FolkMusic: A Mobile Peer-to-Peer Entertainment System

Mikael Wiberg, PhD Interaction Theory Lab, Department of Informatics

Umeå University, 901 87 Umeå, Sweden mwiberg@informatik.umu.se

Abstract

In this paper we present the design of FolkMusic, a mobile peer-to-peer entertainment system. The work reported in this paper builds on the current trends towards: 1) edutainment software, 2) increase in use of peer-to-peer technologies, and, 3) the current trend towards mobile computing solutions. Further on, the research reported in this paper builds on prior research on “folk computing” for which mobile ad-hoc peer-to-peer solutions are a focal concern. Overall, this paper contributes to research on MANETs (Mobile Ad-hoc Networks) and peer-to-peer (P2P) computing by illustrating how mobile peer-to-peer technologies can be applied to both the area of entertainment as well as for educational purposes. Further on, the FolkMusic system contributes to current technical efforts made on ad-hoc networks by providing a good example of how Internet technologies and similar stationary and stable infrastructures can be combined with highly dynamic and wireless networks. 1. Introduction

During the 80s it was very popular to bring a HiFi stereo along while walking down the streets downtown in the cities. The music could be heard in the streets and it was shared within a community, the citizens and their visitors. However, the loud stereos were also quite annoying. When the Walkman was introduced in the beginning of the 90s the music was still on the streets. However, although the Walkman solved the problem of annoying music in the streets it also meant another change. The persons became isolated from each other when it came to music sharing when everybody listens to their own music in earphones.

In the end of the 90s Napster and similar peer-to-peer (P2P) technologies were developed to support sharing of music files (MP3 files) within a community. Today, almost any song can be found on these networks. However, even though Kazaa [7], Gnutella, LimeWire, and similar services offers the opportunity for people to share music it also meant that the music was no longer shared in the streets, and the community members were still isolated, both in terms of what the other community members prefer to listen to, but also geographically. Still, different cultures listens to different songs and music are often associated with geographical places. Bronx for instance is often associated with hip-hop music, whereas the city of Wien might be more associated with classical tunes. However, while walking down the streets in a city that cannot be experienced. We are isolated and it is only our assumptions that tell us what kind of music that is popular in a certain place.

In this paper we have explored how to escape this

isolation and bring the music back on the streets again, and enable children to explore and experience the audio community of different places without annoying its surrounding. The paper builds on the idea of folk computing [3] and contributes to that strand of research by focusing on audio communities rather than text based interaction. The fundamental idea behind folk computing and its basic motivation has been described by Borovoy [3] as follows:

"Folk Computing" is a new way of using

technology to support face-to-face communication and community, modeled on the communicative process of folklore. Although there is an abundance of research on new technologies to support community, very little of it focuses on co-present communities whose dominant means of interacting is face-to-face [3].

Further on, this project has been inspired by the description made by Reingold [12] of how modern and mobile information technology can enhance human interaction and make us become “Smart mobs”. As formulated by Reingold [12]:

“Smart mobs consist of people who are able to act

in concert even if they don’t know each other. The people that make up smart mobs cooperate in ways never before possible because they carry devices that process both communication and computing capabilities.” [12, p. xii].

From a pedagogical perspective our intention was to

design a tool to enable children to, in a “learning-by-doing” fashion, explore the fundamentals of set theory (including elements, sets, unions, and intersections of different sets) and how that can be applied for the purpose of filtering large information sets (e.g. large audio archives such as the online MP3-file sharing system Kazaa [7] that contains several thousands music files), at the same time as they entertain themselves with music. Although other file types such as documents could have been chosen for this particular project we choose music since a lot of research efforts right now is focused on how to design good edutainment software, i.e. software that is both good educational means, but also fun to use.

The paper presents an architecture that supports

mobile and ad hoc exploration and experience of the music that is ”in the air” of a community, how that music can be shared, and how people can follow audio traces to find different kinds of music.

The rest of the paper is structured as follows: in the

next section we present the underlying model for this project followed by a description of the FolkMusic system. We then report on some initial usage observations, present related work, outline future work and discuss progress made so far in the project before concluding the paper.

2. Underlying model: Set theory based services

In this project we have explored how interactive

audio services can be designed based on the fundamental building blocks of set theory. Basically, and according to the concepts of set theory we have chosen to treat a user and his/her near close vicinity (i.e. around 400 meters outdoors in open terrain) as one

set which then contains elements (i.e. MP3-files) carried around on some kind of mobile device (e.g. PDA or Smartphone). When two persons then meet a union of the two sets is made. Further on, an intersection between these two sets indicates which MP3-files they have in common (i.e. if the file exists on two both devices). However, beyond the concept of proximity triggered services (in the Bluetooth standard handled in the SDP – Service Detection Protocol), which is basically the union of two or several sets, set theory allows us to explore complementary sets, chains of complementary sets, unions, intersections, and elements.

Further on, a set of elements can belong to a face,

i.e. a community member or, in fact, a person’s mobile device, or a place, i.e. a geographic place. That means that a place can be loaded with music files at the same time as the persons who are present at the place might carry around, and drop to each other, or to the specific place different tunes or pieces of music. This also means that the music elements can be geographically positioned (i.e. belonging to a set defined by its geographical location), or relative positioned (i.e. belong to a person that is moving around). All these combinations leave traces of music in the air for others to experience.

The concept of traces as a way of distinguishing

between several interaction modalities was first introduced by Sorensen [13, p. 125]. According to him, interaction, and if it leaves traces can be categorized as follows:

Figure 1. Four different interaction modalities. In Sorensen [13, p. 125].

According to the figure 1 the audio traces of the

faces and places left are persistent so that other people can experience them, and unobtrusive as to avoid the traces to be as annoying as loud stere os in the streets.

3. Peer-to-peer goes mobile

Kazaa [7] and similar services uses peer-to-peer

technology for sharing music (MP3 files). Currently (i.e. 2003-09-10) there are 3,121,237 90 users online, sharing 631,138,762 files (5,128,704 GB) available solely within the Kazaa [7] P2P network community. Each computer in the network is a peer that has a predefined public folder containing music that can be shared with others. See figure 2:

Figure 2. The Kazaa client with a set of shared MP3 music files.

As seen in figure 2 a single search on “Madonna”

generated a long list of different songs played by her. These files can then be shared to others by simply downloading each song over the Internet. Every file that is downloaded is stored in a public folder as the one in figure 3 below. In this way each downloaded song will be possible for others to download from yet another node in the peer-to-peer network.

However, the problem of isolation as highlighted

above still remains. Even though each user has access to almost any music file in the world it is hard to come up with ideas about what to download. We are isolated with the music we know so well that we can make an explicit search for the artists name or the title of the song.

In the modern society of today we hear music

around us everyday, all the time on the radio, in city malls, on the TV, etc. However, there are no easy links between the music played and the files on systems like Kazaa. The user needs to find out the name and title of that piece of music themselves. Further, to be able to

experience the music the MP3 file needs to be downloaded first.

Figure 3. The public folder containing MP3-music files that can be shared to other Napster or Kazaa users.

In this project we are exploring the use of persons

and their geographical locations together with geographic places as peers in mobile ad hoc audio networks. In combination with the set theory based services outlined before this means that a user can walk down the streets in a city and experience the music that is played in the surrounding.

4. FolkMusic: Experiences audio traces of faces and places through mobile ad hoc networking

To explore how people can learn to filter large

information sets, and get to know communit ies with use of set theory, at the same time as they entertain themselves with music we have developed a prototype system called FolkMusic.

The basic idea underlying the FolkMusic system can

be view as follows: As a person (A) walks down the street in a city He/She will meet or pass by several persons on his/her way. For each face (or person) met their public music files is represented to the user in a playlist of ”folk tunes”. Similary, geographical places loaded with music files (i.e. music files associated with GPS-coordinates) will be displayed to the user in the playlist. See figure 4 for a graphical illustration of this scenario:

Figure 4. Basic idea underlying the FolkMusic system. Person A is passing through the crowd of people which constitutes a dynamic audio landscape of available sets of music to explore and listen to while walking around.

When a user selects a song from the playlist the

music is not downloaded as in the case of peer-to-peer clients such as the Kazaa client. Instead we use streaming MP3 so that the music starts instantly when the song is selected. In this way the user can easily and quickly brows through the files available to find out if there is anything interesting “in the air” without having to wait for the file to download.

5. FolkMusic hardware

In the current version the FolkMusic system it is

implemented on laptop computers equipped with WLAN (IEEE 802.11b) PC cards. The WLAN pc cards are used to determine which computers are within the close vicinity (i.e. within 400 meters), and thus should be included in the ad hoc network.

The laptops are also equipped with GPS receivers.

The GPS receivers are used to leave audio traces on geographic locations by associating a music file with a specific location. When a user then enters that location the MP3-files associated with that location is displayed to the user. If a geographic location is associated to a music file the GPS coordinates of that specific place is distributed to the other clients in the network. Each client checks 10 times/minute if the current location is associated with any music files.

6. The FolkMusic software client

FolkMusic is a PC based software client that is

integrated with Kazaa. The FolkMusic system shares the public file folder with Kazaa and Kazaa is also used to download new MP3-files to the system. However, Kazaa is only used when a user decides to save a music file on his/her own computer. To just experience the music currently “in the air” there is no need for Kazaa. Instead of first havin g to download each song over the

ad hoc network and then execute the file to listen to the music we have chosen to implement streaming MP3 over a TCP/IP socket so that just one click on a song that is currently stored on another device starts the playing of that song immediately.

The FolkMusic client was programmed in Delphi and

the user interface consists of a personal play list and a list of folk tunes displaying the MP3-files currently available given the specific location and the people that are currently on that location. The default setting in the FolkMusic client is to display every MP3-files in the union of co-located devices as in the case of figure 5 below. However, more complex questions can be stated. E.g. to only show the elements (i.e. MP3-files) that are shared with others than the user him/herself (i.e. the intersections between other co-located devices and their complementary sets which are not part of an intersection between the set that the user represents and its nearby nodes).

Figure 5. The FolkMusic software client.

Figure 5 illustrates the FolkMusic client. At the top

of the interface the user find buttons for starting and stopping the music files played. By clicking on the files in the personal playlist the user can choose to lis ten to his/her own music previously downloaded with Kazaa.

Further on, the user can also choose to play MP3s

from devices that are in the close vicinity or associated with the current location of the user by selecting files from the Folk tunes playlist. As described above we implemented this as streaming MP3 over a TCP/IP socket in the ad hoc network.

7. Selecting music by set theory based requests to the system

To enable the users to explore set theory while

experiencing music we developed a set theory based request interface to filter the surrounding audio files available. See fig. 6:

Figure 6. Set theory based request interface. Every computer in the MANET (Mobile Ad-hoc Network) is represented as a set in a dynamic VENN-diagram.

In figure 6 the user is represented as A and the

nearby faces and audio enhanced places (i.e. other nearby computers in the ad-hoc network) are represented as B, C, D, and E.

The user can alter the range of what is displayed (i.e.

how many levels of complementary sets) and, by selecting different sets, intersections, and complementary sets the user can filter the surrounding music. Based on the selections of the user the folk tunes (i.e. MP3 files) available changes (see figure 5). For each selection the request is made more specific and, by pressing the Reset button the user can start allover again with a new request to the system. In figure 6 a user has selected the intersection between his own music (A), (B) and (C) together with the intersection between his/her own complementary set Cs intersection with Cs complementary set D.

8. Initial observations from FolkMusic in use & questions for future research

We have explored the use of FolkMusic initially but a large-scale evaluation is planned to take place during the fall of 2003. Figure 7 below shows a user while using the FolkMusic system.

Figure 7. One of the users during the initial evaluation. The user has ear phones for the audio and holds a Libretto subnote PC computer in her right hand.

Figure 7 shows a user as she is holding a subnote

PC in her right hand almost like holding a ordinary Walkman while listening to streaming MP3 files from nearby mobile computers downtown in a city.

During the initial usage tests of FolkMusic several

questions where raised for further research. First, related to the “isolation” problem with Kazaa and similar P2P technologies as outlined in the introduction of this paper it will be interesting to see if it will be possible for users of the FolkMusic system to geographically navigate themselves to music they like by following the audio traces that is “in the air”. Second, we are going to further explore how users will go about making requests to the system compared to geographically move around, which is a filter in itself. Will they learn to see connections between the requests they make to the systems concerning e.g. complementary sets, unions, etc and the faces and places they pass by while walking around? A third question concerns awareness, and especially social awareness [1], i.e. what other people listen to might give guidance about what to listen to oneself. This might be especially interesting to teenagers which are sometimes described as persons who have a strong need to belong to a group, and be very alike the other persons in the group, at the same time as they want to be somewhat different or special in some sense, e.g.

have a new MP3-file from the currently most popular rock band to share with their friends.

Finally, will this kind of mobile IT support have any

implications for what to consider as hit music within a community? When hit music is presented on the radio it is a representation of what someone considers as hit music which is then broadcasted to the community. With this IT support it might be possible to recognize frequent flows of music exchanged between its community members as hit music. Clearly, how the users will go about exploring set theory to filter music will open up for a lot of questions related to how to track hit music, broadly shared music, and how to use set theory to find out how a community ranks music, etc. Probably even more interesting and novel ways of applying set theory to these distributed, temporal music archives will emerge when this system is put to use on a large-scale basis.

During our initial user tests it has been suggested

by the users that this kind of support might also be valuable for more work related applications, such as enable explorations of other kinds of information, e.g. EndNote files or publication reference lists to represent peoples academic viewpoints. 9. Audio community mirrors – Question: “On the wall or in the head?” – Answer: “Preferably learning by walking around”

We have considered the use of shared community

mirrors to represent the dynamics of how elements are shared between different sets, etc. That idea has proved to be very useful before (e.g. [2], [3]). However, providing that overview would also destroy the necessity to move around to be able to experience the music of the community. As an alternative we want to support mobility, i.e. that people are active in moving around as part of the interaction with the system and thus, we hope that the user can create that understanding of the community rather from the inside of the community by moving around and share music.

10. Related work

There have been several efforts made on using

multiple mobile devices to support various kinds of work related activities (e.g. [5], [9], [10]). However, so far there have been only few efforts made on supporting entertainment and CSCL (Computer

Supported Collaborative Learning) with mobile devices and MANETs (Mobile Ad hoc Networks)(e.g. [6]). One such example is Geney [4], a problem solving application to help children explore genetic concepts using Palm computers. There have also been some attempts made on applying set theory for information retrieval. E.g. [8] presents a Wenn-diagram based interfaces designed for the purpose of supporting effective information retrieval. Instead of keywords, the system uses real objects such as books or CDs as keys for retrieval. The system projects a circle around each book; data corresponding the book are then retrieved and projected inside the circle. By moving two or more circles so that the circles intersect each other, the user can compose a Venn diagram interactively on the desk.” However, in their research they have not applied these ideas to enable explorations of the contents of a mobile ad hoc network. Further on, there has been a lot of research made on how to implement mobile ad hoc networks (MANETs) with several mobile devices. However, the work reported so far concerns only technical issues such as: admission control, asymmetry on TCP performance, or broadcast deliveries [11] without discussing or exploring new computer enabled use situations or new application domains, which was the preliminary objective of the research efforts reported in this paper.

11. Discussion

In our project we have come to think about the

FolkMusic system and its intended use as a part of, or a contribution to the SmartMobs movement as described by Reingold [12], in that the system represent a new kind of mobile community support as well as a mobile reputation systems for “stranger interaction” [12]. This “stranger interaction computer support” rimes well with the description of “Smart mobs” as described in the introduction of this paper. In this sense the FolkMusic system serves a similar purpose as the MemeTags system [2] designed to be an “icebreaker” technology for strangers to start to interact with each other. A main difference between these two projects is that there has not been a focal issue in the design of the FolkMusic system to make it a tool for bringing people together, but rather a tool for unifying people’s preferences. Here, other preferences then music could be valuable to explore. One such direction could for instance be sharing of endnote-files (i.e. files with references to publications) within a research community.

There are of course privacy issues related to file sharing and sharing of other personal preferences in ad-hoc networks. As formulated by Reingold [12]:

“A number of social and technical barriers must be

overcome in order for mobile ad hoc communities to self-organize cooperatively. Nobody is going to contribute their personal area network to a community internetwork unless they feel secure about privacy and trust.” [12, p. 172].

In this project we have tried to overcome this

problem by only distributing a single public folder that is already made public on the comp uter for other online users on the peer-to-peer networks. Although we know that there might still be privacy issues related to this approach we think that this way of unfolding ones personal preferences in a public place (e.g. by sharing ones MP3 files) is somewhat comparable to playing the HiFi stereo on the streets like back in the 80s. To further meet these privacy issues associated with this kind of technology we have thought about different “anonymity modes” as an alternative here (e.g. not revealing sender name, etc). Finally, in really crowded public places that might however be less of a problem since it will be harder to track from whom a specific song is distributed in the ad hoc network. Of course this is dependent upon the implementation of different anonymity modes as discussed above.

A completely different question about the

FolkMusic system concerns its potential for being an educational tool. Although we have only very briefly touched upon this issue in this paper we believe that it would be a fruitful direction to explore even further. We believe that the FolkMusic system could be an exiting “learning by doing”-tool in teaching children about the basics of set theory due to the clear entertainment aspects of the Folkmusic system which is also a focal issue for all edutainment software.

Another future direction for the FolkMusic system is

of a more technical kind. In the current version of the system the user needs to geographically follow another person around to continue to listen to a song that is being streamed from the other device as to avoid that the other device gets out of range for the WLAN pc card. Here, some technical research into service handover from this ad hoc network to a backbone infrastructure (e.g. some stable P2P network on the Internet) without having to disrupt the music played would be of great interest to explore further.

12. Conclusions

In this paper we have reported on an ongoing

project that aims to explore how people can learn to filter large information sets, and get to know communities with use of set theory, while walking around downtown in a city at the same time as they entertain themselves with music. In this paper we have presented FolkMusic, a mobile peer-to-peer application designed to as an educational tool for teaching children the basic elements of set theory while they are entertaining themselves with music (i.e. edutainment). The work reported here builds on the current trends towards edutainment software, increase in use of peer-to-peer technologies such as Kazaa [7] for distributing and sharing entertainment, and the current trend towards mobile and ad-hoc computing solutions. The paper presents the background for this project and an architecture designed to support mobile and ad hoc exploration and experience of the music that is currently ”in the air” of a community through a Wenn-diagram inspired interface.

The work reported in this paper contributes to

current research on CSCL (Computer Supported Collaborative Learning), applications for MANETs (Mobile Ad-hoc Networks), and research on edutainment software. More specifically, this paper contributes to research on MANETs and peer-to-peer (P2P) computing by illustrating how mobile peer-to-peer technologies can be fruitfully applied to both the area of entertainment as well as for educational purposes. Further on, the FolkMusic system contributes to current technical efforts made on ad-hoc networks by providing a good example and a case for further research into how Internet technologies and stable infrastructures might be combined with highly dynamic and wireless ad hoc networks.

This paper also contributes to current research on

persistent interaction in two specific ways. First, it contributes to that strand of research by focusing on persistent traces of music that people can interact around rather then with or via. Second, this research contributes to current research on persistent interaction by moving away from the desktop computer and online computer-mediated interaction to instead explore mobile computing augmented or mediated interaction in an outdoor environment. This extend the scope of persistent interaction research to include all various forms of “social belongings technologies” since

persistence (as well as history) is typically an important element of social interaction.

Future work includes porting of the system to

Pocket PCs equipped with WLAN pc cards and GPS, and implementation of the service handover from media streaming over an ad hoc network to an internet based peer-to-peer network for continuing to listen to a song even if the mobile node gets out of range. This development part of the project will then be followed by a large-scale empirical evaluation of the system in use. The evaluation of FolkMusic in a real use setting will take place in the city of Skellefteå in Sweden. Skellefteå has been chosen as an appropriate evaluation setting since they are right now setting up a WLAN that will cover the whole city and will thus provide the infrastructure necessary for the evaluation. We are planning to involve teenagers in the user study since they are early adopters of mobile technology, listen to a lot of music, enjoy being members of communities and, might know less about the fundamental principles of set theory then adults. Finally, we are also planning for a more focused study on the educational aspects of the FolkMusic system and its pros and cons when it comes to meet the purpose of being an edutainment tool for children to in a very direct way “play around” with.

13. References [1] Borghoff, U & Schlichter, J (2000). Computer-supported Cooperative Work, Cloth. Springer-Verlag, UK. [2] Borovoy, R., et al (1998) Meme tags and community mirrors: moving from conferences to collaboration; Proceedings of the ACM 1998 conference on Computer supported cooperative work, 1998. [3] Borovoy, R., et al (2001) Folk computing: revisiting oral tradition as a scaffold for co-present communities; Proceedings of the SIG-CHI on Human factors in computing systems, 2001, Pages 466 – 473.

[4] Danesh, A., et al (2001) GeneyTM: designing a collaborative activity for the palmTM handheld computer; Proceedings of the SIG-CHI on Human factors in computing systems, 2001, Pages 388 – 395. [5] Davis, R., Landay, J., Chen, V., Huang, J., and Lee, R. (1999) NotePals: light weight note sharing by the group, for the group; Proceeding of the CHI 99 conference on Human factors in computing systems: pp. 338-345. [6] Frei, P., Su, V., Mikhak, B., & Ishii, H (2000) Curlybot: designing a new class of computational toys Proceedings of the CHI 2000 conference on Human factors in computing systems, April 2000. [7] Kazaa peer-to-peer network, www.kazaa.com. [8] Koike, H., Sato, Y., Kobayashi, Y., Tobita, H., & Kobayashi, M. (2000) Interactive textbook and interactive Venn diagram: natural and intuitive interfaces on augmented desk system, Proceedings of the SIGCHI conference on Human factors in computing systems, April 2000. [9] Lamming, M. & Flynn, M. (1994) "Forget-me-not" Intimate Computing in Support of Human Memory, In proceedings of FRIEND21, '94 International Symposium on Next Generation Human Interface, 2-4 February 1994, Meguro Gajoen, Japan. [10] Myers, B., Stiel, H., and Gargiulo, R. (1998) Collaboration using multiple PDAs connected to a PC; Proceedings of the ACM 1998 conference on CSCW. [11] Pagani, E. (1999). Providing reliable and fault tolerant broadcast delivery in mobile ad-hoc networks; Mob. Netw. Appl. 4, 3, Pages 175 – 192. [12] Reingold, H. (2002) Smart Mobs – The next social revolution: Transforming cultures and communities in the age of instant access, Perseus Publishing. [13] Sorensen, C., and Ljungberg, F., (2001) Overload: From transaction to interaction, Planet Internet, Studentlitteratur, Sweden.