Most of the existing studies on analyzing the productivity of serial production lines focus on cases where the coefficient of variation ($CV$) for both uptime and downtime is less than 1. Hardly any result is available when$CV>1$, i.e., uptime and downtime of machines exhibit high variability. The improvement of the production lines with high variable uptime and downtime depends on heuristic trial and error due to the lack of analysis method. This article suggests a neural network that approximates the throughput of serial production lines from machine and buffer parameters. Four neural network architectures (multilayer perceptron, recurrent neural network, long short-term memory (LSTM), and gated recurrent unit) are compared to determine the most effective architecture for the throughput approximation task. Training data are obtained from discrete-event simulations, encompassing a wide range of parameters. The results indicate that the LSTM model outperforms the other architecture considered. Furthermore, we present bottleneck identification and continuous improvement scenarios utilizing the model.

Multijob production (MJP) is a class of flexible manufacturing systems, which produces different products within the same production system. MJP is widely used in product assembly, and efficient MJP scheduling is crucial for productivity. Most of the existing MJP scheduling methods are inefficient for multijob serial lines with practical constraints. We propose a deep reinforcement learning (DRL)-driven scheduling framework for multijob serial lines by properly considering the practical constraints of identical machines, finite buffers, machine breakdown, and delayed reward. We analyze the starvation and the blockage time, and derive a DRL-driven scheduling strategy to reduce the blockage time and balance the loads. We validate the proposed framework by using real-world factory data collected over six months from a tier-one vendor of a world top-three automobile company. Our case study shows that the proposed scheduling framework improves the average throughput by 24.2% compared with the conventional approach.

We introduce Phalanx, a failure-resilient truck pla- tooning system, where trucks in a platoon protect each other from sensor failures despite the lack of redundant sensors. For that, we first emulate the failed sensors by collectively utilizing other sensors across the platoon. If the failed sensor cannot be emulated, the control system is instantaneously reconfigured to a cooperative protection mode using only the live sensors. We take a scenario-based approach considering six scenarios with single and dual failures of the essential sensors (i.e., lidar, encoder, and camera) for platooning control. For each scenario, we present a protection method that enables the safe maneuvering of platoons. For the evaluation, Phalanx is implemented using our scale truck testbed instrumented with fault injection modules, demonstrating safe platooning controls for the failure scenarios.

Cyclops, introduced in this paper, is an open research platform for everyone who wants to validate novel ideas and approaches in self-driving heavy-duty vehicle platooning. The platform consists of multiple 1/14 scale semi-trailer trucks equipped with associated computing, communication and control modules that enable self-driving on our scale proving ground. The perception system for each vehicle is composed of a lidar-based object tracking system and a lane detection/control system. The former maintains the gap to the leading vehicle, and the latter maintains the vehicle within the lane by steering control. The lane detection system is optimized for truck platooning, where the field of view of the front-facing camera is severely limited due to a small gap to the leading vehicle. This platform is particularly amenable to validating mitigation strategies for safety-critical situations. Indeed, the simplex architecture is adopted in the computing modules, enabling various fail-safe operations. In particular, we illustrate a scenario where the camera sensor fails in the perception system, but the vehicle is able to operate at a reduced capacity to a graceful stop. Details of Cyclops, including 3D CAD designs and algorithm source codes, are released for those who want to build similar testbeds.

Automated measurement of the machine reliability parameters for a production system enables a continuous update of the mathematical model of the system, which can be used for various analyses toward productivity improvement. However, the continuous update may be impeded by some machines of which automated parameter measurements are out of order. Such a situation has been observed, for instance, when some of the machines in the line cannot save log files or Internet of Things devices that measure these machines stop functioning. In this context, this article addresses the problem of estimating the reliability parameters of those machines while avoiding a direct manual measurement (by humans) of uptime and downtime. It turns out that those parameters can be computed using buffer-related data of the neighboring machines along with the system information. With this, a continuous update of the model is possible even though some machines stop recording their status in an automated manner. The method is indirect as opposed to direct manual measurement. The results are derived for synchronous serial production lines with Bernoulli and also exponential reliability characteristics. Our simulation studies verify the accuracy of the proposed estimation methods.

The digital screens installed inside subway stations generate advertisement revenue by displaying advertisements to passengers. Digital advertisements are increasingly preferred by advertisers because of the high traffic volumes in subway. This paper designs a digital advertising system based on the historical demand information extracted from the smart card data. To this end, we propose a method of designing advertisement products tailored to the digital screens in subway. Next, we consider a reservation system for the designed products with an objective of maximizing the advertisement revenue. The linear programming model is used for the reservation control. If the reservation requests arrive with a Poisson process, the dynamic programming model is used for a more accurate control. The final problem to address is how to schedule the accepted reservations for a maximum exposure to subway passengers. The scheduling problem is the traditional knapsack problem, and the simple greedy method is optimal. Numerical study is performed using our real-life smart card data from Daegu, South Korea. Our data set does not have the demographic information. For the case where this information is available, this paper describes the model for the location-based targeted advertising.

Virtual coupling, which refers to the operation of railway vehicles that enables the merge and separation of vehicles on the move by controlling the gap between the vehicles without any mechanical coupling, is one of the technologies for increasing the transport capacity and enhancing operational efficiency. This paper proposes a robust gap controller based on sliding mode control with a nonlinear train model with uncertainties. Additionally, a gap reference generation scheme is developed that ensures that the merge and separation of two trains is completed before a given location and respects constraints on acceleration and jerk. The position and velocity measurement errors arising from imperfect knowledge of wheel diameters are also considered, and a new error correction scheme is proposed to reduce the perturbation in the gap control performance. The proposed schemes are validated through simulations.

The mean time between failures (MTBF) and mean time to repair (MTTR) of manufacturing equipment (e.g., machines) are used in every quantitative method for production systems performance analysis, continuous improvement, and design. Unfortunately, the literature offers no methods for evaluating the smallest number of up- and downtime measurements necessary and sufficient to calculate reliable estimates of these equipment characteristics. This article is intended to provide such a method. The approach is based on introducing the notion of$(\alpha, \beta)$-precise estimates, where$\alpha $characterizes the estimate’s accuracy and$\beta $its probability. Using this notion, this article evaluates the critical number,$n^{*} (\alpha,\beta)$, of up- and downtime measurements necessary and sufficient to calculate$(\alpha, \beta)$-precise estimates of MTBF and MTTR. In addition, this article derives a probabilistic upper bound of the observation time required to collect$n^{*} (\alpha,\beta)$measurements.Note to Practitioners—To evaluate and predict production systems behavior, managers of manufacturing operations need to know equipment reliability characteristics. Quantifying the equipment status by MTBF and MTTR, this article provides answers to the following questions: Q1: How many measurements of machines up- and downtime are required to obtain reliable estimates of MTBF and MTTR? Q2: How long the observation period must be to collect the desired number of measurements? The answer to Q1 is provided by a rule (formula), which is based on the desired estimate accuracy (characterized by$\alpha $) and its likelihood (quantified by$\beta $). The answer to Q2 consists of selecting a small number of initial measurements, which can be used to calculate an upper bound of the total observation time.

Cyber-physical attacks exploit intrinsic natures of physical systems and can severely damage cyber-physical systems (CPSs) without being detected by the conventional anomaly detector. In this article, based on software-defined networking, we propose a holistic resilient CPS framework that can detect, isolate, and recover from cyber-physical attacks in real time. To show the effectiveness of the proposed framework, we focus on the pole-dynamics attack (PDA), a newly reported stealthy sensor attack that can make the physical system unstable. We develop an efficient detection algorithm for PDA and embed it into the proposed framework. By implementing a testbed, we validate that the proposed framework guarantees resilience of CPS against the PDA.

Cyber-physical system (CPS) provides a well-organized integration betweencommunication,computation, andcontrol(3C) technologies. CPS has been widely used in the vehicular networks and it requires to discover multimodal data from the physical system to make appropriate decisions and actions, for example, congestion warnings, applying brakes, adjusting speed limits, etc. Information discovery and availability at individual network elements is one of the fundamental foundations of CPS. In this paper, we proposed two multimodal network information discovery schemes for vehicular CPS using the Named Data Networking (NDN). One of the proposed schemes simply modifies the pull-based NDN communication mechanism to discover multimodal multi-hop data from the network and the other scheme uses the Interest broadcast suppression (IBS) mechanism. The proposed Interest broadcast suppression scheme adapts the holding time technique to defer the Interest forwarding and its computation involves the hop-count, distance, and other network parameters. Simulation results show that the proposed schemes discover about 172 and 162 percent more multimodal information from approximately 283 and 210 percent more network area by suppressing approximately 50 percent of the Interest broadcast storm in highway and the urban traffic scenarios, respectively.

In this paper, we present a sensor attack on cyber-physical systems (CPSs), which can be constructed with limited knowledge of the target system and can remain stealthy until the attack succeeds. The target CPS consists of a physical plant with unstable linear dynamics and a feedback controller. Specifically, the attack mechanism is to impede the stabilizing function of the feedback controller by injecting false data to the sensors, where the false data are created using the unstable dynamics of the plant. When the only nominal model for the target dynamics is known, the stealthiness is maintained by deploying a mechanism similar to a disturbance observer (DOB) which can be designed to absorb the effect of the mismatch between the nominal and actual dynamics until the attack succeeds. The success of the attack is defined by the norm of the system state exceeding a threshold. Sensor attacks that exploit unstable plant dynamics had been conceived previously. Generation of such attacks require precise knowledge of the target system for stealthiness, i.e., the attack must cancel at the sensor exactly the effect of instability in order to avoid detection. When not exact, the mismatch grows exponentially leading to the detection of abnormality. The attack presented in this paper absorbs the mismatch using the DOB mechanism, where the degree of absorption is selected such that the detection is delayed until the attack succeeds. Thus, the proposed attack, compared to the conventional ones, poses a greater level of threat to CPS. In this paper, generation of the attack is presented, and the effect is analyzed. The consequence of the attack is illustrated and emphasized by simulations on quadrotors and inverted pendulums, respectively.

A cyber-physical system (CPS) is an entanglement of physical and computing systems by real-time information exchange through networking, which can be considered as real-time IoT because of end-to-end real-time performance guarantee. Most societal infrastructures, such as transportation systems, smart power grid, smart factory, and smart buildings, are key application domains of CPS. Though there have been extensive studies on infrastructures from the perspective of cyber security, insufficient research has been conducted from a practical viewpoint of cyber-physical security. In this paper, we focus on train control systems as one of the critical infrastructures. We fully investigate the emerging de facto standard of train control systems, communication-based train control (CBTC). We analyze the cyber-physical vulnerability of CBTC and discover that a man-in-the-middle attack combined with knowledge on train signaling can cause train collisions in CBTC. To resolve the issue, we propose a countermeasure for resiliency of CBTC. By implementing a realistic CBTC testbed, we validate our analysis. To the best of our knowledge, this is the first in-depth empirical study on cyber-physical vulnerability of CBTC systems.

Although many stroke survivors are not fully capable of driving, they drive again without any formal assessment due to an absence of valid screening tools. This leads to an elevated risk of accidents. Although an on-road test is considered a standard assessment method for items relevant to actual driving, it may be dangerous to evaluate all stroke drivers with the on-road test. For safe pre-screening of unsuitable stroke drivers, we propose an automatic Driving Performance Assessment System for Stroke drivers (Driving-PASS). Driving-PASS aims to provide not only information about problematic driving assessment items but also a decision about fitness to drive. The problematic driving items are classified by abnormal classifiers while the decision item is determined by a decision classifier in Driving-PASS. For designing the system, we firstly propose a subjective assessment method consisting of ten assessment items and one decision item. And then, we propose an automated method of the subjective assessment method with a machine learning approach (i.e., ANN and SVM) by using assessment criteria from five expert's judgments. Evaluation results demonstrate that Driving-PASS automatically assess not only the ten assessment items (total average Accuracy of 90% and F1-score of 88%) but also the decision item (Accuracy of 93% and F1-score of 92%). We expect Driving-PASS provides analytical assessment results that can be used in driving rehabilitation programs and contributes to reducing the risk of vehicle accidents by pre-screening unsuitable stroke drivers with high accuracy and reliability.

During the past two decades, we have witnessed a tremendous development in Vehicular networks, while exploring emerging communication technologies such as vehicular cyber-physical systems (VCPS). Basically, VCPS requires multimodal data from the physical system to take appropriate decision and actions, for example, the congestion warnings, applying brakes, adjusting speed limits, etc. However, there are multiple systems interconnected in the VCPS with different communication capabilities and data communication between those systems that lead us to a challenging task. In this paper, we consider named data networking (NDN) as a promising solution to enhance the reachability of Data among multi-hop VCPS. NDN offers a simple pull-based content communication in the network with multiple interfaces and also supports heterogeneity in terms of communications technologies. The proposed NDN forwarding scheme enables vehicles to send one Interest (request) to collect multiple instances of the Data from different content sources in the network. Simulation results show that the proposed scheme can collect information from many nodes that are at longer distance from the information requesting nodes.

With ever increasing number of vehicles, shortage of parking space is becoming a serious problem. Going to shopping, school, and workplace can be a headache as finding an available parking spot is getting harder causing wasted time and gas. In this paper, we present WiParkFind: a low-cost, non-intrusive, and real- time parking occupancy monitoring system based on WiFi signals. The channel state information (CSI) of received WiFi signals is analyzed by using a machine learning technique to capture distinctive characteristics of CSI data that are strongly correlated with the number of empty parking slots in order to detect whether there is an empty slot, and how many empty slots are available. Compared with contemporary approaches based on magnetic sensors deployed on individual parking slots, WiParkFind utilizes low-cost off- the-shelf WiFi devices, dramatically reducing the cost for purchasing, installing, and maintaining a large number of sensors, and backend server systems. A proof-of-concept system of WiParkFind was developed and deployed in a department parking lot. The results demonstrate that the average classification accuracy of WiParkFind over a week of data collection is 78.2%, and the accuracy is improved to 90.8% with a tolerance of one empty slot.

Sleep position is directly related to sleep quality especially in patients with sleep disorders such as sleep apnea or snoring. We propose SleeP-Manager, a wearable embedded system, that is designed to aid Sleep Positional Therapy (SPT). SleeP-Manager with two wristbands monitors the sleep position of the user and gives a vibration feedback when a poor position is detected. We experimentally evaluate the effectiveness of SleeP-Manager. In order to accomplish this, an additional device of chestband is designed. The chestband provides the true sleep position and also measures the response of users to the vibration feedback. The results indicate that the accuracy of sleep position detection higher than 80%, and the ratio of desired sleep position per night increased significantly by the use of SleeP-Manager. Our questionnaire survey shows the wristband-typed device is most preferred for SPT due to the cost-effectiveness, easy-to-wear, and practicality.

Wearable systems are commonly used for fitness purpose as these devices provide activity measurements to motivate daily exercise. With aims to promote improved health, healthcare companies are incentivizing their customers with the amount of exercise that is performed and using readings from wearable devices as a way of proving that the individual met the requirements. However, these devices have a risk of user spoofing attacks as an unauthorized individual can utilize the system. To prevent misuse of the product to gain reward and ultimately promote daily exercise for various types of exercise reward systems, we propose a biometric gait identification approach using a smart earring that we design and develop. In this paper, we preliminary train and test the gait identification system by utilizing a transfer learning, which shows a 100% classification performance for eight participants. We expect the proposed gait identification technique will serve as essential building blocks for reliable exercise reward systems.

Stroke is the fifth leading cause of death in the US. Early recognition and treatment of stroke are essential for a good clinical outcome. It is particularly challenging for Wake-Up Stroke (WUS) to know the time of stroke onset, hence golden time for treatment is easily missed. We propose a Real-tIme StroKe early detection system during Sleep (RISK-Sleep) using wristbands. RISK-Sleep is a solution for early stroke detection tailored for the sleep environment that is cost-effective and practical for daily use. Underneath RISK-Sleep, we define and utilize an abnormal sleep motion model consisting of abnormal intensity and abnormal frequency. The abnormal intensity indicates hemiparesis sleep motion patterns while the abnormal frequency means emergency situations such as full hemiparesis and full paralysis. Based on the model, we seek the best classifier that analyzes the aforementioned two abnormal motion patterns by sliding window in real-time. For performance evaluation, we collect sleep data from 30 healthy people and 14 stroke patients with hemiparesis. Evaluation results show that RISK-Sleep achieves classification accuracy of 96.00% in abnormal intensity with 146-minute window in the KNN classifier with SFS feature selection. In addition, the SVM classifier without feature selection shows classification accuracy of 100% with 108-minute window in abnormal frequency. We expect RISK-Sleep plays a significant role in reducing the incidence of WUS.

With ever-increasing number of car-mounted electronic devices that are accessed, managed, and controlled with smartphones, car apps are becoming an important part of the automotive industry. Audio classification is one of the key components of car apps as a front-end technology to enable human-app interactions. Existing approaches for audio classification, however, fall short as the unique and time-varying audio characteristics of car environments are not appropriately taken into account. Leveraging recent advances in mobile sensing technology that allow for effective and accurate driving environment detection, in this paper, we develop an audio classification framework for mobile apps that categorizes an audio stream into music, speech, speech+music, and noise, adaptably depending on different driving environments. A case study is performed with four different driving environments, i.e., highway, local road, crowded city, and stopped vehicle. More than 420 minutes of audio data are collected including various genres of music, speech, speech+music, and noise from the driving environments. The results demonstrate that the proposed approach improves the average classification accuracy by up to 166%, and 64% for speech, and speech+music, respectively, compared with a non-adaptive approach in our experimental settings.

Sleep plays an important role in recovering physical and mental functions. Sleep position is known to affect sleep quality, hence, managing sleep position is beneficial for patients suffering from sleep disorders. For a long-term sleep management, we propose a sleep position tracking system using two wristbands. From the data collected from the wristbands, the system detects sleep positions and their changes. We define a sleep position motion model that consists of seven transitions between three sleep positions. Then, we propose pre-processing methods to overcome difficulties in analyzing sleep motion data, i.e., discontinuity, uncertainty, and time-variability. We tested experimental data in state-of-art pre-trained convolution neural networks by transfer learning. The accuracy of our proposed system was 96.03% and 88.02% in pilot experiment and on-site sleep experiment, respectively. Our experimental results demonstrate that the proposed system effectively and accurately keeps track of sleep positions without causing any inconvenience to users, and hence, serves as a key building block for cost-effective 24/7 sleep monitoring solutions

Thermal control is critical for real-time systems as overheated processors can result in serious performance degradation or even system breakdown due to hardware throttling. The major challenges in thermal control for real-time systems are (i) the need to enforce both real-time and thermal constraints; (ii) uncertain system dynamics; and (iii) thermal sensor noise. Previous studies have resolved the first two, but the practical issue of sensor noise has not been properly addressed yet. In this paper, we introduce a novel thermal control algorithm that can appropriately handle thermal sensor noise. Our key observation is that even a small zero-mean sensor noise can induce a significant steady-state error between the target and the actual temperature of a processor. This steady-state error is contrary to our intuition that zero-mean sensor noise induces zero-mean fluctuations. We show that an intuitive attempt to resolve this unusual situation is not effective at all. By a rigorous approach, we analyze the underlying mechanism and quantify the noised-induced error in a closed form in terms of noise statistics and system parameters. Based on our analysis, we propose a simple and effective solution for eliminating the error and maintaining the desired processor temperature. Through extensive simulations, we show the advantages of our proposed algorithm, referred to as Thermal Control under Utilization Bound with Virtual Saturation (TCUB-VS).

The reliability of control in cyber-physical systems (CPSs) heavily depends on the network-induced delay. The problem of obtaining a maximum allowable delay bound has been widely studied in the networked control systems (NCS) area. Once the delay bound is derived, the remaining question is how to make a network satisfy the bound. In this paper, we propose a robust path selection algorithm, which exploits multipath diversity for providing robust network performance against intrinsic randomness in delay. Our path selection algorithm gives the required paths for any given robustness level parameterized by the reliability violation probability. Based on extensive experimental results with our testbed, we empirically show that the proposed scheme can provide the required network quality of service (QoS) for system robustness.