Impact of prediction techniques on tactile internet systems
Reconfigurable Computing, FPGA, Tactile Internet, Prediction, Latency, Real Time
In recent years, the development of systems with Internet interaction has improved massively. In this period, there were significant improvements in the quality of communication, which allowed for numerous advances, whether in domestic or industrial environments. These developments were notorious, but there is a barrier to advancing, the latency between the server and the client. This factor is proportional to the distance, the processing time, the complexity of the execution techniques, among others. This latency is negative for constraints applications such as Internet Tactile, which in this case requires the fidelity of qualified and/or tactile interaction. One way to minimize the implications of latency is to use prediction techniques on devices that integrate communication, that is, the master and slave devices. The techniques based on adaptive filters, artificial intelligence, passivity theory and dispersion theory, predictive control, predictors of Smith, among others. However, many other techniques, such as complex implementations and, in some cases, can increase the latency. Thus, this project targets to use systems based on reconfigurable hardware (RH), intrinsically parallel implementation, to improve performance this technique. Systems with general purpose processors (GPP), microcontrollers (uC) and DSPs are present in most applications. Comparative studies will be performed, such as computational complexities of advanced techniques associated with the Tactile Internet. The target is figure out how much prediction techniques can minimize the problem and, at the same time, minimize the latency added to the system after an implementation with RH. Finally, we will propose a factor that relates these two measures, as well as proposals for implementation on RH that minimize the execution time of the prediction techniques. In addition to the possibility of developing IP cores prediction techniques aimed at tactile internet applications.