BP, MF, and EP for Joint Channel Estimation and Detection of MIMO-OFDM Signals

Joint spare channel estimation and decoding for orthogonal frequency division multiplexing using combined message passing

“…The joint channel estimation and decoding iteration is presented in lines (18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36), denoted by global iteration, which includes the detection-decoding (lines [19][20], channel estimation (lines [21][22][23][24][25][26][27][28][29][30][31][32][33][34][35] and noise precision estimation (line 36). Note that the channel estimation and noise precision estimation can be arranged as an inner iteration, i.e.…”

“…Recently, message passing techniques including belief propagation (BP) [11,12], mean field (MF) [13][14][15], expectation propagation (EP) [9,16,17], and approximate message passing [18,19] have emerged as useful tools in designing powerful communication receivers. It is worth mentioning that, by exploiting their virtues while avoiding their drawbacks, BP, EP, and MF have been integrated carefully to design superior receivers with low complexity [20][21][22]. With regard to OFDM receiver design, there has been various message passing based approaches to JCED in the literature [4][5][6][7][8].…”

Joint Frequency Domain Channel Estimation and Equalization Based on Expectation Propagation for Single Carrier Transmissions

“…Such techniques, when formulated as message passing algorithms, have proven themselves to be practical for addressing complex communications systems [8]. In particular, they have been used for channel impulse response (CIR) estimation in multi-carrier systems with hybrid BP-(EP)-MF frameworks [9,10], and with an EP-only frameworks in [11,12]. In these works, FD data symbols are discrete variables (for instance, due to the use of orthogonal frequency domain multiplexing, i.e.…”

“…The technical contribution of this paper is CSI estimation with non-discrete data in the FD with EP, which requires resolving factor nodes where Gaussian variables are multiplied. This problem is addressed by resolving the multiplier node with three approaches: (a) with expectation-maximization (EM) [2], (b) hybrid EP-MF [9,10] and (c) EP extended with the quadratic-approximation (QA) [12]. Moreover, semi-blind receivers need to handle the correlations caused by the CFR interpolation of CIR over the data block through a truncated discrete-Fourier transform (DFT).…”