MMSE antenna diversity equalization of a jammed frequency-selective fading channel

On the performance of the Viterbi equalizer in the presence of channel estimation errors

“…The performance of Viterbi detectors in presence of additive white Gaussian noise (AWGN) has been studied by Forney [3]. This and later contributions [4] assume perfect knowledge of the channel between the emitted data and the received signal.…”

“…and, for simplicity, that all the symbols are equiprobable. For the known channel with AWGN, the ML data estimate is given by (3) Suppose that there exists an interval of size such that all the symbols of are different from the corresponding symbols of whereas the preceding/following symbol is the same for and Define and to be the vectors of symbols corresponding to this interval and the vector of errors Forney called such an occurrence an error event of length A subevent of the error event is that is "better" than in the sense of the ML metric (4) where is the subvector of and is the block of corresponding to the error interval. The probability of allows us to calculate tight bounds for various error statistics such as the bit error rate and mean time between the error events, see e.g., [3].…”

Multisensors receivers using a filtered reference: application to GSM

“…Let us first reformulate the results of [4] in the case of cyclostationary non-circular GBussian jammers. For a white replica b(n), the adaptive structure of figure 1 operating on the modified antennas outputs x(t) and optimized for minimum MSE E[e(n)2] will give (almost) the required frequency responses when the constraint y(0) = 1 is imposed.…”

“…The best receiver for minimizing the Bit Error R.ate (BER) is well known : it is the Maximum Likelihood Sequence Estimation (MLSE) receiver. This receiver and its efficient multichannel implementation have been analyzed in [3] assuming that jammers emit Gaussian circular stationary signals. However, in radio-mobile communication systems like GSM, jamming is essentially due to CCI: these jammers are also communication systems that exhibit the same features as the signal of interest, namely cyclostationarity and non-circularity (for GSM).…”

“…As for Gaussian circular stationary jammers [4], the MLSE receiver involves a ST filter that can be estimated from a replica: in section 3, the resulting adaptive filters for cyclostationary non-circular jammers are compared to the ones operating on stationary circular jammers. A suboptimal practical structure for the receiver is analyzed in section 4, and its limitations are presented.…”