This paper proposes a channel estimation (CE) scheme composed of two successive interference cancelation-inspired approaches for a two-user power domain downlink non-orthogonal multiple access system. One approach is intended for the weaker user, and the other approach is intended for the stronger user. Both CE approaches employ a constant step-size least mean squares (CSS-LMS) algorithm, using partial pilots to estimate a Rayleigh flat fading channel with a Jake's Doppler spectrum. Closed-form analytical expressions for the CE error variance are presented for each approach and the mean square error (MSE) is used as the performance metric. Through simulation, the CE performance of the stronger user and the weaker user is assessed for different probabilities of error in detection of the unknown pilot. A reasonable probability of error in detection is then selected and employed in evaluating the CE performance of both schemes against their theoretical analyses. Overall, the theoretical analysis matches the simulation results.

No abstract