Signal perception and processing with bioinspired sub-micro-systems
H.S. Abdel-Aty-Zohdt, Robert L. Ewing · 2004
Bio-inspired systems utilizing neural networks and genetic algorithms are presented in this paper for applications in pattern discovery, identification, classification, and optimum choices for Signal Perception and Processing (SPP). Sample overview applications are presented for: (I) Integrated Intelligent E-Nose Systems, and (II) Communication Systems dealing with: (1) Connection admission control; (2) Network congestion; (3) Resource management; (4) Priorities and constrains of service control; and (5) Application-specific network adaptation. Bio-inspired systems include: i- Self-Organizing feature Map for discovery, ii-Recurrent Dynamic Neural Networks (NNs), with output neurons feedback and feed forward arrays for noisy signals, iii- Reinforcement NNs for applications with only key features, rather than a known model, ivSpiking NNs that adjust their synapses subject to changes in the environment, and v- Genetic Algorithms for characterization and optimization. This paper summarizes alternative combinations and structures of novel bio-inspired, VLSIC embedded systems for detection and quantification of bio-chemical agents, and optimum performance of SPP.