Delay-Doppler and Angular Domain 4D-Sparse CSI Estimation in OTFS Aided MIMO Systems

Abstract

A convenient delay, Doppler and angular-(DDA) domain representation of the multiple-input multiple-output (MIMO) wireless channel is conceived for deriving the end to end relationship in the delay-Doppler (DD)-domain for orthogonal time frequency space (OTFS)-based communications. Subsequently, a time-domain pilot based model is developed for estimating the DDA-domain channel state information (CSI) of our MIMO OTFS system. The key differentiating feature of the CSI estimation model derived is its ability to exploit the 4-dimensional (4D)-sparsity arising in the DDA-domain, given the limited number of dominant scatterers. Furthermore, the training overhead of the proposed framework is low, and the pilot placement is quite flexible, necessitating no guard-interval. Finally, an orthogonal matching pursuit (OMP) framework is employed for 4D-sparse CSI acquisition, followed by deriving the Oracle minimum mean squared error (Oracle-MMSE) and its Bayesian Cramer-Rao lower bound (BCRLB). Our simulation results confirm the improved CSI estimation performance attained over the benchmarks.

Estimation of Dispersive High-Doppler Channels in the RIS-Aided mmWave Internet of Vehicles

“…Each cascaded two-hop path can be distinguished by its path delay, the AoA at BS or the equivalent AoD at the RIS. The support sets of the three dimensions are sparse, since the number of paths is limited in mmWave channels [37], [38], which is illustrated in Fig. 2.…”

“…0 and (c d ) 0 can be derived from the expressions ( 31) and (37). After a number of iterations, more precise estimates of the equivalent AoD and of the complex gain of each path can be obtained.…”

Low Complexity Detection of Spatial Modulation Aided OTFS in Doubly-Selective Channels

“…According to [8], we have τ i = li M ∆f and ν i = ki N T , where l i and k i denote the normalized delay and Doppler indices of the ith path. We note that if the RAs are hosted by a fixed-location base station (BS) in a uniform linear array (ULA) form, the channel model of ( 4) may be refined as the delay, Doppler and angular (DDA)-domain channel model [20], [21]. Consequently, by considering the array steering vectors at both RA and TA sides, our proposed SM-OTFS scheme can be directly extended to DDA-domain channel-based MIMO-OTFS systems.…”

“…where p(y y y|s s s i ) is given by ( 13), when s s s i is transmitted. It can be readily shown that (20) attains its maximum value in the case of p(y y y|s s s i ) = 1/2 L b , ∀i. Therefore, the DCMC capacity of the SM-OTFS system can be shown to be…”

“…We note that the closed-form DCMC capacity expression is computationally intractable, since there are multidimensional integrals and summations of 2 L b exponential functions in (20) and (21), respectively. Moreover, it can be seen from ( 21) that the DCMC capacity is dependent on all the realisations of s s s i .…”