Suruz Miah

Defending unknown attacks on cyber-physical systems by semi-supervised approach and available unlabeled data

Abstract Cyber-physical systems (CPS) are used increasingly in modern industrial systems. These systems currently encounter a significant threat of malicious activities created by malicious software intent on exploiting the fact that the software of such industrial systems is integrated with hardware and network systems. Malicious codes dynamically and continuously change their internal structure and attack patterns using obfuscation techniques, such as polymorphism and metamorphism, in order to bypass and hide from conventional malware detection engines. This requires continuously updating the database of the malware detection engine, which requires periodic effort from manual experts. This could limit the real-time protection of CPS. In addition, this also makes preserving the availability and integrity of the services provided by CPS against malicious code challenging because there is a demand for the development of specialized malware detection techniques for CPS. In this paper, we propose a semi-supervised approach that automatically integrates the knowledge about unknown malware from already available and cheap unlabeled data into the detection system. The novelty of the proposed approach is that it does not require expert effort to update the database of the detection engine. Instead, the dynamic changes in malware attack patterns are extracted by unsupervised clustering from already available unlabeled data. Then the extracted geometric information about the intrinsic attack characteristics of the clusters is integrated into the classification systems of the detection engine, which updates the detection system automatically. The proposed approach uses global K-means clustering with term-frequency (TF), inverse document frequency (IDF), and cosine similarity as a distance measure for extracting the cluster information and adding it to a support vector machine (SVM) classification system. The proposed approach has been tested extensively on a real malware data set for both static and dynamic malware features. The experiment results show that the proposed semi-supervised approach achieves higher accuracy over the existing supervised approaches for all classifiers. We note that the static feature-based semi-supervised approach can improve detection accuracy significantly. While applying the proposed semi-supervised approach with the run-time characteristics of dynamic feature analysis, the combined effect of dynamic analysis and the proposed approach further increases the detection accuracy of all classifiers by up to a 100% for the SVM and the random forest classifiers, thus exceeding the existing supervised approaches with similar features.

Intelligent Parallel Parking of a Car-like Mobile Robot Using RFID Technology

This paper is devoted to design and implement an intelligent control scheme for a car-like mobile robot that possesses an automatic parallel parking capability using Radio Frequency IDen-tification (RFID) technology. The present manuscript focuses on the navigation module of the overall parking system. Navigation using a few analogue features of an RFID system is a promising alternative to a variety of existing navigation techniques in the state of the art. The proposed approach exploits the ability of a mobile robot to navigate in an unstructured environment and park itself parallel to a wall using RFID tags and some analogue features provided by the RFID system interfaced with the mobile robot. This paper describes how this is achieved by placing the RFID tags in unknown three dimensional positions on the wall within the workspace. A number of computer simulations are carried out to manifest the distinguished features of the proposed technique.

Nonuniform Deployment of Autonomous Agents in Harbor-Like Environments

We propose a nonuniform deployment strategy of a group of homogeneous autonomous agents in harbor-like environments. High value units berthed in the area need to be secured against external attacks. Defenders deployed in the area are expected to monitor, intercept, engage, and neutralize threats. In the framework of decentralized coordinated multi-agent systems, we model and simulate the optimal deployment of a group of mobile autonomous agents that accounts for a risk map of the area and the optimal trajectories that minimize the energy consumed to intercept a threat in a given area of interest. Theoretical results are numerically illustrated through simulations in a realistic harbor protection scenario.

Maximum power point tracking of wind turbines with neural networks and genetic algorithms

In the absence of aerodynamic pitch control, it is required to drive the wind turbine at an optimal speed for a given wind speed to extract maximum power from a wind turbine generator system. Due to unpredictable wind speed fluctuations, operating at maximum power point is a difficult task to undertake. This paper presents a maximum power point tracking (MPPT) algorithm for variable speed wind turbines. The strategy uses neural networks and genetic algorithms to learn the wind turbines nonlinear dynamic model and achieve accurate tracking. As such, robustness to unpredictable wind uncertainties is achieved. Simulation results for different situations highlight the performance of the proposed controller under various wind speed operating conditions.

Adaptive fuzzy logic control of a DC-DC boost converter with large parametric and load uncertainties

In this manuscript, we propose an adaptive fuzzy logic controller for a DC-DC boost converter with parametric and load uncertainties. The control strategy aims to achieve accurate voltage tracking with unknown dynamics, highly parameter and load variations, and no current sensing. Therefore, robustness to uncertainties of large magnitudes is achieved without the inner current control loop, which reduces the number of sensors. Simulations demonstrate the tracking performance of the proposed control technique in the presence of different intensities of uncertainties.

State-of-charge and state-of-health prediction of lead-acid batteries with genetic algorithms

In this paper, a state of charge (SoC) and state of health (SoH) estimator is presented for lead-acid batteries. The estimation strategy is based on adaptive control theory for online parameters identification. To speed up the estimators convergence, the adaptation law is replaced by a genetic algorithm (GA). Therefore, robustness to parameters variation is also achieved and thus, accurate prediction with battery aging. Unlike other estimation strategies, only battery terminal voltage and current measurements are required. Results show high convergence and highlight the performance of the proposed estimator in predicting the SoC and SoH with high accuracy.

Linear time-varying control law for stabilization of hopping robot during flight phase

The well-known Brockets theorem revealed that nonholonomic systems, hopping robots, for example, can not be stabilized by smooth time-invariant state feedback controllers. In this manuscript, we propose a linear time-varying state feedback controller for stabilizing a nonholonomic hopping robot during flight mode in finite time. The current approach is novel in the sense that we modify the Pontryagins minimum principle to formulate the linear state feedback control law. The existence of such a control law and its necessary conditions are presented in detail. The theoretical results are also validated through computer simulations.

Intelligent Range-Only Mapping and Navigation for Mobile Robots

Addressing the navigation (localization and motion control) problem of a mobile robot, coupled with its mapping problem, remains a significant challenge to date. The well-known simultaneous localization and mapping problem of mobile robots has been addressed in the literature without specifically taking into account the robots motion control tasks. Moreover, its implementation can cost more than the robot itself. Robot motion control strategies developed in the literature either 1) rely on sophisticated hardware platforms, 2) assume noise-free environment, or 3) are based on abstract theories, which are validated using computer simulations only. The current work solves the navigation and mapping problems of mobile robots using open-source hardware and range-only measurements from a network of radio sources. The hardware platform used in this work is customized such that it is cost-effective and easy-to-implement, addressing the aforementioned issues in developing motion control strategies. The current strategy permits the robot to estimate its position and orientation and builds a map of its operating environment using radio frequency signals received from radio sources. It then navigates through a path defined by a set of two-dimensional points on the ground using a motion control strategy in cooperation with a computational intelligence tool. The proposed robot navigation and mapping scheme was tested in an indoor laboratory environment and its performance was compared with the simulation counterpart using a commercial robot simulator.

Non-Autonomous Coverage Control With Diffusive Evolving Density

We consider nonuniform coverage optimization with respect to a non-autonomous coverage metric by spatially deploying a platoon of mobile agents in a planar region. Conventional coverage metrics usually encode a density field that weights points in the workspace. We consider a time-varying diffusive density that evolves according to a conservation law, and the induced time-varying coverage. Boundary conditions can model a time-varying flux across the boundary, and/or a time varying boundary density. We propose a decentralized state-feedback control law that maximizes the generalized non-autonomous coverage metric. The current approach of nonuniform deployment of autonomous agents applies to environmental monitoring and intervention, with deployment of mobile sensors in areas affected by penetration of substances governed by diffusion mechanisms, as for example oil in a marine environment, that pose immediate or long-term threats. We establish asymptotic convergence results illustrated by simulations.

Generalized non-autonomous metric optimization for area coverage problems with mobile autonomous agents

Motivated by area coverage optimization problems with time-varying risk densities, in this paper we propose a decentralized control law for a team of autonomous mobile agents in a 2-D area such that their asymptotic configurations optimize a generalized non-autonomous coverage metric. We emphasize that the generalized non-autonomous coverage metric explicitly depends on a nonuniform time-varying measurable scalar field that is defined by the trajectories of a set of mobile targets (distinct from the agents). The time-varying density that we consider here is not directly controllable by agents. We show that under certain conditions on the density defined on a closed bounded region of operation, the agents configure themselves asymptotically to optimize a related generalized non-autonomous coverage metric. A set of simulations illustrates the proposed control.