Investigation of Model-based Parameter Estimation Algorithms for Molecular Communication Systems

Molecular communications (MC) is an emerging research field, which aims to enable communication in challenging environments, for example inside the human body for smart drug delivery. In contrast to conventional wireless communications, MC employs molecules for information transmission, which makes this approach highly bio-compatible and energy-efficient. The analysis and design of MC systems addresses important issues in bio-medical applications such as targeted drug delivery as well as in industrial monitoring applications. Next to theoretical studies, a number of practical MC testbeds has been developed.

A permanent challenge is the match between theoretical (analytical, numerical) models and reality or realistic testbeds. One option to arrive at realistic models is the estimation of channel parameters by the application of learning algorithms to measurements obtained from, e.g., real-world systems or testbeds. After the estimation, these parameters can be used to feed a numerical or analytical system model for testing and for the derivation of communication protocols. Besides data-driven estimation algorithms, model-based estimation algorithms are a promising approach.

In preliminary works, the application of Kalman filters was shown to yield powerful algorithms for the estimation of parameters of MC channels. So far, the estimation algorithms only have been applied to synthetic measurements and the underlying MC channel models does not capture all effects occurring in a real system. Based on these preliminary works, the goal of the research internship is the extension of existing parameter estimation algorithms and the investigation of other algorithms, e.g., unscented Kalman filters and particle filters.


  • Getting started: The previously derived parameter estimation algorithm based on Kalman filters should be studied.
  • Extension: The existing Kalman filter model should be extended and analyzed further. Particularly, the underlying channel model should be extended to a laminar flow channel and the algorithm should be analyzed for its ability to estimate time-varying parameters.
  • Literatur research: A literature research is to be carried out to identify other suitable parameter estimation algorithms. The focus remains on model-based estimation algorithms. Promising techniques are unscented Kalman filters and particle filters, which are strongly related to classical Kalman filtering but provide several advantages.
  • Implementation: The investigated estimation algorithms should be implemented. Their performance should be analyzed and compared to each other.
  • Application: The estimation algorithms should be applied for the estimation of relevant parameters in MC channels, i.e., estimation of physical parameters, distance estimation between (possibly mobile) transmitter and receiver, etc.