Blue3 A Programmable Remote MP3 Sensor

Blue3 A Programmable Remote MP3 Sensor Anonymous Anon Inc. Abstract. We present the concept and the prototypical implementation of Blue3, a programmable device that features a wireless communication interface, a microcontroller, and a sophisticated MP3 encoder/decoder chip. It is capabale of playing back and recording MP3 encoded sound streams, and can therefore act as a high-quality sound sensor and actuator. It is possible to transmit sound files from Bluetooth-enabled clients at high-quality bitrates. Blue3 can be deployed in public spaces as a remote jukebox, it can be used for surveillance purposes, and for two-way communications. 1 Introduction MP3 has become a synonym for pieces of music. This is due to the availability of small, affordable MP3 players, the infamous Napster file sharing service, and the liberally licensed MPEG Layer-3 (a.k.a. MP3) encoding standard [4]. The MP3 file format has become ubiquitously available. Playing back MP3 files is one of the key features that is built into many appliances appearing on the market. Mobile phones, handheld computers, and car radios are all able to play back sound from MP3 files. Most of these devices provide an (exchangable) storage medium for storing a substantial number of files. As the price per storage bit decreases sharply, it seems reasonable to assume that the ability to distribute MP3 files amongst many devices increases. But having MP3 files stored on a portable device is not enough. One also wants to listen to them. The current method of choice is to use headphones or a wired connection to an amplifier/speaker pair. Wireless speaker systems are also available, but they don t use frequencies and encoding schemes compatible with computer hardware (transmission in the 900 MHz band). On the other hand, most PDAs and laptop computers have built-in digital wireless interfaces such as Bluetooth or WiFi. However, transmitting audio data over such a link requires a decoding step at the receiver s side. The idea of Blue3 is to complement the ability of small computing devices to store MP3 data with a wireless playback facility. It provides a Bluetooth communication interface and a decoder for MP3 data. It can directly feed analog audio data into headphones and (amplifying) speakers. It therefore constitutes a remote service point for playing back sound files. The Blue3 prototypical implementation is based upon a BTnode [2], which is connected to a PCB holding the Micronas MAS3587F chip [1]. This chip represents the latest generation of MP3 chips. Its key features are

II MP3 decoding (supporting fixed and variable bitrates), MP3 encoding with a variety of bitrate/sampling combinations, A/D and D/A conversion and amplification, an I 2 C control interface. The BTnode comprises a universal sensor platform with a wireless communication interface. Its main applications are the formation of sensor networks [5] and the creation of smart objects [6]. It is possible to connect from a Bluetooth-enabled computer (such as a PC or a PDA) to a Blue3 module and use it for playing back high-quality audio data. With our prototype implementation, it is possible to handle encoding bitrates of up to 96 kbit/s. This limitation is due to the preliminary state of development of the components. Blue3 is also one of the first applications that makes extensive use of the available bandwidth on the BTnode platform. The encoding features of the MAS chip are not yet employed in our prototype. Adding this feature opens a range of new applications, such as sound monitoring, headset functionality, and other two-way communications. Blue3 is a new appliance that enhances the capabilities of existing computing infrastructures. It provides limited, dedicated functionality and supports the modular construction of federated Bluetooth-based systems. It adds no extra disruptive elements to existing user interfaces, but can be smoothly integrated with applications. In the following sections, we describe the overall concept of Blue3, describe the implementation of a prototype, and evaluate the concept. We conclude with a discussion of related work and some closing remarks. 2 Motivation Mass storage is increasingly available in many small, personal devices like cell phones, MP3 players, and PDAs. A large amount of this storage is often dedicated to storing MP3 music files. When listening to them, the sound experience is usually degraded because only headphones are available, or the built-in speakers offer inferior sound quality. This is conventionally overcome by wiring speakers to the device. An alternative approach is installing fixed speakers in a room that are equipped with a Blue3 module. Besides a power source, no wiring is required. Music files can now be directly transmitted to the Blue3 module, which decodes the data and plays it back on the speakers. If the user s client software is aware of the Blue3 module, the user has only to select Blue3 as output device and can start using her client software as usual. 2.1 Applications Since the MAS chip can not only decode MP3 data, but also encode audio as MP3, Blue3 is usable in a two-way mode. This opens up a wider range of applications:

III Distributed sound sensor Blue3 modules can be used to monitor an environment for sounds. The BTnode can be used to sample sound at a moderate rate. These sound samples can be analyzed for interesting patterns. If a match is found, the BTnode triggers high-quality sound recording by the MAS. The sound data is then transmitted through the sensor network to a base station. The recording can also be triggered by a predefined scheme, or through a remote event. High-quality headset Normal headsets rely on the Bluetooh audio specification [3] that supports the synchronous transmission of digitally encoded voice. This channel is limited to 64 kbit/s, thus it is not usable for the transmission of highquality sound. Blue3 overcomes this limitation by transmitting MP3-compressed audio data over an asynchronous data link. Babyphone Blue3 can be used for a variety of sureveillance tasks, such as monitoring what s going on in a child s room. For example, sound can be recorded to provide entertaining memories in the future. Fact collection Densely deployed Blue3 modules can serve as a fact collecting infrastructure. This is intented for people who like to make recordings of their behaviour and interactions throughout their life. The Blue3 modules record the sounds in a person s surrounding and transmit the encoded data to her personal storage device. While cameras might often be considered too invasive, sound recordings could be acceptable in many situations. 3 Implementation The hardware implementation comprises a PCB that contains the MAS chip and some passive components on a size of about 9 cm 2 (about one third the size of a BTnode). The PCB is connected to the BTnode by a serial link (data and I 2 C) and pulls its power from the BTnode. Speakers can be directly connected to the PCB. The BTnode was not physically modified. The basic idea when building Blue3 was to use the BTnode as a wireless Bluetooth link between an audio source and the decoder. Note that basically the same applications would be possible with a specifically designed platform that would include a microcontroller, a Bluetooth module, and an MP3 decoder. The difference is, however, not crucial. The decision for a modular design has been made simply because the BTnode was already available and a PCB hosting only the MAS is simpler and cheaper than a combined platform. The latter arguments are especially important since we expected to have errors in early designs, and these should be easier to find and cheaper to correct on a dedicated module. On the software side, the BTnode is loaded with the original BTnode systems software, which provides a Bluetooth stack and basic event scheduling. Additionally, a driver was added that provides control routines for configuring the MAS chip. The application running on top of it uses a simple protocol for

IV transferring MP3 data from a remote host (in our implementation, a PC) and playing it back via the MAS chip. On the PC, a simple server was installed. When running, it connects to the (dedicated) BTnode and the data transfer is started. The Blue3 prototype is not yet integrated into larger sensor communities. A possible extension is to transfer MP3 data over multiple hops within a Bluetooth sensor network. For applications like Blue3 as a public space jukebox, client software has been developed but is not yet integrated with the Blue3 prototype. The client software allows the easy management of music titles and manages the communication amongst a group of jukebox users. Blue3 is not able to store large amounts of application data. Internal RAM is used as a buffer for balancing variations in link speed. Received MP3 data is immediately scheduled for playback, and sound recordings would have to be sent out immediately after creation. In future versions, Blue3 has to store additional user information when multiple users are concurrently using its services. However, we expect to be such meta-data of moderate size, thus the capacity of the BTnode would be sufficient. 4 Evaluation It is important to note that the BTnode concept is a research platform, with its system software and hardware under ongoing development. For our prototype implementation, we have used a snapshot of the BTnode modules available at the time we started the implementation (August 2003). The development of the PCB and the basic software components took about 10 weeks (1 person). The development is in a preliminary state, so the results can only give a rough impression of what is possible with Blue3. We were able to send an MP3 stream encoded with a fixed bitrate of 96 kbit/s from a PC to the Blue3 prototype. The transmission and playback of such a data stream is rather stable. The bitrate is sufficient for a high-quality sound experience. In principle, our current Bluetooth module supports transmission rates of up to 460 kbit/s in one direction, so even 256 kbit/s-encoded MP3 streams should be possible to handle. For achieving such high bitrates, the BTnode system software has to be modified and the stability of the involved Bluetooth stacks (BTnode and PC) has to be improved. For interaction with a large number of device types and to reduce the dependency on dedicated client software, a universal soft real-time data transmission protocol could be employed. It is not necessary that the protocol guarantees the full integrity of a data stream, since small gaps in an MP3 stream are usually tolerable. In future versions, we plan to employ a protocol, which offers near real-time data delivery under moderate integrity requirements.

V 5 Related Work One of the key ideas of the Personal Server concept [7] is to employ other devices, which are available in the close environment, for interaction. Personal Server has no user interface and communicates only through a Bluetooth interface. MP3 data stored on such a device could be played back on a Blue3 module, if they are able to establish a Bluetooth connection. Still, a third device is necessary that delivers a playback schedule (such as a playlist), usually on request by the holder of the Personal Server device. There is a large variety of MP3 and media players available off the shelf. These end-user products basically come in two flavors: the minimalistic and the feature-rich flavor. Minimalistic devices are very small and light, offer a basic user interface and a restricted (but increasing) set of features, usually sound playback, recording, and host access. Feature-rich devices have a bigger storage capacity and allow the playback and recording of movies. Wireless interfaces are not yet common, but are technically possible. Bluetooth, however, has the disadvantage of a limited bandwidth, which is not convenient for loading large amounts of data onto such a device. Thus, we consider the Blue3 approach of streaming music data still useful with the advent of more advanced media players. A number of wireless audio transmitting systems are avaialable on the market. To our knowledge, none of them works in the same way as Blue3 does. They often come with dedicated transmitters and receivers and are certainly not able to interact with a dynamic group of heterogeneous audio sources. 6 Conclusion We have shown that it is feasable to use an existing sensor platform with its limited capabilities to transmit high-quality sound data (MP3 encoded at 96 kbit/s) over a Bluetooth connection and hand it over to an extension board for playback. The platform is prepared for recording sound (and transmitting it to interested parties). The cost of the prototypical extension board is approx. 70 Euro. A web site with project details is available and others are encouraged to build on our experience for creating new, challenging applications. References 1. Micronas Semiconductor Holding AG. www.micronas.com. 2. J. Beutel, O. Kasten, F. Mattern, K. Römer, F. Siegemund, and L. Thiele. Prototyping Wireless Sensor Network Applications with BTnodes. In European Workshop on Wireless Sensor Networks. IEEE, 2004. 3. Bluetooth SIG. Specification of the Bluetooth System, 2001. Version 1.1. 4. K. Brandenburg and H. Popp. An Introduction to MPEG Layer-3. EBU Technical Review, June 2000. 5. K. Römer. Tracking Real-World Phenomena with Smart Dust. In European Workshop on Wireless Sensor Networks. IEEE, 2004.

VI 6. F. Siegemund and C. Flörkemeier. Interaction in Pervasive Computing Settings using Bluetooth-enabled Active Tags and Passive RFID Technology together with Mobile Phones. In Proc. of PerCom 2003 (IEEE International Conference on Pervasive Computing and Communications. IEEE, March 2003. 7. R. Want, T. Perrig, G. Danneels, M. Kumar, M. Sundar, and J. Light. The Personal Server: Changing the Way We Think about Ubiquitous Computing. In 4th Int. Conf. on Ubiquitous Computing, volume 2498 of LNCS, page 194 ff. Springer-Verlag, 2002.