Semi-Blind Channel Estimation for Diffusive Molecular Communication

A Survey on Estimation Schemes in Molecular Communications

“…This is different from pilot-based estimation which only considers the received pilot-carrying observations in the estimation process but excludes data-carrying observations. Incorporating datacarrying observations into the estimation can significantly enhance the estimation accuracy and/or improve the data rate [55].…”

“…An expectation maximization (EM)-based estimation scheme was first proposed in [55]. The data vector β constitutes hidden information at the RX.…”

“…This is followed by the maximization step (M-step), in which the posterior probability of the hidden data is maximized to acquire an updated estimate of CIR. [55] also proposed other two semi-blind estimation schemes based on the decision-directed (DD) strategy. The idea of DD strategy is to use the channel estimate acquired through pilot-based estimation for performing data detection.…”

“…The idea of DD strategy is to use the channel estimate acquired through pilot-based estimation for performing data detection. The detected data is in turn treated as a new set of pilots to perform another cycle of channel estimation [55]. We note that the analytical derivation of the CRLB can be very challenging in semi-blind estimation, because the log-likelihood function of the observations becomes complicated when the statistics of the data symbols are taken into account.…”

“…We note that the analytical derivation of the CRLB can be very challenging in semi-blind estimation, because the log-likelihood function of the observations becomes complicated when the statistics of the data symbols are taken into account. [55] applied the Monte-Carlo method to obtain accurate approximations of the Fisher information matrix I −1 (c). The semi-blind CRLB was then obtained by evaluating tr I −1 (c) .…”