Lee, Zhi Hou
(2015)
Improved multiple input multiple output blind equalization algorithms for medical implant communication.
PhD thesis, University of Nottingham.
Abstract
Medical implant sensor that is used to monitor the human physiology signals is helpful to improve the quality of life and prevent severe result from the chronic diseases. In order to achieve this, the wireless implant communication link that delivers the monitored signal to a multiple antennas external device is an essential portion. However, the existing conventional narrow band Medical Implant Communications System (MICS) has low data rate because of the bandlimited channel is allocated. To improve the data rate in the radio frequency communication, ultra-wide band technology has been proposed. However, the ultra-wide band technology is relatively new and requires living human to be the test subject in order to validate the technology performance. In this condition, the test on the new technology can rise ethical challenge. As a solution, we improve the data rate in the conventional narrow band MICS. The improvement of data rate on the narrow band implies the information bandwidth is larger than the allocated channel bandwidth, and therefore the high frequency components of the information can loss. In this case, the signal suffers the intersymbol-interference (ISI). Instead of that, the multiple antennas external device can receive the signal from other transmitting implant sensor which has the same operating frequency. As a result, the signal is further hampered by co-channel interference (CCI). To recover the signal from the ISI and CCI, multiple-input multiple output (MIMO) blind equalization that has source separation ability can be exploited. Cross-Correlation Constant Modulus Algorithm (CC-CMA) is the conventional MIMO blind equalization algorithm that can suppress ISI and CCI and able to perform source separation. However, CC-CMA has only been analyzed and simulated in the modulation of Phase Shift Keying (PSK). The performance of CC-CMA in multi-modulus modulation scheme such as 4-Pulse-amplitude modulation (PAM) and 16-Quadrature amplitude modulation (QAM), which has higher data rate than PSK, has not been analyzed. Therefore, our work is to analysis and optimize CC-CMA on the multi-modulus modulation scheme. From our analysis, we found that the cost function of CC-CMA is biased cost function. Instead of that, from our simulation, CC-CMA introduces an unexpected shrinking effect whereby the amplitudes of the equalizer outputs have been reduced, especially in multi-modulus modulation scheme. This shrinking effect is not severe in PSK because the decision of a PSK symbol is based on phase, but not amplitude. Unfortunately, this is severe in multi-modulus modulation scheme. To overcome this shrinking effect in multi-modulus modulation scheme, we propose Cross-Independent Constant Modulus Algorithm (CI-CMA). Based on the convergence analysis, we identify the new optimum dispersion value and mixing parameter in CI-CMA. From the simulation results, we confirm that CI-CMA is able to perform equalization and source separation in the multi-modulus modulation scheme. In order to improve the steady state performance of CI-CMA, we perform the steady state mean square error (MSE) analysis of CI-CMA using the energy preservation theorem that was developed by Mai and Sayed in 2001, and our result is more accurate than the previous work. From our analysis, only the reduction in adaptation step size can reduce the steady state MSE, but it is well known that the MSE is indeed a tradeoff with the speed of convergence. Therefore without sacrificing convergence speed, our last effort is to propose hybrid algorithms. The hybrid algorithms are done by combining a new adaptive constant modulus algorithm (ACMA), a decision directed algorithm and a cross-correlation function. From the simulation results, we found that the hybrid algorithms can show low steady state error and thereby improve the reliability of the communication link. The main achievement of this thesis is the discovery of new dispersion value through the convergence analysis.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
