Limited-feedback transmit diversity schemes: Proposed methods and impact of feedback delay and channel estimation errors

In this thesis we propose a practical hybrid transmit diversity scheme that applies phase adjustment on signals of stronger subchannels using a few bits of channel information. This scheme requires considerably reduced electronic circuitry at the transmitter compared to the partial phase combining (PPC) and equal gain transmission (EGT) schemes and can outperform them by avoiding energy waste over insignificant subchannels. In 2-transmit antenna systems, the proposed hybrid method adjusts the power of transmit signals using one bit magnitude information along with applying PPC. This scheme offers a very close performance to the optimal beamforming scheme (OBS) by using very few feedback bits and maintaining a simple implementation. We also propose an FFT-based transmit antenna selection scheme which substantially outperforms the conventional antenna selection scheme in correlated fading channels and systems with uniform linear array of omnidirectional antennas. We propose a fast almost-optimal procedure for codeword extraction of PPC method which significantly reduces computational complexity.; We analyze the joint and separate impact of channel estimation errors and feedback delay on the performance of PPC, hybrid and transmit antenna selection schemes. We derive the average SNR expression for the these methods under channel estimation errors and feedback delay. We show that the channel estimation errors and feedback delay have separable impacts on the average SNR performance and the average SNR degradation due to estimation errors is identical in these methods. We also derive the exact closed-form expressions for the average BER of transmit antenna selection under channel estimation errors and feedback delay in the 2 x 1 case. We show that for all the transmit diversity schemes, channel estimation errors do not change the diversity gain and degrade the performance by reducing the coding gain. We also show that feedback delay results in diversity gain reduction and if the channel estimation is perfect, feedback delay reduces the diversity gain to 1.