UNSW Canberra

Publication Search Results

Now showing 1 - 10 of 72
  • (2021) Keshk, Marwa
    Thesis
    With the rapid evolution of cyber attack techniques, the security and privacy of Cyber-Physical Systems (CPSs) have become key challenges. CPS environments have several properties that make them unique in efforts to appropriately secure them when compared with the processes, techniques and processes that have evolved for traditional IT networks and platforms. CPS ecosystems are comprised of heterogeneous systems, each with long lifespans. They use multitudes of operating systems and communication protocols and are often designed without security as a consideration. From a privacy perspective, there are also additional challenges. It is hard to capture and filter the heterogeneous data sources of CPSs, especially power systems, as their data should include network traffic and the sensing data of sensors. Protecting such data during the stages of collection, analysis and publication still open the possibility of new cyber threats disrupting the operational loops of power systems. Moreover, while protecting the original data of CPSs, identifying cyberattacks requires intrusion detection that produces high false alarm rates. This thesis significantly contributes to the protection of heterogeneous data sources, along with the high performance of discovering cyber-attacks in CPSs, especially smart power networks (i.e., power systems and their networks). For achieving high data privacy, innovative privacy-preserving techniques based on Artificial Intelligence (AI) are proposed to protect the original and sensitive data generated by CPSs and their networks. For cyber-attack discovery, meanwhile applying privacy-preserving techniques, new anomaly detection algorithms are developed to ensure high performances in terms of data utility and accuracy detection. The first main contribution of this dissertation is the development of a privacy preservation intrusion detection methodology that uses the correlation coefficient, independent component analysis, and Expectation Maximisation (EM) clustering algorithms to select significant data portions and discover cyber attacks against power networks. Before and after applying this technique, machine learning algorithms are used to assess their capabilities to classify normal and suspicious vectors. The second core contribution of this work is the design of a new privacy-preserving anomaly detection technique protecting the confidential information of CPSs and discovering malicious observations. Firstly, a data pre-processing technique filters and transforms data into a new format that accomplishes the aim of preserving privacy. Secondly, an anomaly detection technique using a Gaussian mixture model which fits selected features, and a Kalman filter technique that accurately computes the posterior probabilities of legitimate and anomalous events are employed. The third significant contribution of this thesis is developing a novel privacy-preserving framework for achieving the privacy and security criteria of smart power networks. In the first module, a two-level privacy module is developed, including an enhanced proof of work technique-based blockchain for accomplishing data integrity and a variational autoencoder approach for changing the data to an encoded data format to prevent inference attacks. In the second module, a long short-term memory deep learning algorithm is employed in anomaly detection to train and validate the outputs from the two-level privacy modules.

  • (2021) Zhang, Junpeng
    Thesis
    Satellite remote sensing videos have become an emerging valuable data source for city-scale surveillance from space. Moving Object Detection (MOD) and Multiple Object Tracking (MOT) methods serve as the stone-steps for the related applications. However, they are challenged by the low spatial resolution and low signal and noise ratio. This thesis aims to meet the challenges and develop effective MOD and MOT techniques for satellite remote sensing videos. For MOD, an Extended Low-rank and Structured Sparse Decomposition (E-LSD) model is proposed to suppress the effect of random noises. E-LSD models moving objects by a sparse foreground matrix, and the structured sparse regularization is imposed on it for exploring the spatial priors on moving objects. This alleviates the false alarm problem caused by noises on satellite remote sensing videos. To promote online processing, an Online Low-rank and Structured Sparse Decomposition (O-LSD) is developed. O-LSD reformulates the E-LSD problem that combines all frames in a video to a sequence of frame-wise separable counterparts by adopting the matrix factorization approximation and stochastic optimization techniques. In this way, extracting moving objects on a frame relies only on the information from current frame and its predecessor frames. In addition to random noises, the local misalignment caused by the motion of satellite platform is another primary source of false alarms in MOD on satellite remote sensing videos. To separate moving objects from it, a Moving-Confidence-assisted Matrix Decomposition (MCMD) model is developed by integrating motion information on moving objects into the foreground regularization. This is an improvement to E-LSD model in suppressing the effect of moving satellite platform in MOD. For MOT, an Incremental Successive Shortest Path (ISSP) tracker is developed. It defines a Maximum A-Posterior (MAP) problem for selecting and linking an optimal set of detection observations for trajectory formation. By fusing the information along the extracted trajectories, the proposed ISSP tracker reduces the fragmentations in the generated trajectories while automatically discarding the false alarms, which makes it more adaptive to the scenarios with inadequate accuracy. The videos captured by SkySat and Jilin-1 satellites were utilized for the testing and evaluation. The experimental results presented in this thesis demonstrate the proposed solutions perform well, which confirms the feasibility of applying satellite remote sensing videos for reliable MOD and MOT.

  • (2021) Huang, Qiuxiang
    Thesis
    Fluid-structure interactions (FSIs) of fluid-conveying collapsible tubes produce rich physiologically significant phenomena in many biological systems. Despite the significant progress made in recent years, the physical mechanisms responsible for the onset of self-excited oscillations of collapsible tubes remain unclear. To study nonlinear dynamic behaviors of collapsible tubes, a numerical framework based on the immersed boundary-lattice Boltzmann method (IB-LBM) for the simulation of FSI in collapsible tubes is developed, and then applied to investigate the physical mechanisms behind self-excited oscillations in a two-dimensional (2D) collapsible channel and a three-dimensional (3D) collapsible tube. In the proposed numerical framework, the lattice Boltzmann method (LBM) is employed to solve the fluid dynamics, while the structural equations for the 2D collapsible channel and 3D collapsible tube are solved by the finite difference method (FDM) and finite element method (FEM), respectively. The immersed boundary method (IBM) is adopted for the fluid and structural coupling. A power-law non-Newtonian fluid model is used to model the non-Newtonian fluid, and large eddy simulation (LES) is used to capture the turbulence in the FSI system. The nonlinear dynamics of a two-sided collapsible channel flow is first investigated. The stability of the hydrodynamic flow and collapsible channel walls are examined for a wide range of Reynolds number (100 < Re < 3000), structure-to-fluid mass ratio (0.3 < M < 20), external pressure (1 < Pe < 10) and wall thickness (0.01D < h < 0.1D). Chaotic behaviors of the collapsible channel flow are characterized and possible routes to chaos as well as physical mechanisms responsible for the onset of self-excited oscillations are identified. Nonlinear and rich dynamic behaviors of the collapsible system are newly observed. Specifically, the system experiences a supercritical Hopf bifurcation leading to period-1 limit cycle oscillations as the Reynolds number increases. The existence of chaotic behavior of the collapsible channel walls is confirmed by a positive dominant Lyapunov exponent and a chaotic attractor in the velocity-displacement phase portrait of the mid-point of the collapsible channel wall. Chaos in the system can be reached via period-doubling and quasi-periodic bifurcations. In addition, it is found that symmetry breaking is not a prerequisite for the onset of self-excited oscillations, but symmetry breaking induced by large mass ratio and external pressure may lead the system into a chaotic state. Unbalanced transmural pressure, wall inertia and shear layer instabilities in the vorticity waves have been found to induce the onset of the self-excited oscillations of the collapsible system. The period-doubling, quasi-periodic and chaotic oscillations are closely associated with vortex pairing and merging of adjacent vortices, and interactions between the upper and lower vorticity rows. The IB-LBM solver is applied to examine the nonlinear dynamics of unsteady flows in 3D collapsible tubes. The stability of the hydrodynamic flow and collapsible tube walls are examined for a range of Reynolds numbers (100 < Re < 1000) from laminar to turbulent. The effects of non-Newtonian rheology on the dynamic behaviors of the collapsible system are examined with a power-law model, and LES is used to model turbulent effects. For the effects of Reynolds number, it is found that the periodic vortex shedding downstream the throat of the elastic tube is responsible for small-amplitude and quasi-periodic self-excited oscillations developed at Re=200. During these oscillations, two regions of wall-thickening are developed near the end of downstream elastic tube (non-dimensional axial distance z/R=7.25) due to compressive stress concentrations, suggesting the potential for fatigue failures at these two regions. A small secondary buckling pattern develops at the downstream end of the elastic tube which has not been observed in the 2D model. The reverse-flow region is open and fills the entire symmetry plane with the major axis in the 3D model while it is closed and cannot fill the entire height of the channel in the 2D model. For the effects of non-Newtonian fluid, it is found that shear-thickening fluids tend to stabilize the collapsible system while shear-thinning fluids will trigger the onset of self-excited oscillations. The deformation of the tube decreases as the power-law index n increases. The tube snaps from a fully constricted state at n=1.6 to an unconstricted state at n=1.7. For turbulent flows at Re=1000, flow bifurcation occurs and the system settles into large-amplitude and quasi-periodic self-excited oscillations which are regular and repetitive. These oscillations are caused by the periodic shedding of vortices downstream of the throat of the elastic tube. The shed vortices feed back periodic perturbations through the elastic tube wall. At two monitor points placed in the downstream rigid tube, small secondary oscillations were found in the time history of pressure and streamwise velocity, which are caused by two jets merging together and interactions at the middle section of the downstream rigid tube.

  • (2021) Neilsen, Rhiannon
    Thesis
    In the contemporary digital age, mass atrocity crimes are increasingly promoted and organised online. Yet, little attention has been afforded to the question of whether proactive cyberspace operations might be used for human protection purposes. Beginning with the framework of the Responsibility to Protect (R2P), this thesis asks: How might cyber-operations be used ethically to protect populations from mass atrocity crimes? To answer this question, I introduce the concept of ‘cyber humanitarian interventions’, and argue that such measures can be used to disrupt potential perpetrators’ means and motivations for atrocities. Specifically, I contend that cyber humanitarian interventions can be used to frustrate potential perpetrators’ communication channels, logistical supply chains, and funding, as well as to stymie potential perpetrators’ desire for violence via online, targeted, tailor-made campaigns based on their big data. These capabilities can be used in an ethically acceptable manner, and thus ought to be pursued prior to the resort to other more forceful measures to protect. Moreover, and perhaps more controversially, I argue that, in some circumstances, there is a qualified responsibility to deceive potential perpetrators – via online disinformation – in order to fulfil responsibilities to protect. This thesis seeks to make three key contributions. First, it contributes to extant literatures on R2P, atrocity prevention, and cyberspace by offering cyber humanitarian interventions as a hitherto neglected tool for human protection. Second, it furthers ethical debates on atrocity prevention by providing an in-depth analysis of how cyber humanitarian interventions can be deployed ethically. Third, it challenges prevailing conceptions of disinformation by arguing that that there is, in fact, a qualified responsibility to deceive potential perpetrators into not committing atrocities via online disinformation. In sum, this thesis aims to bring 21st century capabilities to bear on centuries-old crimes, and highlights cyber humanitarian interventions as a more peaceful, cost-effective, and politically palatable tool to protect vulnerable populations from mass atrocity crimes.

  • (2021) Hassan Zadeh Koohi, Tahereh
    Thesis
    Medical image segmentation is a procedure to analyse an image’s content to find an organ, cancer, tumour, or possible abnormalities. Since hospitals and medical centres generate billions of images daily worldwide, manual analysis of the images is frustrating. Therefore there is a need to improve automatic techniques to examine the content of images. Deep Convolutional Neural Networks (DCNNs) are one of the most reliable and successful approaches to analyse images’ content. However, the main problem is a lack of rules to design a network, and trial and error is the usual approach to find a network structure along with its training parameters. Regarding the diversity of medical images, existing with various types of noises and artefacts, the limited number of available labelled medical images, and limited available computational facilities, designing a CNN for medical image analysis is even more complicated. Because of the importance of medical image segmentation, during the last decade, various CNNs are designed manually; however, most of these networks work well for the segmentation of a specific dataset or application. One of the solutions to address this problem is developing networks automatically. Neuroevolution, which is the combination of an evolutionary algorithm and Neural Networks (NNs), can automatically design a network. Evolutionary algorithms are relatively easy to understand and implement; however, they need considerable computation to evolve a network. Since Nerouevolution is computationally demanding, there is very limited previous work regarding applying Neuroevolution for medical image segmentation. Existing works just set up a part of the parameters to develop a network and have been applied to a limited number of datasets. The most significant drawbacks of existing works are lack of robustness and generalizability; also, most of them are computationally expensive. In this thesis, several Neroevolutionary-based frameworks are developed for 2D and 3D medical image segmentation. Firstly, a new block-based encoding model is developed to generate variable length 2D Deep Convolutional Neural Networks (DCNNs). The proposed encoding model could find appropriate values for several hyperparameters to create and train a DCNN. Also, a Genetic Algorithm (GA) is employed to evolve the generated networks. Besides, a comprehensive analysis is done to find an appropriate population size and generations, and consequently, an improved model is developed. In addition, to improve the results’ quality, an ensemble of found networks is utilised for final segmentation. Then to find a 3D evolutionary network, two approaches are examined. According to the proposed 2D model, a 3D model is developed to generate a population of 3D networks and evolve the 3D networks to find an appropriate 3D network for 3D medical image segmentation. Since evolving 3D networks is computationally expensive, a second approach is also introduced. In the second approach, the possibility of using a 2D evolutionary model to create a 3D network is examined and named Converted 3D network. Because of the diversity of medical images and the complexity of medical image analysis, sometimes more complicated CNN is needed. To address this issue, also another evolutionary model is developed in this thesis to generate more accurate and complex DCNNs using the combination of Dense and Residual blocks. In the proposed DenseRes model, a new encoding model is introduced, which is able to create a variable-length network with variable filter sizes within a block. In the DenseRes model, all required parameters to generate and train a network are included in the search. Most of the time, the Region Of Interest (ROI) is a small part of a medical image with almost the similar colour and texture of the surrounding organs. Therefore, more precise network architectures, like attention networks, are needed to process the images. To do so, two different approaches are introduced in this thesis to develop evolutionary attention networks. First, a 2D evolutionary attention model is proposed that is able to find an appropriate attention gate to transfer the block’s input to its output. Since some useful information will be lost during the downsampling in DCNNs, another 2D and 3D evolutionary attention framework is introduced to address this issue. In this model, besides creating a network structure along with its training parameters, an evolutionary algorithm is employed to find an appropriate model to recover and transfer feature maps from downsampling to the upsampling part of a network. The effectiveness of the proposed models is examined using various publicly available datasets. Results are compared with multiple manual and automatically designed models. The significant findings of this thesis can summarise as: (1) the proposed models obtain much better segmentation accuracy compared to state-of-the-art models, also, the proposed models are computationally cheap, even for developing 3D evolutionary networks; (2) converting a 2D evolutionary model to a 3D model is a reliable, fast, and accurate approach to create 3D networks; (3) including more constructive parameters in the search space can lead to more precise networks; (4) the initial population plays a significant role in the final results and decreasing training time; moreover, using variable filter sizes within a block can obtain better results compared to using a fixed one; (5) recovering a downsampling’s feature maps and transferring them to the corresponding upsampling part can considerably improve segmentation accuracy; (6) the proposed models are robust and general such that they can be applied for the segmentation of various medical images (CT and MRI) for different organs and tumour segmentation; (7) all the proposed encoding models are compatible with conventional crossover and mutation techniques, without any extra effort to create a new crossover technique or using a method to check the correctness of layers’ sequences.

  • (2021) Qiu, Huanneng
    Thesis
    The ability of a robot to adapt in-mission to achieve an assigned goal is highly desirable. This thesis project places an emphasis on employing learning-based intelligent control methodologies to the development and implementation of an autonomous unmanned aerial vehicle (UAV). Flight control is carried out by evolving spiking neural networks (SNNs) with Hebbian plasticity. The proposed implementation is capable of learning and self-adaptation to model variations and uncertainties when the controller learned in simulation is deployed on a physical platform. Controller development for small multicopters often relies on simulations as an intermediate step, providing cheap, parallelisable, observable and reproducible optimisation with no risk of damage to hardware. Although model-based approaches have been widely utilised in the process of development, loss of performance can be observed on the target platform due to simplification of system dynamics in simulation (e.g., aerodynamics, servo dynamics, sensor uncertainties). Ignorance of these effects in simulation can significantly deteriorate performance when the controller is deployed. Previous approaches often require mathematical or simulation models with a high level of accuracy which can be difficult to obtain. This thesis, on the other hand, attempts to cross the reality gap between a low-fidelity simulation and the real platform. This is done using synaptic plasticity to adapt the SNN controller evolved in simulation to the actual UAV dynamics. The primary contribution of this work is the implementation of a procedural methodology for SNN control that integrates bioinspired learning mechanisms with artificial evolution, with an SNN library package (i.e. eSpinn) developed by the author. Distinct from existing SNN simulators that mainly focus on large-scale neuron interactions and learning mechanisms from a neuroscience perspective, the eSpinn library draws particular attention to embedded implementations on hardware that is applicable for problems in the robotic domain. This C++ software package is not only able to support simulations in the MATLAB and Python environment, allowing rapid prototyping and validation in simulation; but also capable of seamless transition between simulation and deployment on the embedded platforms. This work implements a modified version of the NEAT neuroevolution algorithm and leverages the power of evolutionary computation to discover functional controller compositions and optimise plasticity mechanisms for online adaptation. With the eSpinn software package the development of spiking neurocontrollers for all degrees of freedom of the UAV is demonstrated in simulation. Plastic height control is carried out on a physical hexacopter platform. Through a set of experiments it is shown that the evolved plastic controller can maintain its functionality by self-adapting to model changes and uncertainties that take place after evolutionary training, and consequently exhibit better performance than its non-plastic counterpart.

  • (2021) Wang, Chenman
    Thesis
    The effectiveness of superplasticizers on the workability of fly ash-based geopolymer (FAG) pastes has been known to be far less obvious than in the case of cement pastes. An understanding of what happens to such pastes at the early stage of the reactions is extremely important for the manufacture of consistent and repeatable FAG concrete. This study investigates the effects of superplasticizers on FAG pastes by analysing the dispersion ability of superplasticizers, the properties of pore solutions, and the ongoing reaction processes. For the first time, the chemical reaction, which occurs between superplasticizers and Si, Al species on fly ash surface, was confirmed by using X-ray photoelectron spectroscopy (XPS). Meanwhile, this research has been pioneering the observation of the formation and evolution of gel material in the FAG pore solutions and presented analysis of that gel material. And also, in this pioneering study, in situ X-ray diffraction (XRD) was used to observe the phase transformation in the wet pastes with the presence of superplasticizers. Based on the above experimental work, a better understanding of the mechanisms that superplasticizers perform has been established. Four commercial superplasticizers including a lignosulfonate (LS), a polynaphthalene (PNS) and two polycarboxylates (PCE-1 and PCE-2) based superplasticizers were studied. Results show that absorption, bridging flocculation, decomposition and being trapped in the reactions are the main factors which resulted in the weak dispersion ability of the superplasticizers. Electrostatic repulsion and steric hindrance induced by the superplasticizers contributed to improving their dispersion ability. The reaction processes, including dissolution, inter-reaction, gelation and polymerisation, were also influenced by the superplasticizers. The superplasticizers promoted the dissolution of the fly ash in the activator solution. Most superplasticizers containing carboxyl groups probably reacted with dissolved Si, Al species, thus hindering the inter-reaction or the gelation. PNS, which contains no carboxyl groups, was the only superplasticizer that accelerated both the inter-reaction and the gelation processes. However, the polymerisation process was hindered by all the superplasticizers.

  • (2021) Pallithadathil Nazer, Muhammed Ajmal
    Thesis
    High-resolution optical spectroscopy in solids has attracted an enormous amount of attention over the past few years on the basis of a wide potential range of applications in the fields of communication, biomedical, defence and industrial applications. Transient spectral hole-burning, a laser-based spectroscopic technique, is used throughout this thesis to investigate the optical properties of the chromium(III) systems such as ruby (Al2O4:Cr3+) and alexandrite (BeAl2O4:Cr3+) crystals at their 4A2 to 2E spin-flip transitions (R1-line) at ~2 K. A spin-level pumping enhanced slow light based on the transient spectral hole-burning is reported. The frequency locking and stabilization of an external-cavity diode laser using the Pound-Drever-Hall technique is described. Time-resolved transient spectral hole-burning is performed in the R1-line of different ruby samples in zero field and low magnetic fields at ~2.2 K. The generation of slow light in ruby samples was then investigated using the spectral hole-burning technique. Slow light is caused by the spectral hole in the absorption spectrum of the crystal leading to a strong frequency-dependent normal dispersion, resulting in the delay in pulse propagating through the crystal. An external magnetic field further reduces the width of the spectral hole by reducing the electronic spin-flip rates and even slower light is then observed. A spin-level pumping results in an even slower group velocity of the pulse propagating through the ruby sample when compared without the pumping process. A strong pump pulse applied to the R1(±1/2) line modifies the optical density of the R1(±3/2) line, with the hole-burning performed at the R1(±3/2) line. In addition, the spin-lattice relaxation time of ruby is accurately measured. High-resolution laser transmission measurements are conducted at the R1m-line of alexandrite at 2.1 K, and its strong luminescent properties are studied by fluorescence and fluorescence decay measurements. Polarized time-resolved transient spectral hole-burning is conducted at the R1m-line of alexandrite at 2.1 K in zero and low magnetic fields BIIa, BIIb and BIIc (B < 12 mT). The Zeeman-level splitting, proportional to the applied field strength, is directly observed from the hole patterns and this allows the accurate determination of the g-factor in the excited state.

  • (2021) Kabir, Md Imran
    Thesis
    Recently, structural applications of the High Strength Steel (HSS) sections have gained widespread attention among the engineers as their remarkably high yield strength reduces the sizes of the structural components and allows engineers to design and build elegant and iconic structures. While the use of HSS offers the flexibility of designing thinner sections, it comes with the drawback that the slenderness of the steel section will be increased significantly. Therefore, when comparing with a normal strength I–beam, a laterally unrestrained HSS I–beam, depending on the slenderness of the section, is more vulnerable to both local buckling and lateral torsional buckling (LTB) as well as their combined effects. As a result, the HSS I–beam is more likely to fail by LTB and/or local buckling before the section develops its full plastic moment capacity and thereby, the advantages of using the HSS section could not be fully utilized. In order to prevent the onset of these structural instabilities, this research introduces a new type of Engineered Cementitious Composites (ECC)–Light Weight Concrete (LWC) encased steel composite beams. The main focus of this research was to investigate the flexural behaviour of the proposed beams experimentally, numerically, and analytically. For the experimental study, fifteen full–scale beams of different encasement configurations, steel grades, and steel classes were prepared and tested under four–point bending. The performances of the beams were investigated in terms of load–deflection behaviour, failure modes, load–strain responses, bond–slip behaviour, and crack propagation history. For the numerical study, a three–dimensional nonlinear finite element (FE) model was developed using ABAQUS to simulate the flexural behaviour of the proposed beam by considering both material and geometric nonlinearities. The developed FE model was first validated against the test results. It was found to agree well with the test results in terms of overall load–deflection response, initial stiffness, failure modes, peak–load and corresponding deformation, and post–peak response. The validated FE model was then adopted for a parametric study. Finally, a detailed analytical investigation was carried out to determine the flexural strength of the proposed beams.

  • (2021) Mahmud, Md Rasel
    Thesis
    The conventional power system undergoes a massive change owing to the penetration of distributed generators into distribution networks and the standalone operation of microgrids. The characteristics of distributed generators largely depend on power electronic interfaces that are not matched with conventional synchronous machine-based generation systems. The changing behaviours of the distribution side of the electrical network continue to provide new challenges to power system design and operation engineers. To overcome such issues, the design of high-performance controllers plays a crucial role. In addition, investigation of the dynamic behaviours of the power electronic interfaces, including switching frequency is essential, which is the critical characteristic of the distributed generators. The first contribution of this dissertation is the design of a robust nonlinear controller to enhance the transient stability of distributed generators. In this work, the feedback linearization technique is used to linearize the standalone solar photovoltaic including battery energy storage, grid-connected fuel cell, and grid-connected solar photovoltaic including battery energy storage systems. Robust H_\infty mixed-sensitivity loop-shaping controllers are designed to stabilize the linearized part of the nonlinear feedback linearized control laws. The proposed robust H_\infty controller for the feedback linearized control scheme provides several benefits over the existing feedback linearization that is stabilized by the conventional PI controller. The second and unique contribution of this thesis is the developed single-input two-output feedback linearized control scheme based on the conventional single- input single-output feedback linearization approach. The third contribution of this dissertation is single-input two-output controller design for the islanded DC, AC, and hybrid DC/AC microgrids to enhance the voltage stability and minimize the circulating current among parallel-connected distributed generators. The fourth and final contribution of this work is the developed multi-frequency averaging based dynamic phasor model of a standalone, grid-connected, and two-stage distributed photovoltaic generators. The developed dynamic phasor models are provided with switching frequency-dependent characteristics, showing the switching frequency effect on the distributed generators. This contribution provides the idea of a suitable switching frequency range for power electronic interfaces.