Anomaly Detection in Molecular Communications With Applications to Health Monitoring Networks

Abstract

Early detection of diseases such as cancer plays a crucial role in their successful treatment. Motivated by this, anomaly detection in molecular nano-networks is studied. The proposed anomaly detection is, in fact, a two-tier network of artificial cells (ACs) in the first tier and a bio-cyber interface (BC) in the second tier. The ACs detect anomaly by variation in concentration of the biomarker molecules released by diseased cells (DCs) to the channel ending to ACs. This channel is modeled by a molecular communication (MC) paradigm. In the second tier, the ACs transmit a molecular message to a BC through the cardiovascular network, which is modeled again by the MC paradigm. A decision is made in the BC, which is implemented on the body skin, based on the received messages from different ACs. Due to the nature of the problem, a suboptimum design of detectors in the first and second tiers are provided. We use a Neyman-Pearson (NP) framework to analyze the detection performance of a health monitoring network. Based on bounding likelihood ratio (LR) a lower bound for the probability of detection and an upper bound for false alarm of each AC is derived. Next, taking into account the effect of ACs-BC channels, the overall performance of anomaly detection is analyzed in terms of probabilities of detection and false alarm.