| Summary: | As technologies around us are emerging at a rapid rate, wireless body sensor networks (WBSN)s are increasingly being deployed to provide comfort and
safety to patients. WBSNs can monitor the patient's health and transmit the collected data to a remote location where it can be assessed.
Such data is collected and transmitted using low battery devices such as specialized sensors or even smart phones. To elongate the battery life, the energy spent on acquiring, processing and transmitting the data should be minimized. The thesis addresses the case of electroencephalogram (EEG)signals. It studies the energy spent in the sensor node, mainly in the processing stage, i.e. after acquiring the data and before transmitting it to a certain receiver-end in a wireless fashion. To minimize this energy, the number of bits to be processed and transmitted must be minimized. Compressive sampling (CS) is ideal for such a purpose as it requires minimal number of computations to compress a signal.
For transmitting the signals acquired by CS, we studied their quantization followed by 2 different schemes. Scheme 1 applies lossless Huffman coding for further compression that allows perfect reconstruction. This is followed by a Reed-Solomon (RS) code to protect the data from errors during transmission. Scheme 2 does not apply any further compression. It only quantizes the data and applies the RS code to it.
Both schemes were then enhanced by adding an interleaver and deinterleaver that improved the results. The data was then sent in packets over a transmission channel which was simulated using a 2-state Markov model.
Under ideal channel conditions, Scheme 1 with Huffman compression decreased the total number of bits sent by 5.45 %. The best scheme however was scheme 2 followed by an interleaver. It achieved the best signal reconstruction results under normal or noisy channel conditions. === Applied Science, Faculty of === Electrical and Computer Engineering, Department of === Graduate
|
|---|
