Metric combinations in non-coherent signal detection for molecular communication

In emerging Internet-of-Nano-Things (IoNT) and molecular communication (MC), digital information will be embedded in chemical molecules in order to propagate through complex and absorbing media. However, two main challenges in signal detection are the intensive inter-symbol-interference (ISI) and the severe background noise, which may heavily deteriorate the reliability of molecular communications. This may be even worse when facing unknown channel impulse response (CIR), which is common in the industrial practice of MCs, rendering most of coherent signal detection schemes in traditional wireless communications less attractive. In this paper, we propose an improved non-coherent signal detection scheme, aiming at addressing ISI and noise contamination in the absence of CIR. Specifically, we utilized three non-coherent metrics to characterize transient features of received signals, and furthermore, designed an optimized combination scheme for more reliable signal detections. Given that the significance of each non-coherent metric remains unknown, we devise a self-adaption method premised on the concept of deflection coefficient, which can recursively determine the optimal combination weights. Numerical simulations demonstrate that our proposed scheme significantly improves the detection performance in terms of bit error rate (BER), and outperforms the adaptive threshold detection (ATD) method over 7 dB.