| EEG is widely used in medical field as a means of diagnosing the brain disease. Inthe wireless multi-channel EEG data acquisition system, to achieve the high samplingrate at a high resolution, the data rate will exceed the highest wireless communicationrate when EEG data are transmitted through the wireless transmitter. In this case, it isnecessary to reduce the volume of the transmitted data. On the other hand, Yates et. al.has shown that the primary source of power dissipation in the wireless EEG system is thewireless transmission which directly depends on the volume of the transmitted data. Thereduction in volume can translate into a lower total energy consumption, which canprolong the life of the system.For a long time, people have been committed to researches of reducing EEG data.There are generally two ways to reduce the data volume. One is compressing the EEGdata by reducing the data redundancy with compression methods, and the other achievesthe goal by transmitting incomplete signals. In the second method, they just transmit the‘important’ regions through detecting the information content of EEG signals[2]. Thealgorithms to select ‘relevant’ activity are imperfect, so the second method is seldom usedin applications. Compression methods are relatively available for reducing the datavolume.In the wireless EEG data acquisition system, the ADC the system uses has an about10KHz sampling rate and a10-bit precision. In this case, the data acquisition rate of thesystem is about200Kbps. However, not only the EEG data but also the wireless protocolpackets are transmitted through the transmitter. The data part accounts for about70%in atransmitted data frame. Hence, the transmission rate of the system is about290Kbps.However, the wireless RF zig-bee requires that the transmission rate cannot exceed250kbps, so the system will be overloaded if we do not reduce the EEG data. It alsocannot transmit or receive data in time. The system requires the realtime more than thehigher CR, as the realtime can ensure the system transmits the EEG data within a shortdelay. A high CR provides the system a better opportunity for transmitting the data fast.However, the complexity of the algorithm that achieves a high CR is usually high as well. As the algorithm can meet requirements when CR is about2:1and it is achieved on thestm32w108chip, the algorithm we adopt cannot be complex and the compression timeshould be very short.As a result, we adopt a lossless compression method because of its simpleness.In thealgorithm, EEG are first coded by DPCM that removes the temporal redundancy of thedata, then adaptive Huffman coding is used to compress the data further by removing thestatistical redundancy.By analyzing the data coded by DPCM, we can see that the data should berepresented with a6-bit precision. However, they all take up one byte of memory. Inorder to make full use of the system resource, we can make the data after being coded byDPCM take up more bits memory by reducing the sampling rate. The method greatlyreduces the system resource consumption and expenses at the same compression method. |
