Wireless transmission of digital data has gained much importance recently and a variety of wireless technology have evolved. One of these is Bluetooth, a technology which aims at replacing the cabling of PC peripherals and communication devices. Phonak, our industry-partner and one of the leading manufacturers of hearing aids, plans to launch projects which will add wireless-features to their products.

This work had the goal to transmit digitally coded audio data with Bluetooth technology. Compared to conventional analogue transmission, Bluetooth offers additional features such as:

• Each individual sender and receiver has to register by the master before being enabled to take part in communications.
• Addressing of individual devices within the Bluetooth-Net.
• Improved privacy due to standardised encryption algorithms.

Two options for the perceptual coding algorithms had been offered to us:

• Ogg-Vorbis, a completely free technology.
• MP3, a patented ISO-MPEG-standard.

We decided to implement MP3 due to its wide spread, and because there exist a large variety of encoders and decoders, some of which are being part of open-source software projects. Phonak intends to develop a completely new proprietary sound technology for their future products

In the system we built, an MP3 encoder is used to encode a 64kBit/s digital bit stream representing audio data. A sampling frequency of 16kHz has been used. The MPEG-bitstream then is fed to the Bluetooth device, using a serial interface and a Bluetooth protocol stack. From the protocol stack, the data is transfered to the radio part. Bluetooth offers data- and voice-channels. We used a data channel in order to transmit the data. On the receiver's end, the bits are handed over again to the receiver's protocol stack. From the stack data is again transferred by the use of a serial link to a second PC. There, an mp3-decoder program runs and outputs the audio information, using the PC-sound-card.

What surface (or space) does a cow need in order to have a cowshed being called 'appropriate' for the species ?

To answer this question the Swiss government offers a research grant by the end of 2001. The project should provide substantial contribution to the question on how much space cows need for feeling well. During an experimental stage, data on each individual cow are being collected in a dedicated cowshed.

In regular time intervals the cow's name, its ID, its co-ordinates and its current position within the cowshed (i.e. resting area, feeding area) are collected. Up to now, staff-members had to monitor the cows, day and night which is a very time-consuming work and additionally results sometimes have been inconsistent due the observer's fatigue. The goal of our diploma project was to simplify the monitoring such that an image processing system, comprising a camera, an image capturing hardware and a PC would take care of the surveillance of the cows, day and night.

The monitoring system, we developed, is based on the following technologies: Each cow in the cowshed is equipped with a transmitter, which transmits different blink-codes in regular intervals by using an infrared LED. A digital camera with an infrared-filter is installed at the cowshed's ceiling and captures within the overall image, the blink-code of each individual cow and copies it's information to the PC's memory where our monitoring software can access all the information.

The monitoring software 'filters' each blink-codes out of the captured images and will make it available for further data treatment. For each cow, a time-stamp, the position are generated within a software task and finally, a lower prioritised task will store the information in a database, linked to our image processing software. Each cow's position can be displayed on screen in a dedicated zone which has been pre-defined by the user prior to the analysis by the use of a simple line-drawing interface.

Phonak AG, our Industry Partner for this diploma project, is one of the leading companies for hearing aids. Phonak intends to develop a proprietary and extremely low-delay coding scheme, offering ''graceful degradation''. Within the many coding schemes, applicable for such a use, Advanced Audio Coding (AAC) is one of the most up-to-date coding schemes, being part of the MPEG-4 standard. AAC is a perceptual coding scheme, taking advantage of psychoacoustic phenomena such as simultaneous and temporal masking of the human auditory system. The goal of any perceptual coder is to reduce redundancy but also irrelevancy. The idea is to decompose the incoming audio signal into subbands, to analize them and finally apply a quantisation scheme which takes care of keeping the resulting quantisation noise beyond the masking threshold of the human ear. Additionally, AAC takes advantage of a redundancy reduction, using Huffman coding on the subband samples. One of the achievements of AAC is the prediction. Using suitable algorithms one tries to determine the next value. Thus only differences are to be transferred to the preceding value. The prediction is used both in the time domain as well as in the frequency domain. In our diploma-work we develop an AAC-Player, based on a code which was available within the faad open source project. It can play AAC-files in real-time and is tested for MPEG-conformance, using MPEG-AAC reference bitstreams. Phonak aims at using their coding scheme for wireless transmission where ''graceful degradation'' is a very important feature, especially for hearing-impaired persons. ''Graceful degradation'' means, that the digital transfer behaves similarly as the analog. If the bitrate decreases, the audio signal should not abort, but only get qualitatively worse.