TY - JOUR
T1 - CSI-independent Non-linear Signal Detection in Molecular Communications
AU - Li, Bin
AU - Guo, Weisi
AU - Wang, Xiang
AU - Deng, Yansha
AU - Lan, Yueheng
AU - Zhao, Chenglin
AU - Nallanathan, Arumugam
PY - 2019/12/4
Y1 - 2019/12/4
N2 - Molecular communications rely on diffusive propagation to transport information, which is attractive for a variety of nano-scale applications. Due to the long-tail channel response, spatial-temporal coding of information may lead to severe inter-symbol interference (ISI). Classical linear signal processing in wireless communications is usually operating with high complexity and high signal-to-noise ratios, whereas signal processing in molecular communication system requires operating in opposite conditions. In this work, we propose a novel signal processing paradigm inspired by the biological principle, which enables low-complexity signal detection in extremely noisy environments. We first propose a non-linear filter inspired by stochastic resonance, which is found in a variety of biological systems, and it can significantly improve the output SNR by converting noise to useful signals. Then, we design a non-coherent detection method, one which exploits the generally transient trend of observed signals (i.e. quick-rising and slow-decaying) rather than hidden channel state information (CSI), thus excluding CSI estimation and involving only summations. Implementation issues are also discussed, including parameters configuration and adaptive threshold. Numerical results show that the proposed bio-inspired scheme can improve the performance remarkably over classical approaches. Even compared with the optimal linear methods, the required SNR of the proposed scheme can be reduced by 7 dB, which reaffirms why it can be used in noisy biological environments. As the first attempt to design bio-inspired molecular signal detectors, the proposed non-linear processing paradigm may provide the great promise to the emerging nano-machine applications.
AB - Molecular communications rely on diffusive propagation to transport information, which is attractive for a variety of nano-scale applications. Due to the long-tail channel response, spatial-temporal coding of information may lead to severe inter-symbol interference (ISI). Classical linear signal processing in wireless communications is usually operating with high complexity and high signal-to-noise ratios, whereas signal processing in molecular communication system requires operating in opposite conditions. In this work, we propose a novel signal processing paradigm inspired by the biological principle, which enables low-complexity signal detection in extremely noisy environments. We first propose a non-linear filter inspired by stochastic resonance, which is found in a variety of biological systems, and it can significantly improve the output SNR by converting noise to useful signals. Then, we design a non-coherent detection method, one which exploits the generally transient trend of observed signals (i.e. quick-rising and slow-decaying) rather than hidden channel state information (CSI), thus excluding CSI estimation and involving only summations. Implementation issues are also discussed, including parameters configuration and adaptive threshold. Numerical results show that the proposed bio-inspired scheme can improve the performance remarkably over classical approaches. Even compared with the optimal linear methods, the required SNR of the proposed scheme can be reduced by 7 dB, which reaffirms why it can be used in noisy biological environments. As the first attempt to design bio-inspired molecular signal detectors, the proposed non-linear processing paradigm may provide the great promise to the emerging nano-machine applications.
KW - CSI-independent
KW - Molecular communications
KW - non-coherent detector
KW - non-linear signal processing
KW - stochastic resonance
KW - transient feature
UR - https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8922790
UR - http://www.scopus.com/inward/record.url?scp=85077792725&partnerID=8YFLogxK
U2 - 10.1109/TSP.2019.2957636
DO - 10.1109/TSP.2019.2957636
M3 - Article
SN - 1053-587X
VL - 68
SP - 97
EP - 112
JO - IEEE Transactions on Signal Processing
JF - IEEE Transactions on Signal Processing
M1 - 8922790
ER -