Efficient Digital Realization of Endocrine Pancreatic β-Cells

The accurate implementation of biological neural networks, which is one of the important areas of research in the field of neuromorphic, can be studied in the case of diseases, embedded systems, the study of the function of neurons in the nervous system, and so on. The pancreas is one of the main organs of human that performs important and vital functions in the body. One part of the pancreas is an endocrine gland and produces insulin, while another part is an exocrine gland that produces enzymes for digesting fats, proteins and carbohydrates. In this paper, an optimal digital hardware implementation for pancreatic β-cells, which is the endocrine type, is presented. Since the equations of the original model include nonlinear functions, and the implementation of these functions results in greater use of hardware resources as well as deceleration, to achieve optimal implementation, we have approximated these nonlinear functions using the base-2 functions and LUT. The results of dynamic analysis and simulation show the accuracy of the proposed model compared to the original model. Analysis of the synthesis results of the proposed model on the Spartan-3 XC3S50 (5TQ144) reconfigurable board (FPGA) shows the superiority of the proposed model over the original model. These advantages include using fewer hardware resources, a performance almost twice as fast, and 19% less power consumption, than the original model.