Probabilistic data association for symbol detection in MIMO systems

A Novel Probabilistic Data Association Based MIMO Detector Using Joint Detection of Consecutive Symbol Vectors

“…The performances of the PDA-JD detector, the conventional PDA detector proposed in [4] and the ML detector are compared in Fig. 3.…”

“…Both the PDA detectors do not use the optimal detection sequence and refinements we have mentioned in the above section. This is done in order to provide the same comparison conditions as in [4]. It is easy to find the advantage of the PDA-JD detector over the conventional PDA algorithm.…”

“…When quasi-static channel is taken into account, we have . Following the PDA principle ("Gaussian approximation") in [4], the pseudo noise j′ U is approximated by a complex Gaussian distribution with matched mean…”

“…he probabilistic data association (PDA) algorithm is a reduced complexity approximation to the a posteriori probability (APP) detector, and has attracted great interests for suboptimal detection of CDMA [1] and spatial multiplexing multiple-input multiple-output (MIMO) systems [2]- [4].…”

“…The key feature of the conventional PDA algorithms in [1]- [4] is a repeated conversion of a multimodal Gaussian mixture probability structure, the true probability function for the noise and interference, to a single Gaussian with matched mean and covariance. Consequently, an implicit gap exists between the "Gaussian approximation" and the Gaussian mixture especially when the number of transmit antennas is small [7], with the result that the conventional PDA detectors are not generally near optimal.…”