Mission-critical Applications Research Articles

A Novel Approach for Model Interpretability and Domain Aware Fine-Tuning in AdaBoost

AbstractThe success of machine learning in real-world use cases has increased its demand in mission-critical applications such as autonomous vehicles, healthcare and medical diagnosis, aviation and flight safety, natural disaster prediction, early warning systems, etc. Adaptive Boosting (AdaBoost) is an ensemble learning method that has gained much traction in such applications. Inherently being a non-interpretable model, the interpretability of the AdaBoost algorithm has been a research topic for many years. Furthermore, most of the research being conducted till now is aimed at explaining AdaBoost using perturbation-based techniques. The paper presents a technique to interpret the AdaBoost algorithm from a data perspective using deletion diagnostics and Cook’s distance. The technique achieves interpretability by detecting the most influential data instances and their impact on the feature importance of the model. This interpretability enables domain experts to accurately modify the significance of specific features in a trained AdaBoost model depending on the data instances. Unlike explaining AdaBoost using perturbation-based techniques, interpreting from a data perspective will enable it to debug data-related biases, errors and to impart the knowledge of the domain experts into the model through domain aware fine-tuning. Experimental studies were conducted with diverse real-world multi-feature datasets to demonstrate interpretability and knowledge integration through domain-aware fine-tuning.

A method for evaluating deep generative models of images for hallucinations in high-order spatial context

Deep generative models (DGMs) have the potential to revolutionize diagnostic imaging. Generative adversarial networks (GANs) are one kind of DGM which are widely employed. The overarching problem with deploying any sort of DGM in mission-critical applications is a lack of adequate and/or automatic means of assessing the domain-specific quality of generated images. In this work, we demonstrate several objective and human-interpretable tests of images output by two popular DGMs. These tests serve two goals: (i) ruling out DGMs for downstream, domain-specific applications, and (ii) quantifying hallucinations in the expected spatial context in DGM-generated images. The designed datasets are made public and the proposed tests could also serve as benchmarks and aid the prototyping of emerging DGMs. Although these tests are demonstrated on GANs, they can be employed as a benchmark for evaluating any DGM. Specifically, we designed several stochastic context models (SCMs) of distinct image features that can be recovered after generation by a trained DGM. Together, these SCMs encode features as per-image constraints in prevalence, position, intensity, and/or texture. Several of these features are high-order, algorithmic pixel-arrangement rules which are not readily expressed in covariance matrices. We designed and validated statistical classifiers to detect specific effects of the known arrangement rules. We then tested the rates at which two different DGMs correctly reproduced the feature context under a variety of training scenarios, and degrees of feature-class similarity. We found that ensembles of generated images can appear largely accurate visually, and show high accuracy in ensemble measures, while not exhibiting the known spatial arrangements. The main conclusion is that SCMs can be engineered, and serve as benchmarks, to quantify numerous per image errors, i.e., hallucinations, that may not be captured in ensemble statistics but plausibly can affect subsequent use of the DGM-generated images.

A novel lightweight multi-factor authentication scheme for MQTT-based IoT applications

The present authentication solutions employed in the Internet of Things (IoT) are either inadequate or computationally intensive, given the resource-constrained nature of IoT devices. This challenges the researchers to devise efficient solutions to embed an important security tenet like authentication. In IoT, the most popular machine-to-machine communication protocol used at the application layer is Message Queuing Telemetry Transport (MQTT). However, the MQTT protocol inherently lacks security-related functions, like authentication, authorization, confidentiality, access control, and data integrity, which is unacceptable for IoT-driven mission-critical applications when connected over public networks. In such a situation, the security is hardened by employing a transport layer security protocol like TLS, which entails significant computational overheads. This paper presents a novel scheme to enhance MQTT security by providing a lightweight multi-factor authentication scheme based on Elliptical curve cryptography. The proposed scheme uses a low-cost signature and a fuzzy extractor to correct errors in imprinted biometrics in noisy environments. This scheme attains mutual authentication, generates a securely agreed-upon session key for secret communication, and guarantees perfect forward secrecy. Furthermore, the rigorous informal security analysis shows the proposed scheme resists cryptographic attacks, including known session critical attacks. Furthermore, an empirical study has been carried out to assess the effectiveness of the proposed scheme in the Cooja simulated environment.

Dynamic Bias-Temperature Instability Testing in SiC MOSFETs

For power converter development in mission critical applications, the attractive performances of SiC power MOSFETs are shadowed by reliability concerns, particularly those induced by the defects at the gate dielectric. Charge trapping at the oxide-semiconductor interface can lead to threshold voltage drift, degrading the power converter efficiency and lifetime. The scope of this contribution is to show a testing methodology under development to understand SiC power MOSFET threshold voltage stability under dynamic and accelerated operating conditions. The presented testing methodology relies on switching the device under test at high-voltage and current, simultaneously applying a gate stress and extracting threshold voltage from switching transients. The paper outlines the setup description, its operating modes and intended design of experiment to assess SiC MOSFET threshold voltage stability.

Cybersecurity for tactical 6G networks: Threats, architecture, and intelligence

Edge intelligence, network autonomy, broadband satellite connectivity, and other concepts for private 6G networks are enabling new applications for public safety authorities, e.g., for police and rescue personnel. Enriched situational awareness, group communications with high-quality video, large scale IoT, and remote control of vehicles and robots will become available in any location and situation. We analyze cybersecurity in intelligent tactical bubbles, i.e., in autonomous rapidly deployable mobile networks for public safety operations. Machine learning plays major roles in enabling these networks to be rapidly orchestrated for different operations and in securing these networks from emerging threats, but also in enlarging the threat landscape. We explore applicability of different threat and risk analysis methods for mission-critical networked applications. We present the results of a joint risk prioritization study. We survey security solutions and propose a security architecture, which is founded on the current standardization activities for terrestrial and non-terrestrial 6G and leverages the concepts of machine learning-based security to protect mission-critical assets at the edge of the network.

Reliability estimation for individual predictions in machine learning systems: A model reliability-based approach

The conventional aggregated performance measure (i.e., mean squared error) with respect to the whole dataset would not provide desired safety and quality assurance for each individual prediction made by a machine learning model in risk-sensitive regression problems. In this paper, we propose an informative indicator ℛx to quantify model reliability for individual prediction (MRIP) for the purpose of safeguarding the usage of machine learning (ML) models in mission-critical applications. Specifically, we define the reliability of a ML model with respect to its prediction on each individual input x as the probability of the observed difference between the prediction of ML model and the actual observation falling within a small interval when the input x varies within a small range subject to a preset distance constraint, namely ℛx=Py∗−ŷ∗≤εx∗∈Bx, where y∗ denotes the observed target value for the input x∗,ŷ∗ denotes the model prediction for the input x∗, and x∗ is an input in the neighborhood of x subject to the constraint Bx=x∗x∗−x≤δ. The developed MRIP indicator ℛx provides a direct, objective, quantitative, and general-purpose measure of “reliability” or the probability of success of the ML model for each individual prediction by fully exploiting the local information associated with the input x and ML model. Next, to mitigate the intensive computational effort involved in MRIP estimation, we develop a two-stage ML-based framework to directly learn the relationship between x and its MRIP ℛx, thus enabling to provide the reliability estimate ℛx for any unseen input instantly. Thirdly, we propose an information gain-based approach to help determine a threshold value pertaing to ℛx in support of decision makings on when to accept or abstain from counting on the ML model prediction. Comprehensive computational experiments and quantitative comparisons with existing methods on a broad range of real-world datasets reveal that the developed ML-based framework for MRIP estimation shows a robust performance in improving the reliability estimate of individual prediction, and the MRIP indicator ℛx thus provides an essential layer of safety net when adopting ML models in risk-sensitive environments.

Operational Amplifiers Defect Detection and Localization Using Digital Injectors and Observer Circuits

Operational amplifiers (op amps) are fundamental blocks that find wide application both as stand-alone devices and as crucial blocks embedded in various Systems on Chips (SoCs). Achieving high defect coverage, as well as performing defect localization in these circuits, has proven to be a difficult/expensive task, even with sophisticated testing circuitry. The ISO 26262 standard for functional safety (FuSa) includes the stringent requirement that an automotive IC must have a very high defect coverage. This reinforces the need to ensure the functionality of analog and mixed (AMS) circuits, especially in mission critical applications. This paper presents an all-digital op amp defect detection, diagnosis, and localization method that can be used both for production and in-field tests and discusses various implementation of the proposed method. We validated our results using extensive transistor-level simulations of multiple op amp architectures using TSMC 180 nm technology. Across op amp architectures and multiple implementation approaches, we achieved a worst-case and best-case defect coverage of 94.5% and 99%, respectively. Furthermore, in this work, we also propose a defect diagnosis and localization strategy using recorded bit streams from states of digital injectors and detectors.

Reliable and Delay Efficient Multi-Path RPL for Mission Critical IoT Applications

Reliable and Delay Efficient Multi-Path RPL for Mission Critical IoT Applications

Experimental analysis for comparison of wireless transmission technologies: Wi-Fi, Bluetooth, ZigBee and LoRa for mobile multi-robot in hostile sites

This research paper conducts a thorough comparison of four prominent transmission technologies suitable for mobile robots operating in challenging environments. Emphasizing key factors such as signal strength, noise resistance, and data transfer efficiency, the study aims to identify the optimal communication solution in hostile conditions. The exploration delves into the intricacies of received signal strength indication (RSSI) and signal-to-noise ratio (SNR), revealing distinctive traits and trade-offs among the technologies. Navigating through the complexities of frequency bands, modulation types, and communication topologies, the paper examines the impact of obstacles, energy consumption dynamics, and potential real-world applications. Beyond contributing to the fields of robotics and communication, the study offers practical insights for stakeholders seeking resilient and efficient transmission methods for mobile robotic applications. Advocating for long range (LoRa) as the preferred transmission technology in hostile environments, the paper highlights its unmatched immunity to noise, stability, and minimal energy consumption. These findings provide valuable guidance for technology choices in collaborative mobile robot operations under challenging conditions. This research sets the stage for future developments in robotic communication, underscoring the crucial role of selecting the right transmission means for mission-critical applications in hostile environments.

Algebraic Speed Estimation for Sensorless Induction Motor Control: Insights from an Electric Vehicle Drive Cycle

Induction motors (IMs) must meet high reliability and safety standards in mission-critical applications, such as electric vehicles (EVs), where sensorless control strategies are fundamental. However, sensorless rotor speed estimation demands improvements to overcome filtering distortions, tuning complexities, and sensitivity to IM model mismatch. Algebraic methods offer inherent filtering capabilities and design flexibility to address these challenges without introducing additional dynamics into the control system. The objective of this paper is to provide an algebraic estimation strategy that yields an accurate rotor speed estimate for sensorless IM control. The strategy includes an algebraic estimator with single-parameter tuning and inherent filtering action. We propose an EV case study to experimentally evaluate and compare its performance with a typical drive cycle and a dynamic torque load that emulates a small-scale EV power train. The algebraic estimator exhibited a signal-to-noise ratio (SNR) of 43 dB. The closed-loop experiment for the EV case study showed average tracking errors below 1 rad/s and similar performance compared to a well-known sensorless strategy. Our results show that the proposed algebraic estimation strategy works effectively in a nominal speed range for a practical IM sensorless application. The algebraic estimator only requires single-parameter tuning and potentially facilitates IM model updates using a resetting scheme.

A novel trusted hardware-based scalable security framework for IoT edge devices

The Internet of Things (IoT) devices are pervasively deployed and embedded into our daily lives. Over several years, the massive assimilation of IoT devices has given rise to smart cities, smart factories, smart farms, etc. While IoTs can collect immense amounts of user and environmental information, the lack of adequate security measures is among the major concerns about information security and privacy. With the ever-increasing presence of IoTs, information flow between IoT devices in a single or multi-layered network is becoming increasingly vulnerable to attacks. Although several software-based solutions have been developed over the years that improve the security of resource-constrained IoT devices, there are limited efforts on hardware-oriented IoT security solutions. This paper proposes a Trusted HArdware-based Scalable Secure Architecture (THASSA) for IoT Edge systems. Leveraging standard cryptographic algorithms, THASSA provides a secure mechanism for data transfer and off-loads cryptography operations to trusted hardware. The modularized THASSA architecture allows devices to be added or removed without user intervention. Also, THASSA could be scaled to multiple subsystems using multiple trusted hardware to form IoT clusters with secure data path(s) for mission-critical applications. An experimental study has verified the correctness and effectiveness of THASSA.

Non-invasive real-time diagnosis of PMSM faults implemented in motor control software for mission critical applications

The machine’s health should be continuously monitored, or a test should be applied on a regular basis to predict failure before fatal damage is incurred. Unexpected failures, particularly in mission-critical applications, can cause irreversible damage to the system, or even human life. This paper introduces a novel non-intrusive, real-time, online Condition Monitoring (CM), and Fault Diagnosis (FD) system for Permanent Magnet Synchronous Machines (PMSMs). Only the motor drive’s built-in sensors, such as current and position sensors, are used to detect three types of faults: inter-turn short circuit, partial demagnetization, and static eccentricity. It encompasses the implementation of algorithms within a motor drive system and the creation of failure mode models. The proposed solution adopts a hardware-free approach, utilizing current/voltage signature analysis for cost-effectiveness. It requires a small memory and short execution time, allowing it to be implemented on a simple motor controller with limited memory and calculation power. The drive system is intended for mission critical applications, therefore, computation load, code size, memory allocation, run-time optimization, etc. are the key focuses for real-time operation. It offers immediate insights into motor’s health without interrupting the drive operation. Additionally, it ensures rapid processing with modest computational requirements, making it adaptable for implementation on any PMSM controller. The non-intrusive nature of this diagnostic approach has the potential to enhance safety in systems reliant on PMSM drives. The proposed method has a high detection accuracy of 98%, is computationally efficient and can detect and classify the fault accurately. Simulation and experimental results demonstrate the efficiency of the FD algorithm for online identification and classification of machine faults. Theoretical hypotheses are proven based on experimental data.

Adaptive boosting with fairness-aware reweighting technique for fair classification

Machine learning methods based on AdaBoost have been widely applied to various classification problems across many mission-critical applications including healthcare, law and finance. However, there is a growing concern about the unfairness and discrimination of data-driven classification models, which is inevitable for classical algorithms including AdaBoost. In order to achieve fair classification, a novel fair AdaBoost (FAB) approach is proposed that is an interpretable fairness-improving variant of AdaBoost. We mainly investigate binary classification problems and focus on the fairness of three different indicators (i.e., accuracy, false positive rate and false negative rate). By utilizing a fairness-aware reweighting technique for base classifiers, the proposed FAB approach can achieve fair classification while maintaining the advantage of AdaBoost with negligible sacrifice of predictive performance. In addition, a hyperparameter is introduced in FAB to show preferences for the fairness-accuracy trade-off. An upper bound for the target loss function that quantifies error rate and unfairness is theoretically derived for FAB, which provides a strict theoretical support for the fairness-improving methods designed for AdaBoost. The effectiveness of the proposed method is demonstrated on three real-world datasets (i.e., Adult, COMPAS and HSLS) with respect to the three fairness indicators. The results are accordant with theoretic analyses, and show that (i) FAB significantly improves classification fairness at a small cost of accuracy compared with AdaBoost; and (ii) FAB approach can achieve low levels of fairness loss while maintaining high accuracy compared with state-of-the-art fair classification methods including equalized odds method, exponentiated gradient method, grid search reduction method and disparate mistreatment method.

Statistical Methods for Detecting Recycled Electronics: From ICs to PCBs and Beyond

<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Globalization of semiconductor industry has led to unprecedented challenges in design, verification, test, security and trustworthiness of contemporary electronics. With parts sourced from various suppliers across the globe, who are beyond the auspices of the original designers and system integrators, serious concerns arise regarding integrity and trustworthiness of electronic circuits, which can be compromised by malicious actors at any stage of the component manufacturing and system assembly process. Among the variety of such concerns, recycled electronics sold as new by malicious suppliers/vendors pose a key threat to mission-critical end applications such as military systems, financial institutions, transportation security, and poser distribution infrastructure. Recycled electronics may experience performance degradation due to aging and, more importantly, may have shorter-than-expected lifetime, thereby raising serious reliability and safety issues. To combat this problem, statistical methods have been proposed in the past decade as a time-efficient and cost-effective approach to identify recycled electronics sold as new. Initially demonstrating in the context of detecting recycled Integrated Circuits (ICs), such methods are based on using statistics and machine learning to outline the parametric signatures of known brand-new ICs and, thereby, weed out ICs that do not fit the expected profile. This article summarizes the current state of knowledge in recycled IC detection schemes and discusses how these methods have inspired, over time, the development of similar solutions in a plethora of domains, including identification of recycled Field Programmable Gate Arrays (FPGAs), recycled hardware components in a network, and recycled Static Random-Access Memories (SRAMs)/Flash memories</i> .

The concept of network resource control of a 5G cluster focused on the smart city's critical infrastructure needs

The article proposes a concept of organizing access to the guaranteed resource of the 5 G cluster information environment, structured in the form of a hierarchy of logical network segments. This framework presents an interrupt access control scheme that exhibits survivability characteristics important to mission-critical applications. Analytically, the author's concept is represented by an irreversible Markov process, for which a system of equilibrium equations and stationary probabilities are defined. Based on the latter, a qualitative metric was formulated, which included such indicators as the average number of requests accepted for service in the 5 G cluster, the probability of losing an incoming request, and the average volume of network resource units used in the 5 G cluster. For the single input data that characterizes a real NS 5 G cluster, a comparison of the author's and known competing access control schemes in terms of formulated qualitative metrics is implemented. The obtained results showed that the author's concept prevails over the competitor in the range of low loads. Note that in this range, the advantage of the author's scheme over the competitor significantly increases with the increase in the number of network segments into which the information environment of the 5 G cluster is divided.

Penetralium: Privacy-preserving and memory-efficient neural network inference at the edge

The proliferation of artificial intelligence and edge computing has led to an increase in the deployment of proprietary deep learning models on third-party edge servers or devices to power mission-critical applications. However, this trend raises concerns about model privacy, particularly on untrusted edge platforms. Protecting model privacy in such scenarios requires addressing challenges such as untrustworthy model deployment environments, resource-constrained Trusted Execution Environments (TEE), and vulnerability to privacy inference attacks. To address these challenges, this paper proposes Penetralium, a system-algorithm jointly optimized model inference system on edge computing platforms. Penetralium runs models in the TEE by building an underlying computational engine. We propose an adaptive decomposition algorithm that builds a computing pipeline for models, which adapts to the underlying trusted components. Additionally, Penetralium uses a lightweight confidence score perturbation policy to protect against advanced privacy inference attacks on deep learning models. Experimental results demonstrate that Penetralium provides strong security guarantees with reasonable performance. The system not only reduces inference latency and memory consumption overhead but also improves the overall robustness of the system against advanced attacks.

Decentralized Architectures: A Paradigm for Scalable, Fault-Tolerant, and Efficient Distributed Systems

The topic "Decentralized Architectures: A Paradigm for Scalable, Fault-Tolerant, and Efficient Distributed Systems" explores the foundational principles and advantages of decentralized architectures in the context of distributed systems. In essence, decentralized architectures represent a paradigm shift in designing systems that offer scalability, fault tolerance, and efficiency. Scalability is addressed through the distribution of computing tasks across a network of nodes, allowing the system to handle increasing workloads by adding more nodes. This modular approach enables seamless expansion without compromising performance. Additionally, fault tolerance is achieved by decentralizing control, reducing the impact of individual node failures on the overall system. This robustness enhances system reliability and ensures uninterrupted operation, crucial in mission-critical applications. Efficiency is a key focus, as decentralized architectures streamline communication pathways and minimize bottlenecks. Peer-to-peer communication models, consensus algorithms, and distributed databases play pivotal roles in optimizing resource utilization and response times. The abstract emphasizes how decentralization aligns with the demands of contemporary distributed systems, addressing challenges posed by the ever-growing scale and complexity of modern applications. In conclusion, the abstract underscores the significance of decentralized architectures as a transformative paradigm, offering a robust foundation for building scalable, fault-tolerant, and efficient distributed systems in the face of evolving technological landscapes.

Reliable Solder Columns to Replace Solder Balls in Large 2.5D Heterogeneous Packages

Solder balls are well suited for interconnecting normal sized ball grid array (BGA) components to printed circuit boards (PCB). However, solder balls are prone to cracking (failure) caused by excessive stresses inherently found in advanced packaging such as large-sized heterogeneous 2.5D packages where multiple dies are placed side-by-side. The growth trend is to make larger and larger BGA processors exceeding 60x60mm in size for AI and data centers. A plethora of new processor substrate and interposer designs are coming on line such as Chip-on-Wafer-on-Substrates (CoWoS). For that reason, more instances of solder ball delamination and cracking on large-sized BGA packages are probable. Solder columns were produced more than 50 years ago to reduce cracking in large ceramic packages, especially for use in harsh environments found in aerospace and defense applications. However, such legacy solder column technology is limited to Tin-Lead (SnPb) alloys. The challenge is to make Lead Free solder columns that satisfy demand for RoHS compliance, while still providing adequate stress relief (compliancy) for extending the operational life of large sized heterogeneous 2.5D processors for mission critical applications. This paper describes our ongoing work to innovate non-collapsible, compliant Lead-Free Braided Copper Solder Columns to reduce strain caused by the Coefficient of Thermal Expansion (CTE) mismatch between large BGA packages and FR4 PCBs. A braid, consisting of 16 fine copper alloy wires, serves as an exoskeleton sleeve around a solder column to provide mechanical support as well as compliancy. This novel solder column technology supports market trends to scale up the size of chip packaging while maintaining reliability. Our focus was to quickly characterize Lead-Free solder column core materials and braiding alloys by employing “Fail-Fast” destructive accelerated mechanical bend testing in our lab. Since there is an ongoing interest in Tin-Lead alloys, this paper also discusses data on Tin-Lead alloys for comparative purposes.

Artificial intelligence empowered physical layer security for 6G: State-of-the-art, challenges, and opportunities

With the commercial deployment of the 5G system, researchers from both academia and industry are moving attention to the blueprint of the 6G system. The space–air–ground–sea integrated 6G system is envisioned to support various new applications with very diverse requirements in terms of quality of service and security. Due to the open nature of the wireless channel, highly dynamic network topology, and mission-critical applications, 6G will face various security threats. The conventional upper-layer cryptography-based security mechanisms face multiple challenges to resist physical layer attacks in the context of large numbers computing, power-limited low-cost Internet of Thing (IoT) devices. The Physical layer security (PLS) mechanisms by exploiting the random nature of the wireless channel and intrinsic hardware imperfection provide complementary approaches to secure the 6G system. Meanwhile, the quick development and successful application of artificial intelligence facilitate the advancement of PLS solutions. In this paper, we make a comprehensive overview of machine learning-empowered PLS techniques toward 6G. We first introduce the vision of the 6G system, typical applications, key enabling radio technologies, security threats, and security requirements to 6G. Then typical machine learning (ML) methods and the driving forces for introducing intelligent PLS in 6G are presented. Afterward, the major academic works on PLS oriented to key 6G radio techniques including cell-free massive MIMO, visible light communication, Terahertz technology, configurable intelligent surfaces, molecular communication, and ambient backscatter communication are discussed. Then, the latest works on ML-enabled PLS solutions including physical layer key generation, secure wireless transmission, physical layer authentication, intelligent anti-jamming, and ML-enabled attack and detection are systematically reviewed. Finally, we identify important open issues and future research opportunities to inspire further studies to realize a more intelligent and secure 6G system.

Unveiling the Black Box: A Unified XAI Framework for Signal-Based Deep Learning Models

Condition monitoring (CM) is essential for maintaining operational reliability and safety in complex machinery, particularly in robotic systems. Despite the potential of deep learning (DL) in CM, its ‘black box’ nature restricts its broader adoption, especially in mission-critical applications. Addressing this challenge, our research introduces a robust, four-phase framework explicitly designed for DL-based CM in robotic systems. (1) Feature extraction utilizes advanced Fourier and wavelet transformations to enhance both the model’s accuracy and explainability. (2) Fault diagnosis employs a specialized Convolutional Long Short-Term Memory (CLSTM) model, trained on the features to classify signals effectively. (3) Model refinement uses SHAP (SHapley Additive exPlanation) values for pruning nonessential features, thereby simplifying the model and reducing data dimensionality. (4) CM interpretation develops a system offering insightful explanations of the model’s decision-making process for operators. This framework is rigorously evaluated against five existing fault diagnosis architectures, utilizing two distinct datasets: one involving torque measurements from a robotic arm for safety assessment and another capturing vibration signals from an electric motor with multiple fault types. The results affirm our framework’s superior optimization, reduced training and inference times, and effectiveness in transparently visualizing fault patterns.