Source localization for molecular communication via diffusion

Molecular communication is a type of communication that provides communication in mediums such as underwater, where traditional communication paradigms are insufficient, and in which information is carried by molecules in fluid environments. However, the fact that the information particles in molecular communication propagates by diffusion prevents us from solving problems such as finding the location of the transmitter with traditional methods. Especially, when the number of transmitters is more than one, the solution becomes more complicated. The solution described in this thesis is to find the locations of the transmitters by employing the coordinates of the molecules hitting the receiver in a fluid environment where there are multiple transmitters and a spherical receiver that absorbs the hitting molecules completely. For this localization solution, first the coordinates of the hitting molecules are clustered with models such as Gaussian, Bayesian mixture models, and K-means. By calculating the average values of the clustered data, the direction of the corresponding transmitter is determined. At the same time, the distance is determined based on the number of data belonging to separate clusters and the probability of the particles hitting the receiver. The results show that the most promising cluster algorithm is K-Means. By calculating the direction and the distance of the locations via clustered data, we can estimate the transmitter locations.