Field-Programmable Gate Array Implementation of Adaptive Neuro-Fuzzy System Using Sensors Monitoring Health-Care Medicinal Internet of Things

Abstract

In this study an artificial intelligent algorithm that can be used for monitoring health-care MIoT (Medicinal Internet of Things) and predicting system based on Adaptive Neuro-Fuzzy Inferences System Architecture (ANFIS) is proposed. We contribute with a new modification for ANFIS architecture and implement it in Field-programmable Gate Array (FPGA) using High-Level Synthesis (HLS) approach for monitoring predicting temperature and humidity. The proposed modification for intelligent algorithm is done by extending the ANFIS standard architecture to six-layer adaptive instead of five-layer in order to minimize the number of linear parameters that need to adapt in the defuzzification output layer and hardware utilization resources that used within the FPGA environment. The performance of proposed architecture has been evaluated and tested in term of mean square error between the real outputs of the modified algorithm (that are taken from hardware ANFIS-IP core) and the desired targets (optimal outputs that are taken from Matlab simulation). The modifying architecture provides a high precision in the training phase and acceptable precision in the testing phase when compared with a standard Matlab toolbox. While the number of hardware resources within our proposed embedded system are decreased by 55% when compared with other works that untiled the same approach.