Robot Communication: Network Traffic Classification Based on Deep Neural Network

With the rapid popularization of robots, the risks brought by robot communication have also attracted the attention of researchers. Because current traffic classification methods based on plaintext cannot classify encrypted traffic, other methods based on statistical analysis require manual extraction of features. This paper proposes (i) a traffic classification framework based on a capsule neural network. This method has a multilayer neural network that can automatically learn the characteristics of the data stream. It uses capsule vectors instead of a single scalar input to effectively classify encrypted network traffic. (ii) For different network structures, a classification network structure combining convolution neural network and long short-term memory network is proposed. This structure has the characteristics of learning network traffic time and space characteristics. Experimental results show that the network model can classify encrypted traffic and does not require manual feature extraction. And on the basis of the previous tool, the recognition accuracy rate has increased by 8%

COMBINED CONTROL SYSTEM FOR THE OUTPUT OF A LINEAR PLANT WITH UN-KNOWN RELATIVE ORDER

We consider a system of combined control over the output of a linear structurally and parametrically indeterminate plant, synthesized within the framework of applying the hyperstability criterion and L-dissipativity conditions, as well as by using two filter correctors in a scheme with an implicit standard. The functioning of the plant is examined in the presence of external limited interference, when only its scalar input and output, and not their derivatives, are available for measurement

SUBOPTIMAL CONTROL OVER LINEAR NON-STATIONARY PLANT WITH DELAY

The article focuses on the fact that the problem of suboptimal control over parametrically and functionally undefined linear non-stationary plant has been solved. It is supposed that only scalar input and output of plant are determinable. The purpose of control consists in subminimization of the integral with an infinite top limit from square-law sub-integral function dependent on a scalar input of the plant and a control signal. The algorithm received is simple and does not require difficult analytical accounts of parameters of a control system. The serviceability of the received algorithms is illustrated on numerical examples.

Synthesis of Tracking Algorithms for Linear Objects with a Generative Model of the Reference Signal

In the article the class of linear stationary objects with a scalar input is considered. The purpose of control is formulated in the form of tracking the output of the control object for a given input reference. At the same time, the principle of the internal model is used to form a generating model of the input reference. The solution of the representation subtask with a predetermined finite precision of a time-discretized input reference in the form of a linear expansion on the basic functions corresponding to the roots of the desired characteristic polynomial of a discrete linear dynamic system (generator) is considered. By using the continualization, a continuous, linear generating model of the input reference with non-zero initial conditions and input is constructed in the state space, coinciding in dimension with the model of the control object. The generating model makes it possible to formulate the control goal in the form of tracking the state vector of a closed system with the state vector of the generating model. In general, the generating model may not be stable. Therefore, the desired rate of convergence of the tracking error vector is given by the Hurwitz reference model. The developed method of generating the generating model is considered in the context of solving the general problem of the synthesis of the tracking algorithm.

Modified Backstepping Algorithm and its Application to Control of Distillation Column

A novel robust algorithm is proposed for control of plants under parametric and structural uncertainties, as well as, external bounded disturbances. The algorithm design is based on the modified backstepping approach that allows to compensate mismatched disturbances under presence of nonlinearities. The obtained results are extended to control of network systems with nonlinear agents and with nonlinear links in the presence of mismatched disturbances. Effectiveness of the proposed algorithm is demonstrated on control of a distillation column which is described by parametric and structurally uncertain differential equation in presence of external bounded disturbances. It is assumed that only scalar input and output of the distillation column are available for measurement, but not their derivatives. The developed algorithm provides output tracking of a smooth bounded reference signal with a required accuracy at a finite time. The synthesis of control algorithm is separated into ρ steps, where ρ is an upper bound of the relative degree of the distillation column model. Therefore, the dynamical order of the proposed algorithm is equal to ρ. The sufficient conditions of the closed-loop stability is formulated and proved by using methods of stability of singular perturbed differential equations and Lyapunov functions. The simulations illustrate effectiveness of the proposed algorithm and confirm analytical results.

Adaptive Robust Stabilization of Rossler System With Time-Varying Mismatched Parameters Via Scalar Input

This paper presents an adaptive robust controller for a class of uncertain chaotic Rossler system with time-varying mismatched parameters. The proposed controller is designed based on Lyapunov stability theory, and it is shown that using this controller all signals of the closed-loop system are uniformly ultimately bounded (UUB). In addition, the proposed scheme is such that it does not require a priori information about the bound of uncertainties. Furthermore, since all the signals are UUB, the control signal is smooth and feasible to implement. Simulation results on a third-order Rossler system with time-varying parameters confirm the effectiveness of the proposed controller.