Diffusive MIMO Molecular Communications: Channel Estimation, Equalization and Detection

TL;DR

This paper explores channel estimation and equalization techniques for diffusive MIMO molecular communication systems, addressing inter-symbol and inter-link interference to improve data rates and reliability.

Contribution

It introduces training-based channel estimation methods, designs optimal training sequences, and evaluates equalization strategies specifically for diffusive MIMO molecular communication.

Findings

Maximum likelihood and least-squares estimators effectively estimate the channel.

Decision feedback equalizer structures mitigate ISI and ILI.

Time interleaving significantly reduces inter-link interference.

Abstract

In diffusion-based communication, as for molecular systems, the achievable data rate is low due to the stochastic nature of diffusion which exhibits a severe inter-symbol-interference (ISI). Multiple-Input Multiple-Output (MIMO) multiplexing improves the data rate at the expense of an inter-link interference (ILI). This paper investigates training-based channel estimation schemes for diffusive MIMO (D-MIMO) systems and corresponding equalization methods. Maximum likelihood and least-squares estimators of mean channel are derived, and the training sequence is designed to minimize the mean square error (MSE). Numerical validations in terms of MSE are compared with Cramer-Rao bound derived herein. Equalization is based on decision feedback equalizer (DFE) structure as this is effective in mitigating diffusive ISI/ILI. Zero-forcing, minimum MSE and least-squares criteria have been paired to DFE, and their performances are evaluated in terms of bit error probability. Since D-MIMO systems are severely affected by the ILI because of short transmitters inter-distance, D-MIMO time interleaving is exploited as countermeasure to mitigate the ILI with remarkable performance improvements. The feasibility of a block-type communication including training and data equalization is explored for D-MIMO, and system-level performances are numerically derived.