Reconfigurable computing applied to reduce latency in control and prediction systems associated with tactile internet
Tactile Internet, Takagi - Sugeno Fuzzy-PI, FPGA, Multilayer Perceptron, Artificial Neural Networks, Hardware, Prediction
Tactile internet is the current technological advance for the Internet. The devices associated with this new internet paradigm will be responsible for man-machine applications with the sending of touch information and the other stimuli already sent. Thus, it is necessary to guarantee an extremely low latency between the devices that make up the tactile interaction. This latency is associated with propagating information through the communication channel, processing power of local devices, and complexity of the techniques being executed, among others. Accordingly, this work proposes using dedicated hardware-based reconfigurable computing to reduce latency in control and prediction systems applied to tactile Internet. Two approaches are proposed to address the problem of latency. In the first approach, reconfigurable hardware is proposed for an intelligent control system based on Fuzzy logic. The system is a Takagi - Sugeno Fuzzy-PI type controller that aims to reduce the latency associated with processing data using a tactile tool. The implementation uses a fully parallel strategy associated with a hybrid bit format scheme (fixed-point and floating-point). In the second approach, the implementation in reconfigurable hardware of linear and nonlinear prediction techniques is proposed. In the nonlinear case, a technique based on multilayer Perceptron artificial neural networks is used. In this approach, prediction techniques are used to minimize the impacts caused by delays and loss of information associated with tactile Internet. The proposals were tested for a field-programmable gate array (FPGA) on the Virtex 6 xc6vlx240t-1ff1156 platform. Data related to hardware occupation and throughput associated with the target platform are presented, and a comparison between results through simulation and implementations in dedicated hardware. The results are superior to those presented in other studies in the literature.