Abstract

You can view a living cell simply as a soup of chemical molecules, but a living cell can sense the environment, make decisions, repair itself, communicate with other cells and do many other amazing things. From a computing perspective, a cell can be thought of as a computation, communication and control device. Cells realise these functions by using networks of chemical reactions (which are also known as molecular circuits) and transport. Chemical reactions are inherently stochastic, so a research question in biology is to understand how cells use molecules and chemical reactions to compute and communicate. With the development of synthetic biology, there is now an opportunity to engineer novel molecular circuits and bio-nano devices, which are based on biological materials rather than silicon. With this background in mind, our research aims to develop a framework to understand, analyse and engineer molecular communication. In this seminar, we examine molecular communication from a Bayesian inference point of view. Inspired by temporal coding found in living cells, we formulate an optimal decoding problem with a pair of molecular circuit based transmitter and receiver. By modelling the end-to-end communication system with a reaction-diffusion master equation and by solving an optimal Bayesian filtering problem, we derive the optimal decoder in the form of a bank of analogue filters. An important insight from our solution is that molecular communication is event-based with information communicated through binding and unbinding events. We will also discuss the problems of signal and molecular circuit design.

This talk is predominantly based on these two publications: (1) A Markovian approach to the optimal demodulation of diffusion-based molecular communication networks. IEEE Trans. on Communications, Oct 2015. (2) Maximum a posteriori decoding for diffusion-based molecular communication using analog filters. IEEE Trans on Nanotechnology, Dec 2015.