Low Power UWB Signaling over a Time-Varying Multipath Channel using Multiple Receiver Antennas
Abstract
This thesis explore a digital wireless communication system which is intended for commu- nication from the abdominal region to an on-body receiver. Where a capsule is meant for image capturing of the intestine for better diagnosis of a patient. As the capsule carries limited power, which must supply power to light, image capturing and processing of data, the power limits for transmission is evident. In addition to the power limitation, the path loss through human tissue is significant. To obtain reliable communication, an investment in large bandwidth is therefore needed. Where in this thesis pulse-position modulation (PPM) is investigated.Even though the transmitter is subject to large limitations, the receiver on the body surface has unlimited resources and use of multiple receiver antennas are examined. Where the signals are combined using different schemes, and when only the signals with the best signal- to-noise ratio (SNR) are considered. As the capsule will travel through the digestive system the unknown channel will be time-varying. Hence, an adaptive equalizer will be investigated, to prevent distortions made by the expected multipath channel. For adaptive equalization, the least mean square (LMS) and recursive least square (RLS) algorithm are investigated.The results showed that the main obstacle for obtaining a reliable communication link, is the power limitation in combination with high path loss on the channel. Simulation showed that the signal distortion from the multipath channel was significantly mitigated, when multiple receiver antennas were used. Where combining only four out of ten receiver antennas, gave almost equal performance, as for combining all the antennas. In fact, when multiple receiver antennas were used, receivers using equalization did not obtain a better bit error rate (BER) than without equalization. However, the equalizers were not optimized for all signal-to-noise ratios, and are in general less efficient than nonlinear equalizers which should be reviewed in future work. Still, the performance gain in terms of BER the optimal equalizer could achieve, was found to be around 2-3 dB, for bit error probability 10−4. To avoid using designated power for equalizer adaption, blind equalizers should be investigated in future work.