 ms to 1 s and measure power consumption with an application that logs the battery level. Then， we analyze the data on a computer （Fig. 4b）. Each time， we close the other applications on the same tablet， run the application， and record the battery level every hour.
The energy consumption using each method is clear in our experiments. The tablet’s own video player consumes the most power. WeWatch consumes less power than the tablet’s own video player when the WeWatch update interval is 200 ms. When this interval is changed to 1 s， WeWatch can last more than half an hour. As this update interval is increases， WeWatch’s power consumption decreases. In terms of user experience， the difference between 200 ms and 1 s update intervals is not perceptible to the human eye. So we optimize WeWatch by collect the RSSI every second. We also conducted an experiment on power consumption where we turned off all applications and the screen was dark. The tablet was in standby mode for about one week. The result showed that the screen consumes the most energy.
4.3 User Study
Current applications can detect user habits and usage context [33]， [34]. This capability affects user experience and the user’s preferences for their device. The motivation of our study is to improve user viewing experience. We gave our system to 27 students to try. In general， predicting user experience is quite difficult because there are so many factors affecting the interaction of users with devices. We formulated a questionnaire （Fig. 5） that would help us evaluate the user experience of WeWatch. Fig. 6 shows the results of this questionnaire.
More than 50% of participants gave WeWatch a score of nine points or higher out of a possible ten points because they appreciated the WeWatch concept. Some of these participants commented that the app was novel and truly improved the viewing experience. All participants viewed our application favorably， i.e.， gave it a score of at least 6 points， but some commented that we still have a lot of work to do to improve the demonstration. The question many asked is how to eliminate the physical frame between the screens of the two devices. The frames of different devices are often different widths and colors， and this detracts from the dual?screen viewing experience. This problem will be addressed over the next few years. Concept mobile devices with no frame have already been showcased. For example， SHARP released a new frameless screen at CES 2014 [35]， and we believe the no?frame products will be mainstream soon. Participant in our trial also suggested using distance control to download and transmit videos across devices and introducing greater flexibility in the direction that devices can be moved from each other in order to use these functions.
5 Conclusion
In this paper， we have presented an Android application called WeWatch that combines two to improve the video viewing experience. The basis of this idea is loT， i.e.， making devices connect to one another and become an Internet. WeWatch users can place two devices close to each other to create a bigger screen over which synchronized videos can be played. Functions such as play， pause and stop are controlled by moving the devices towards or away from each other. The app architecture comprises positioning module， adaptation algorithm， and synchronization module. We installed WeWatch on two tablets and tested it. To properly control the video player， we used a distance?estimation model based on realistic data and analyzed the appropriate distance ranges to be set. We experimented with WeWatch to determine how much energy it consumed， and we compared this with another video player application. We then compared the battery life with different update intervals in WeWatch to optimize energy efficiency.
We studied dual?screen viewing across two devices because this is the most basic case. WeWatch only focuses on left and right adaptation. In the short?term， we plan to extend this so that playback can be controlled through vertical movements of a device and enable videos to be matched from all four sides of the screen. In this way， we can extent WeWatch to more screens. In the long term， we intend to expand WeWatch so that is can support real?time video transmission across three or more devices.
References
[1] S. Kristoffersen and F. Ljungberg， “‘Making place’ to make IT work： empirical explorations of HCI for mobile CSCW，” in International ACM SIGGROUP Conference on Supporting Group Work， New York， USA， 1999， pp. 276-285. [2] R. Mohan， J. R. Smith， and C.?S. Li， “Adapting multimedia Internet content for universal access，” IEEE Transactions on Multimedia， vol. 1， no. 1， pp. 104-114， Mar. 1999. doi： 10.1109/6046.748175.
[3] J. Krumm and K. Hinckley， “The Nearme wireless proximity server，” in UbiComp ’04： Ubiquitous Computing， Nottingham， UK， Sept. 2004， pp. 283-300. doi： 10.1007/978?3?540?30119?6_17.
[4] V. M. Olivera， J. M. C. Plaza， and O. S. Serrano， “WiFi localization methods for autonomous robots，” Robotica， vol. 24， no. 4， pp. 455-461， Jul. 2006. doi： 10.1017/S0263574705002468.
[5] B. Roberts and K. Pahlavan， “Site?specific RSS signature modeling for WiFi localization，” in IEEE Global Telecommunications Conference， Hawaii， USA， Dec. 2009， pp. 1-6. doi： 10.1109/GLOCOM.2009.5425740.
[6] WeWatch Demo [Online]. Available： v.youku.com/v_show/id_XNzQwNjAyNzcy.html
[7] E. Oliver， “A survey of platforms for mobile networks research，” ACM SIGMOBILE Mobile Computing and Communications Review， vol. 12， no. 4， pp. 56-63， 2008. doi： 10.1145/1508285.1508292.
[8] M Butler， “Android： changing the mobile landscape，” IEEE Pervasive Computing， vol. 10， no. 1， pp.4-7， 2011. doi： 10.1109/MPRV.2011.1.
[9] ITU. （2005， Nov. 17）. ITU internet reports 2005： the internet of things [Online]. Available： www.itu.int/osg/spu/publications/internetofthings
[10] L. Atzori， A. Iera， and G. Morabito， “The internet of things： a survey，” Computer Networks， vol. 54， no. 15， pp. 2787-2805， Oct. 2010.
[11] L. Tan and N. Wang， “Future internet： the internet of things，” in 3rd International Conference on Advanced Computer Theory and Engineering （ICACTE）， Chengdu， China， Jun. 2010， pp. 376?380.
[12] L. Zeng， “The internet of things technological application in the region of fire control，” Computer Programming Skills & Maintenance， no. 16， pp. 118-119， Aug. 2012.
[13] L. Cai， “Intelligent logistics in the era of the internet of things，” Logistics Sci?Tech， vol. 33， no. 12， 2010.
[14] Q. Sun， J. Liu， S. Li， et al.， “Internet of things： summarize on concepts， architecture and key technology problem，” Journal of Beijing University of Posts and Telecommunications， vol. 33， no. 3， pp. 1-9， 2010. doi： 10.13190/jbupt.201003.1.038.
[15] B. Zhang， X. Yang， J. Li， et al.， “Overview on internet of things，” in 17th Annual Conference of China Information Theory Society， Xi’an， China， 2010.
[16] C. K. Toh， Ad Hoc Mobile Wireless Networks： Protocols and Systems. Upper Saddle River， USA： Prentice Hall， 2001. [17] C. de M. Cordeiro and D. P. Agrawal， Ad Hoc and Sensor Networks： Theory and Applications， 2nd ed. Singapore： World Scientific Publishing Company， 2011.
[18] S. R. Subramanya and B. K. Yi， “Enhancing the user experience in mobile phones，” Computer， vol. 40， no. 12， pp. 114-117， Dec. 2007. doi：10.1109/MC.2007.420.
[19] H. Wigelius and H. V??t?j?， “Dimensions of context affecting user experience in mobile work，” 12th IFIP TC 13 International Conference， Uppsala， Sweden， Aug. 2009， pp. 604-617. doi： 10.1007/978?3?642?03658?3_65.
[20] S. Bendel and D. Schuster， “WatchMyPhone—providing developer support for shared user interface objects in collaborative mobile applications，” in IEEE International Conference on Pervasive Computing and Communications Workshops， Lugano， Switzerland， Mar. 2012， pp. 166-171. doi： 10.1109/PerComW.2012.6197471.
[21] G. Zhai， G. C. Fox， M. Pierce， et al.， “eSports： collaborative and synchronous video annotation system in grid computing environment，” in Seventh IEEE International Symposium on Multimedia， Irvine， USA， Dec. 2005， pp. 95-103. doi： 10.1109/ISM.2005.55.
[22] R. Want， T. Pering， S. Sud， and B. Rosario， “Dynamic composable computing，” 9th Workshop on Mobile Computing Systems and Applications， Napa Valley， USA， Feb. 2008， pp. 17-21. doi： 10.1145/1411759.1411765.
[23] B. Xing， K. Seada， and N. Venkatasubramanian， “Proximiter： enabling mobile proximity?based content sharing on portable devices，” in IEEE International Conference on Pervasive Computing and Communications， Galveston， USA， Mar. 2009， pp. 1-3. doi： 10.1109/PERCOM.2009.4912802.
[24] S. Liu， Y. Jiang， and A. Striegel， “Face?to?face proximity estimation using bluetooth on smartphones，” IEEE Transactions on Mobile Computing， vol. 13， no. 4， pp. 811-823， Apr. 2014. doi： 10.1109/TMC.2013.44.
[25] G. Shen， Y. Li， and Y. Zhang， “Mobius： enable together?viewing video experience across two mobile devices，” in 5th international conference on Mobile Systems， Applications and Services， San Juan， Puerto Rico， Jun. 2007， pp. 30-42. doi： 10.1145/1247660.1247667.
[26] R. Mohan， J. R. Smith， and C.?S. Li， “Adapting multimedia Internet content for universal access”， IEEE Transactions on Multimedia， vol. 1， no. 1， pp. 104-114， Mar. 1999. doi： 10.1109/6046.748175.
[27] H. Liu， X. Xie， W.?Y. Ma， and H.?J. Zhang， “Automatic browsing of large pictures on mobile devices，” in 11th ACM International Conference on Multimedia， Berkeley， USA， Nov. 2003. pp. 148-155. doi： 10.1145/957013.957045. [28] T. Camp， J. Boleng， and V. Davies， “A survey of mobility models for ad hoc network research，” Wireless Communications and Mobile Computing， vol. 2， no. 5， pp. 483-502， Aug. 2002. doi： 10.1002/wcm.72.
[29] S. Basagni， M. Conti， S. Giordano， and I. Stojmenovic， Mobile Ad Hoc Networking. Hoboken， USA： Wiley?IEEE Press， 2004.
[30] B. Xing， K. Seada， and N. Venkatasubramanian， “An experimental study on Wi?Fi ad?hoc mode for mobile device?to?device video delivery，” in IEEE INFOCOM Workshops 2009， Rio de Janeiro， Brazil， Apr. 2009， pp. 1 -6. doi： 10.1109/INFCOMW.2009.5072111.
[31] J. Figueiras and S. Frattasi， Mobile Positioning and Tracking. Hoboken， USA： Wiley， 2010.
[32] C. Sweet， V. Devarapalli， and D. Sidhu. （2004）. IEEE 802.11 performance in an ad?hoc environment [Online]. Available： www.csee.umbc.edu/～sweet/papers/80211adhoc.pdf
[33] R. L. Wakefield and D. Whitten， “Mobile computing： a user study on hedonic/utilitarian mobile device usage，” European Journal of Information Systems， vol. 15， pp. 292-300， Jun. 2006. doi： 10.1057/palgrave.ejis.3000619.
[34] S. Consolvo， L. Arnstein， and B. R. Franza， “User study techniques in the design and evaluation of a ubicomp environment，” 4th International Conference on Ubiquitous Computing， G?teborg， Sweden， Sept. 2002， pp. 73-90. doi： 10.1007/3?540?45809?3_6.
[35] CES. （2014）. 2014 international CES [Online]. Available： http：//www.cesweb.org/
Manuscript received： 2015?03?05
In recent years， high?resolution video has developed rapidly and widescreen smart devices have become popular. We present an Android application called WeWatch that enables high?resolution video to be shared across two mobile devices when they are close to each other. This concept has its inspiration in machine?to?machine connections in the Internet of Things （loT）. We ensure that the two parts of the video are the same size over both screens and are synchronous. Further， a user can play， pause， or stop the video by moving one device a certain distance from the other. We decide on appropriate distances through experimentation. We implemented WeWatch on Android operating system and then optimize Watch so battery consumption is reduced. The user experience provided by WeWatch was evaluated by students through a questionnaire， and the reviews indicated that WeWatch does improve the viewing experience.
Keywords
together watching experience； screen adaptation； internet of things； distance estimation； energy efficiency
W1 Introduction
ith the rapid development of the mobile industry and growth in the number of high?resolution videos， screen size has become an important factor in product design. Widescreen mobile devices have become more and more popular， but this has led to many problems； for example， a large screen makes one?handed operation more difficult [1]. Nowadays， researchers are trying to improve the viewing experience on smart phones and tablets. The following example highlights the goals we are trying to achieve：
Henry has an old smart phone with a small， low?resolution screen， so he buys a smart phone with a bigger screen. He is watching a movie when：
·He puts the old phone and new phone side by side to get a bigger screen， and the movie plays across the two screens. In this way， he can enjoy the movie on a larger overall screen.
·Someone knocks at the door. He moves one phone away from the other and the video pauses. When he comes back， he only needs to put the two phones together again to pick up where the movie left off.
In this paper we propose WeWatch， an application for viewing videos across two devices with different screen sizes and resolutions （Fig. 1）.
WeWatch provides a better viewing experience than a video player on a single device. It encourages people to use idle devices and share videos. WeWatch has two unique features： screen size adaptation and the use of distance to control video playback. Previous studies on video adaptation mainly focus on optimally matching the resources and capabilities of diverse client devices [2]. However， in order to aggregate resources， these client devices need collaboration and synchronization. This is difficult because different devices have different screen sizes and resolutions， and it is also difficult to transform resources so that they match different devices. To solve this problem， we propose an adaptation algorithm that only matches two devices instead of matching both resources and devices. Our adaptation algorithm can match the size of video windows on both devices so that the video only needs be divided into two parts of the same size. When WeWatch is launched on one device， the device transmits its screen size and resolution to the other device， which automatically adjusts itself.
We have devised a special way of using our application： changing the distance between devices. With WeWatch， a video can be paused， played or stopped by moving the devices certain distances from each other. The received signal strength indicator （RSSI） of Wi?Fi is used to measure these distances [3]. Wi?Fi signals are usually used for location?based services [4]， [5] but are rarely used for managing applications.
A demo of WeWatch is available online [6]. WeWatch works on two smart devices running Android [7]， [8]. We choose this platform because it is popular and open?source.
We have also tested the power consumption of WeWatch and found that it consumes about 40% less power than a normal video player， both in single?screen mode and dual?screen mode. Different parameters have obvious effects on battery consumption. To find the empirical Wi?Fi RSSI values of distances， we determine the relationship between them through real?life simulations. The results help us improve the implementation of the application.
The contributions of this paper are：
·It describes one of the first attempts to combine screens for a better viewing experience. Our scheme does not require any special hardware； it can be implemented on existing mobile devices.
·It describes a new adaptation algorithm that swiftly matches the screen size and resolution of different mobile devices.
·It describes a positioning method， which is used for determining suitable distances for playback control， and a synchronization method for smoother viewing experience.
·It describes a prototype of the screen?sharing system that has been evaluated for energy efficiency and user experience. 2 Related Work
The Internet of Things （IoT） comprises interconnected， uniquely identifiable computing devices embedded within the existing Internet infrastructure. Communication will expand to human?to?thing and thing?to?thing communication. Combining devices is a basic concept in the loT. Much attention has been paid to the IoT in recent years [9]-[11]； however， IoT applications are largely limited to intellectual storage management， public security， and automation [12]， [13]. Not enough attention has been paid to using this new technology in daily life. WeWatch draws on the idea of the loT to combine two mobile devices and give people a better viewing experience.
The loT includes RFID technology， sensor network， detection technology， intelligence technology， and nanotechnology [14]， [15]. Machine?to?machine connections are an important part of the loT， and WeWatch combines two smart devices using Wi?Fi ad hoc technology [16]， [17].
There are many existing systems and toolkits that enhance user experience on mobile devices [18]， [19]. However， most of these are mainly designed for one?way connections， i.e.， resources connecting to devices. WeWatch provides a higher level of communication by grouping entities together， i.e.， devices connecting to other devices. WeWatch can be used directly between portable devices and does not require Internet access.
Some toolkits are used to develop collaborative mobile applications [20]， collaborative and synchronous video annotation platforms [21]， and even impromptu assemblies of a logical computer from the best set of wireless components available nearby [22]. We have the same starting point， i.e.， to provide convenience through collaborative devices.
A proximeter [23] enables mobile proximity?based content sharing on portable devices. This sharing occurs when consumers of content move within the proximity of the owner of the content. A proximiter also creates opportunities for users to chat with their neighbors through their phones. The proximeter prototype is based on Meamo v4.0， which is an operating system in its infancy. Meamo is only supported by Nokia， and there is a lack of applications for the operating system.
Shu Liu and Yingxin Jiang [24] focused on face?to?face proximity estimation using smartphones and Bluetooth. They explored the relationship between Bluetooth RSSI and distances through a series of experiments. They compared the accuracy of Bluetooth， Wi?Fi and GPS using real?world measurements. However， it is impractical to use Bluetooth to measure the distance between devices. MobiUS [25] has developed a video application for displaying a high?resolution video across two mobile devices. The application is based on an efficient， collaborative half?frame decoding scheme. Because the collaborative system architecture is tightly coupled， the application only works on devices with the same screen size and resolution.
Content adaptation for mobile devices is usually directed towards online multimedia [26] or pictures [27]. With content adaptation， content of different sizes and resolutions can be adapted to different screens. However， researchers have rarely considered the implications of combining screens.
3 System Design
We develop WeWatch from both the client side and server side. Our application can decide the role of each device automatically when launched. Fig. 2 shows the WeWatch architecture.
The client and server are linked by Wi?Fi ad hoc networking technologies， which enable direct device?to?device communication. This helps reduce the traffic load in the wireless infrastructure and improve user experience. The role of devices can be easily decided because the hotspot and Wi?Fi cannot be turned on at the same time [28]， [29]. Some researchers have examined the trade?off associated with using Wi?Fi ad hoc and infrastructure modes. Their experiments show that mobile devices are stable when communicating directly with each other [30]. With our WeWatch， a user can choose a device to be a hotspot， with other connecting to that hotspot. We separate the ad?hoc function from our application so that a user can exploit it to make new functions.
The client and server sides can send requests to share videos， and these requests are executed after being authenticated by the valid distance from empirical RSSI values.
WeWatch comprises three modules： positioning， adaptation， and synchronization. The positioning module at the client side records the RSSI values and uses these to estimate the distances. Then， it changes the video mode and notices other modules. The adaptation module sends the screen size and resolution between the two devices so that the display across the two screens is the same size. The synchronization module ensures both screens are playing the same video frames at the same time.
3.1 Positioning Method
The positioning method used in WeWatch frees our hands from having to click buttons. Specific functions are initiated by dragging one device away from the other， changing the distance between them. Different distances are estimated according to RSSI values. For example， the video is played when the distance between devices is within 10 cm， the video and the video is paused when the distance is more than 10 cm. To quit dual?viewing mode， one of the devices device has to be moved more than 30 cm away. All these operations are reversible. Device?to?device Wi?Fi ad hoc communication enables the client side to be directly connected to the server side. The client side records the signal values sent from the server side. These values are reliable enough to estimate the distance between them. Wi?Fi triangulation is not accurate enough to detect exact position within a very close range [31]； however， we can use it to detect approximate position within a range， e.g.， within 10 cm， 10 cm-30 cm， and over 30 cm. These ranges are enough for distinguishing our functions in a video player.
3.2 Adaptation Algorithm
Most studies on phone screen interaction are focused on particular equipment— there are no comparative studies of mobile devices with different screen sizes. Screen sizes and resolution are not necessarily related. Some devices have a small， high?resolution screen. For example， the HTC DNA smartphone has a 5?inch， 1920 × 1080 pixel screen， which is higher than some computers. The high resolution in a small screen means it has more pixel dots per inch （dpi）. Therefore， we need to analyze them.
To provide a good viewing experience over two screens， we design a novel sharing algorithm that involves two groups of parameters： one from the target screen and the other from the current screen. With Android， the real screen size cannot be read directly and must be calculated using the pixel and dpi values of screen. To facilitate adaptation， the pixels and dpi of the target screen’s width and height are sent to the current device. Current device use these received data and corresponding parameters of itself to finish adaptation.
Because the dpi of the width and height are different， they need to be calculated separately. [Wratio] is the ratio of the dpi of the width of the target screen T to that of the current screen C：
[Wratio=Twdpi/Cwdpi] （1）
[Hratio] is the ratio of the dpi of the height of the target screen to that of the current screen：
[Hratio=Thdpi/Chdpi] （2）
The width and height pixels shown of the current screen are estimated according to the ratios in （1） and （2）， respectively， and pixels from the target screen. The adaptation models are：
[Cwidth=Wratio×Twidth] （3）
[Cheight=Wratio×Theight] （4）
where [Cwidth] is the width of the current screen， and [Cheight] is the height of the current screen. The window of the video player is modified according to these new pixel values. In this way， the two parts of a video are the same size. 3.3 Synchronization Method
We synchronize the two devices by immediately sharing the current time of the video between the devices. In this way， the video frames are accurately coordinated across the two screens. The current time of the video is sent from one device to the other， which adjusts itself to the same time. Although the data is transmitted very quickly， the timing of the video is slightly asynchronous， and delay in the video may be obvious. WeWatch pauses the video at both sides until the time is synchronized.
Synchronization is controlled by signals from the positioning module. We synchronize according to the three distance ranges mentioned in subsection 3.1. There are three video modes： combined， pause， and full screen. When the devices are within 10 cm of each other， each device plays half the frame of the video， and data is transmitted. When the devices are 10 cm-30 cm from each other， the video is paused and data is not transmitted， but the video is still split between the two screens. When the devices are more than 30 cm from each other， the video is shown in full?screen mode on both devices， and the two devices are disconnected from each other.
4 Performance Evaluation
We ran WeWatch on two tablets： Samsung GT?N5110 and Samsung SM?P601. The former had Android 4.1 and an 8?inch 1280 × 800 pixel WXGA screen. The latter had Android 4.3 and a 10.1?inch 2560 × 1600 pixel screen.
We had three primary expectations： real?time synchronous playback， screen size adaptation， and video controlled by distance. Because WeWatch need to open Wi?Fi and the screen should be light all the time， we need to consider the energy efficiency in the real world.
4.1 RSSI vs. Distance
Most works on Wi?Fi ranging through RSSI focuses on routing issues and are based on simulations [32]. The performance of peer?to?peer communication using real?world settings has not been adequately studied. Therefore， we measure the relationship between RSSI values and distances through experimentation.
We run our experiments on an office desk， a common place where people watch videos. Fig. 3 shows Wi?Fi RSSI for distances 0 cm-30 cm. We calculate the average RSSI from 300 values for each distance. Overall， the RSSI decreases as distance increases. The RSSI does not conform exactly to the theoretical values we had predicted， which decrease smoothly. There is an anomalous increase in the mean RSSI over 5 cm-10 cm. This anomaly occurs again over 25 cm-30 cm. This may be because the Wi?Fi signals are disturbed or shielded in such a closed environment. From our experiments， we find that 10 cm and 30 cm are two critical distances， and it is appropriate to use their RSSIs to divide distance. In this way， the video player can be controlled by distance. 4.2 Power Comparison
In mobile devices， battery power is very limited， and this is an important consideration when developing mobile applications. Before we determine how often the Wi?Fi signals should be collected， we compare the normal play function in WeWatch with that of the tablet’s own video player. When no data is being transmitted in WeWatch， the player lasts more than three hours longer than the Samsung player （Fig. 4a）. A possible reason for this is that WeWatch only has common functions， and we do not open any extra background programs. Then we evaluate the video?sharing function. A data connection consumes a lot of power， so the device last two hours less than before. We change the frequency of signal collection from 200 ms to 1 s and measure power consumption with an application that logs the battery level. Then， we analyze the data on a computer （Fig. 4b）. Each time， we close the other applications on the same tablet， run the application， and record the battery level every hour.
The energy consumption using each method is clear in our experiments. The tablet’s own video player consumes the most power. WeWatch consumes less power than the tablet’s own video player when the WeWatch update interval is 200 ms. When this interval is changed to 1 s， WeWatch can last more than half an hour. As this update interval is increases， WeWatch’s power consumption decreases. In terms of user experience， the difference between 200 ms and 1 s update intervals is not perceptible to the human eye. So we optimize WeWatch by collect the RSSI every second. We also conducted an experiment on power consumption where we turned off all applications and the screen was dark. The tablet was in standby mode for about one week. The result showed that the screen consumes the most energy.
4.3 User Study
Current applications can detect user habits and usage context [33]， [34]. This capability affects user experience and the user’s preferences for their device. The motivation of our study is to improve user viewing experience. We gave our system to 27 students to try. In general， predicting user experience is quite difficult because there are so many factors affecting the interaction of users with devices. We formulated a questionnaire （Fig. 5） that would help us evaluate the user experience of WeWatch. Fig. 6 shows the results of this questionnaire.
More than 50% of participants gave WeWatch a score of nine points or higher out of a possible ten points because they appreciated the WeWatch concept. Some of these participants commented that the app was novel and truly improved the viewing experience. All participants viewed our application favorably， i.e.， gave it a score of at least 6 points， but some commented that we still have a lot of work to do to improve the demonstration. The question many asked is how to eliminate the physical frame between the screens of the two devices. The frames of different devices are often different widths and colors， and this detracts from the dual?screen viewing experience. This problem will be addressed over the next few years. Concept mobile devices with no frame have already been showcased. For example， SHARP released a new frameless screen at CES 2014 [35]， and we believe the no?frame products will be mainstream soon. Participant in our trial also suggested using distance control to download and transmit videos across devices and introducing greater flexibility in the direction that devices can be moved from each other in order to use these functions.
5 Conclusion
In this paper， we have presented an Android application called WeWatch that combines two to improve the video viewing experience. The basis of this idea is loT， i.e.， making devices connect to one another and become an Internet. WeWatch users can place two devices close to each other to create a bigger screen over which synchronized videos can be played. Functions such as play， pause and stop are controlled by moving the devices towards or away from each other. The app architecture comprises positioning module， adaptation algorithm， and synchronization module. We installed WeWatch on two tablets and tested it. To properly control the video player， we used a distance?estimation model based on realistic data and analyzed the appropriate distance ranges to be set. We experimented with WeWatch to determine how much energy it consumed， and we compared this with another video player application. We then compared the battery life with different update intervals in WeWatch to optimize energy efficiency.
We studied dual?screen viewing across two devices because this is the most basic case. WeWatch only focuses on left and right adaptation. In the short?term， we plan to extend this so that playback can be controlled through vertical movements of a device and enable videos to be matched from all four sides of the screen. In this way， we can extent WeWatch to more screens. In the long term， we intend to expand WeWatch so that is can support real?time video transmission across three or more devices.
References
[1] S. Kristoffersen and F. Ljungberg， “‘Making place’ to make IT work： empirical explorations of HCI for mobile CSCW，” in International ACM SIGGROUP Conference on Supporting Group Work， New York， USA， 1999， pp. 276-285. [2] R. Mohan， J. R. Smith， and C.?S. Li， “Adapting multimedia Internet content for universal access，” IEEE Transactions on Multimedia， vol. 1， no. 1， pp. 104-114， Mar. 1999. doi： 10.1109/6046.748175.
[3] J. Krumm and K. Hinckley， “The Nearme wireless proximity server，” in UbiComp ’04： Ubiquitous Computing， Nottingham， UK， Sept. 2004， pp. 283-300. doi： 10.1007/978?3?540?30119?6_17.
[4] V. M. Olivera， J. M. C. Plaza， and O. S. Serrano， “WiFi localization methods for autonomous robots，” Robotica， vol. 24， no. 4， pp. 455-461， Jul. 2006. doi： 10.1017/S0263574705002468.
[5] B. Roberts and K. Pahlavan， “Site?specific RSS signature modeling for WiFi localization，” in IEEE Global Telecommunications Conference， Hawaii， USA， Dec. 2009， pp. 1-6. doi： 10.1109/GLOCOM.2009.5425740.
[6] WeWatch Demo [Online]. Available： v.youku.com/v_show/id_XNzQwNjAyNzcy.html
[7] E. Oliver， “A survey of platforms for mobile networks research，” ACM SIGMOBILE Mobile Computing and Communications Review， vol. 12， no. 4， pp. 56-63， 2008. doi： 10.1145/1508285.1508292.
[8] M Butler， “Android： changing the mobile landscape，” IEEE Pervasive Computing， vol. 10， no. 1， pp.4-7， 2011. doi： 10.1109/MPRV.2011.1.
[9] ITU. （2005， Nov. 17）. ITU internet reports 2005： the internet of things [Online]. Available： www.itu.int/osg/spu/publications/internetofthings
[10] L. Atzori， A. Iera， and G. Morabito， “The internet of things： a survey，” Computer Networks， vol. 54， no. 15， pp. 2787-2805， Oct. 2010.
[11] L. Tan and N. Wang， “Future internet： the internet of things，” in 3rd International Conference on Advanced Computer Theory and Engineering （ICACTE）， Chengdu， China， Jun. 2010， pp. 376?380.
[12] L. Zeng， “The internet of things technological application in the region of fire control，” Computer Programming Skills & Maintenance， no. 16， pp. 118-119， Aug. 2012.
[13] L. Cai， “Intelligent logistics in the era of the internet of things，” Logistics Sci?Tech， vol. 33， no. 12， 2010.
[14] Q. Sun， J. Liu， S. Li， et al.， “Internet of things： summarize on concepts， architecture and key technology problem，” Journal of Beijing University of Posts and Telecommunications， vol. 33， no. 3， pp. 1-9， 2010. doi： 10.13190/jbupt.201003.1.038.
[15] B. Zhang， X. Yang， J. Li， et al.， “Overview on internet of things，” in 17th Annual Conference of China Information Theory Society， Xi’an， China， 2010.
[16] C. K. Toh， Ad Hoc Mobile Wireless Networks： Protocols and Systems. Upper Saddle River， USA： Prentice Hall， 2001. [17] C. de M. Cordeiro and D. P. Agrawal， Ad Hoc and Sensor Networks： Theory and Applications， 2nd ed. Singapore： World Scientific Publishing Company， 2011.
[18] S. R. Subramanya and B. K. Yi， “Enhancing the user experience in mobile phones，” Computer， vol. 40， no. 12， pp. 114-117， Dec. 2007. doi：10.1109/MC.2007.420.
[19] H. Wigelius and H. V??t?j?， “Dimensions of context affecting user experience in mobile work，” 12th IFIP TC 13 International Conference， Uppsala， Sweden， Aug. 2009， pp. 604-617. doi： 10.1007/978?3?642?03658?3_65.
[20] S. Bendel and D. Schuster， “WatchMyPhone—providing developer support for shared user interface objects in collaborative mobile applications，” in IEEE International Conference on Pervasive Computing and Communications Workshops， Lugano， Switzerland， Mar. 2012， pp. 166-171. doi： 10.1109/PerComW.2012.6197471.
[21] G. Zhai， G. C. Fox， M. Pierce， et al.， “eSports： collaborative and synchronous video annotation system in grid computing environment，” in Seventh IEEE International Symposium on Multimedia， Irvine， USA， Dec. 2005， pp. 95-103. doi： 10.1109/ISM.2005.55.
[22] R. Want， T. Pering， S. Sud， and B. Rosario， “Dynamic composable computing，” 9th Workshop on Mobile Computing Systems and Applications， Napa Valley， USA， Feb. 2008， pp. 17-21. doi： 10.1145/1411759.1411765.
[23] B. Xing， K. Seada， and N. Venkatasubramanian， “Proximiter： enabling mobile proximity?based content sharing on portable devices，” in IEEE International Conference on Pervasive Computing and Communications， Galveston， USA， Mar. 2009， pp. 1-3. doi： 10.1109/PERCOM.2009.4912802.
[24] S. Liu， Y. Jiang， and A. Striegel， “Face?to?face proximity estimation using bluetooth on smartphones，” IEEE Transactions on Mobile Computing， vol. 13， no. 4， pp. 811-823， Apr. 2014. doi： 10.1109/TMC.2013.44.
[25] G. Shen， Y. Li， and Y. Zhang， “Mobius： enable together?viewing video experience across two mobile devices，” in 5th international conference on Mobile Systems， Applications and Services， San Juan， Puerto Rico， Jun. 2007， pp. 30-42. doi： 10.1145/1247660.1247667.
[26] R. Mohan， J. R. Smith， and C.?S. Li， “Adapting multimedia Internet content for universal access”， IEEE Transactions on Multimedia， vol. 1， no. 1， pp. 104-114， Mar. 1999. doi： 10.1109/6046.748175.
[27] H. Liu， X. Xie， W.?Y. Ma， and H.?J. Zhang， “Automatic browsing of large pictures on mobile devices，” in 11th ACM International Conference on Multimedia， Berkeley， USA， Nov. 2003. pp. 148-155. doi： 10.1145/957013.957045. [28] T. Camp， J. Boleng， and V. Davies， “A survey of mobility models for ad hoc network research，” Wireless Communications and Mobile Computing， vol. 2， no. 5， pp. 483-502， Aug. 2002. doi： 10.1002/wcm.72.
[29] S. Basagni， M. Conti， S. Giordano， and I. Stojmenovic， Mobile Ad Hoc Networking. Hoboken， USA： Wiley?IEEE Press， 2004.
[30] B. Xing， K. Seada， and N. Venkatasubramanian， “An experimental study on Wi?Fi ad?hoc mode for mobile device?to?device video delivery，” in IEEE INFOCOM Workshops 2009， Rio de Janeiro， Brazil， Apr. 2009， pp. 1 -6. doi： 10.1109/INFCOMW.2009.5072111.
[31] J. Figueiras and S. Frattasi， Mobile Positioning and Tracking. Hoboken， USA： Wiley， 2010.
[32] C. Sweet， V. Devarapalli， and D. Sidhu. （2004）. IEEE 802.11 performance in an ad?hoc environment [Online]. Available： www.csee.umbc.edu/～sweet/papers/80211adhoc.pdf
[33] R. L. Wakefield and D. Whitten， “Mobile computing： a user study on hedonic/utilitarian mobile device usage，” European Journal of Information Systems， vol. 15， pp. 292-300， Jun. 2006. doi： 10.1057/palgrave.ejis.3000619.
[34] S. Consolvo， L. Arnstein， and B. R. Franza， “User study techniques in the design and evaluation of a ubicomp environment，” 4th International Conference on Ubiquitous Computing， G?teborg， Sweden， Sept. 2002， pp. 73-90. doi： 10.1007/3?540?45809?3_6.
[35] CES. （2014）. 2014 international CES [Online]. Available： http：//www.cesweb.org/
Manuscript received： 2015?03?05