| Summary: | Two novel optimized delay diversity (ODD) schemes for suboptimum equalization
are proposed in this thesis. In [1, 2], an ODD scheme was proposed based
on the Chernoff bound on the pairwise error probability (PEP) for maximum-likelihood
sequence estimation (MLSE) [3]. It was shown that the ODD scheme
outperforms the generalized delay diversity (GDD) scheme proposed in [4] in
frequency-selective fading channels. However, the MLSE scheme is too complex
for most practical applications. Therefore, low-complexity equalization
schemes such as decision-feedback equalization (DFE) [5] or even linear equalization
(LE) [6] have to be used. In this work, two novel ODD schemes
are investigated. The ODD transmit filters of the two novel schemes are
optimized for correlated multiple-input multiple-output (MIMO) frequencyselective
Rayleigh fading channels with suboptimum DFE or LE employed at
the receiver, respectively. An equivalent discrete-time channel model containing
the DD transmit filters, the pulse shaping filters, the mobile channel, and
the receiver input filters is first given. Then, the worst-case pairwise error
probabilities (PEPs) for both DFE and LE are derived based on the discretetime
channel model and the error variances of the two schemes. Finally, a
stochastic gradient algorithm for optimization of the ODD filter coefficients is
proposed. The algorithm assumes knowledge of the channel impulse response
(CIR) at the receiver while only the statistics of the CIRs are required at
the transmitter. The proposed algorithm takes into account the equivalent
discrete-time channel, the operating signal-to-noise ratio (SNR), the modulation
scheme, the length of the ODD transmit filters as well as the correlations of the transmit and receive antennas. The resulting ODD filters are applied
to GSM1 [7, 8] and EDGE2 [9, 10]. Simulation results show that the ODD
filters obtained in this work achieve a lower bit error rate (BER) than those
obtained in [1, 2, 4] when DFE and LE are used at the receiver, respectively.
The results of this thesis have been summarized in [11, 12]. === Applied Science, Faculty of === Electrical and Computer Engineering, Department of === Graduate
|
|---|
