The Method of Decoding Signals of Temperature Sensors of Communication Network Equipment Based on the Use of Nonlinear Volterra Integral Models

The paper considers the problem of control of temperature regimes of chips of computer network switching equipment. To do this, a temperature measuring system with a temperature sensor is used. When the temperature sensor is located inside the chip, the speed of the control system’s response to temperature changes is satisfactory. However, the system response speed is quite low when it is placed outside the chip due to thermal contact inertness. In such a case, the control system performance becomes unsatisfactory. To overcome the temperature sensor inertness, it is offered to perform the recovery of the signal of the non-linear subsystem "chip – temperature sensor" based on the use of its mathematical model in the form of a partial sum of the Volterra integro-degree series. T mathematical model of the mentioned subsystem is identified out within a finite time through a series of experiments using test signals. The paper considers a method of reducing the number of test signals based on taking into account the specifics of the nonlinearity influence on the experiment results. The obtained model is the basis for solving the inverse problem of recovering the temperature influence signal at the sensor input. Since this problem is incorrect, it is offered to complement the model with a regularization parameter and reduce the problem to a correct one. However, the obtained model can be used only within a finite time during which the kernels of the Volterra integro-degree series have been identified. To utilize the model within an infinite time, a computer simulation method is suggested using the computing process restarts made in several streams with their shift in time. The calculation result is formed by combining fragments from different streams. To check the reliability of the results obtained by applying the developed method, the solutions of model problems are given in the article.