MMSE based intialized time domain equalizers

MMSE Based Initialized Time Domain Equalizers Intersymbol interference (ISI) due to multipath propagation is a major obstacle to high-speed digital radio communication. This thesis aims to improve the performance of time domain adaptive equalizers used to mitigate ISI, especially when the equalizer training is difficult in high complexity and/or in the low SNR values. With this purpose, MMSE based initialization technique is proposed. The initialization technique improves especially training features of stochastic gradient algorithms (i.e. LMS). As a consequence training time is shortened, stability of training algorithms increased, misadjustment of equalizers reduced and BER performance improved. The channel impulse response, which is necessary for the calculation of initial values, can be obtained in many applications by combining the synchronization and channel estimation algorithms. The additional calculation complexity which comes with the technique is acceptable. The first section is an introduction, including basic definitions and the conventional techniques. The second section consists of four parts. The first part investigates the purposed initialization for the SISO systems, the second part evaluates the initialization for the MIMO systems. In the third part DSSS system is chosen to examine the system performance in the low SNR region. And lastly, the forth part presents a real-time video transmission as an implementation of techniques by using TMS320C6201 DSP boards from The Texas Instruments. The DSP boards are integrated with a daughter boards composed by a FPGA chip and dual ADC/DAC group. FPGA chip is adjusted to be FIFO and frame synchronizer. Key Words: Equalization, initialization, MMSE, LMS, RLS, MIMO, low-power digital- radio, DSSS, DSP implementation VII