Adaptive Anomaly Detection for Internet of Things in Hierarchical Edge Computing: A Contextual-Bandit Approach

The advances in deep neural networks (DNN) have significantly enhanced real-time detection of anomalous data in IoT applications. However, the complexity-accuracy-delay dilemma persists: Complex DNN models offer higher accuracy, but typical IoT devices can barely afford the computation load, and the remedy of offloading the load to the cloud incurs long delay. In this article, we address this challenge by proposing an adaptive anomaly detection scheme with hierarchical edge computing (HEC). Specifically, we first construct multiple anomaly detection DNN models with increasing complexity and associate each of them to a corresponding HEC layer. Then, we design an adaptive model selection scheme that is formulated as a contextual-bandit problem and solved by using a reinforcement learning policy network . We also incorporate a parallelism policy training method to accelerate the training process by taking advantage of distributed models. We build an HEC testbed using real IoT devices and implement and evaluate our contextual-bandit approach with both univariate and multivariate IoT datasets. In comparison with both baseline and state-of-the-art schemes, our adaptive approach strikes the best accuracy-delay tradeoff on the univariate dataset and achieves the best accuracy and F1-score on the multivariate dataset with only negligibly longer delay than the best (but inflexible) scheme.

An Event-Triggered Model Predictive Control for Energy Efficiency and Thermal Comfort Optimization in Buildings

Model predictive control (MPC) is a promising optimal control technique for building automation. However, the high computation load to solve the optimization problem of MPC is challenging its implementation for real-time building control. Typical MPC systems employ the time-triggered mechanism (TTM), which conducts the optimization periodically at each control interval regardless of the necessity. This study proposes an event-triggered mechanism (ETM) for MPC, which conducts the optimization only when there is a triggering event that necessitates it. Contrasting to the conventional ETM that bases only on the current information, the proposed ETM bases on the cost function considering the past, current and future information. An event-triggered model predictive control (ETMPC) system is developed using the proposed ETM. In a simulation environment, the ETMPC system is implemented to control an air-conditioning system. The ETMPC is compared to a MPC employing TTM and a conventional thermostat. The ETMPC improved the computation efficiency by 77.6% - 88.2% as compared to the MPC while achieving similar energy performance as the MPC does (both achieved more than 9% energy savings over the thermostat). The ETMPC only degraded the thermal comfort performance slightly as compared to the MPC but is still much better than the thermostat.

Ir-UNet: Irregular Segmentation U-Shape Network for Wheat Yellow Rust Detection by UAV Multispectral Imagery

Crop disease is widely considered as one of the most pressing challenges for food crops, and therefore an accurate crop disease detection algorithm is highly desirable for its sustainable management. The recent use of remote sensing and deep learning is drawing increasing research interests in wheat yellow rust disease detection. However, current solutions on yellow rust detection are generally addressed by RGB images and the basic semantic segmentation algorithms (e.g., UNet), which do not consider the irregular and blurred boundary problems of yellow rust area therein, restricting the disease segmentation performance. Therefore, this work aims to develop an automatic yellow rust disease detection algorithm to cope with these boundary problems. An improved algorithm entitled Ir-UNet by embedding irregular encoder module (IEM), irregular decoder module (IDM) and content-aware channel re-weight module (CCRM) is proposed and compared against the basic UNet while with various input features. The recently collected dataset by DJI M100 UAV equipped with RedEdge multispectral camera is used to evaluate the algorithm performance. Comparative results show that the Ir-UNet with five raw bands outperforms the basic UNet, achieving the highest overall accuracy (OA) score (97.13%) among various inputs. Moreover, the use of three selected bands, Red-NIR-RE, in the proposed Ir-UNet can obtain a comparable result (OA: 96.83%) while with fewer spectral bands and less computation load. It is anticipated that this study by seamlessly integrating the Ir-UNet network and UAV multispectral images can pave the way for automated yellow rust detection at farmland scales.

Efficient Transformer for Remote Sensing Image Segmentation

Semantic segmentation for remote sensing images (RSIs) is widely applied in geological surveys, urban resources management, and disaster monitoring. Recent solutions on remote sensing segmentation tasks are generally addressed by CNN-based models and transformer-based models. In particular, transformer-based architecture generally struggles with two main problems: a high computation load and inaccurate edge classification. Therefore, to overcome these problems, we propose a novel transformer model to realize lightweight edge classification. First, based on a Swin transformer backbone, a pure Efficient transformer with mlphead is proposed to accelerate the inference speed. Moreover, explicit and implicit edge enhancement methods are proposed to cope with object edge problems. The experimental results evaluated on the Potsdam and Vaihingen datasets present that the proposed approach significantly improved the final accuracy, achieving a trade-off between computational complexity (Flops) and accuracy (Efficient-L obtaining 3.23% mIoU improvement on Vaihingen and 2.46% mIoU improvement on Potsdam compared with HRCNet_W48). As a result, it is believed that the proposed Efficient transformer will have an advantage in dealing with remote sensing image segmentation problems.

The effects of using variable lengths for degraded signal acquisition in GPS receivers

The signal acquisition in GPS receivers is the first and very crucial process that may affect the overall performance of a navigation receiver. Acquisition program initiates a searching operation on received navigation signals to detect and identify the visible satellites. However, signal acquisition becomes a very challenging task in a degraded environment (i.e, dense urban) and the receiver may not be able to detect the satellites present in radio-vicinity, thus cannot estimate an accurate position solution. In such environments, satellite signals are attenuated and fluctuated due to fading introduced by Multipath and NLOS reception. To perform signal acquisition in such degraded environments, larger data accumulation can be effective in enhancing SNR, which tradeoff huge computational load, prolonged acquisition time and high cost of receiver. This paper highlights the effects of fading on satellite signal acquisition in GPS receiver through variable data lengths and SNR comparison, and then develops a statistical relationship between satellite visibility and SNR. Furthermore it also analyzes/investigates the tradeoff between computation load and signal data length.

Geometric Optimization of an Extracorporeal Centrifugal Blood Pump with an Unshrouded Impeller Concerning Both Hydraulic Performance and Shear Stress

Centrifugal blood pumps have provided a powerful artificial support system for patients with vascular diseases. In the design process, geometrical optimization is usually needed to acquire a more biocompatible model for clinical uses. In the current paper, we propose a method for multi-objective optimization concerning both the hydraulic and the hemolytic performances of the pump based on the near-orthogonal array in which the traditional hemolysis index (HI) is replaced with the maximum scalar shear stress criteria to reduce the computation load. The method is demonstrated with the optimization of an extracorporeal centrifugal blood pump with an unshrouded impeller. CFD studies on the original and nine modified pump models are carried out. The calculated hydraulic performances of the optimized model are also compared against the experiments for validation of the numeric method, with an error of 3.6% at the original design point. The resulting blood pump with low maximum scalar shear stress (132.2 Pa) shows a low degree of calculated HI (1.69 × 10−3).

Distributed moving horizon state estimation for sensor networks with low computation capabilities

This paper focuses on distributed state estimation for sensor network observing a discrete-time linear system. The provided solution is based on a Distributed Moving Horizon Estimation (DMHE) algorithm considering a pre-estimating Luenberger observer in the formulation of the local problem solved by each sensor. This leads to reduce the computation load, while preserving the accuracy of the estimation. Moreover, observability properties of local sensors are used for tuning the weights related to consensus information fusion built on a rank-based condition, in order to improve the convergence of the estimation error. Results obtained by Monte Carlo simulations are provided to compare the performance with existing approaches, in terms of accuracy of the estimations and computation time.

Insect-Computer Hybrid System for Autonomous Search and Rescue Mission

There is still a long way to go before artificial mini robots are really used for search and rescue missions in disaster-hit areas due to hindrance in power consumption, computation load of the locomotion, and obstacle-avoidance system. Insect–computer hybrid system, which is the fusion of living insect platform and microcontroller, emerges as an alternative solution. This study demonstrates the first-ever insect–computer hybrid system conceived for search and rescue missions, which is capable of autonomous navigation and human presence detection in an unstructured environment. Customized navigation control algorithm utilizing the insect’s intrinsic navigation capability achieved exploration and negotiation of complex terrains. On-board high-accuracy human presence detection using infrared camera was achieved with a custom machine learning model. Low power consumption suggests system suitability for hour-long operations and its potential for realization in real-life missions.

Improved Event-Based Pi Controller for Limit Cycles Avoidance

One of the issues in designing event-based proportional-integral (PI) controller using aggressive tuning rules is the possible occurrence of limit cycles. To date, there is no adequate simple event-based PI controller technique able to explicitly use aggressive tuning rule. In this paper an improved simple event-based PI controller is proposed to address this issue. By analysing the discrete PI algorithm, an improved triggering condition is introduced. To test the effectiveness of the approach, extensive simulations are carried out by introducing the proposed method to process control under various sampling period, triggering limit, and different tuning rules namely, AMIGO, SIMC and One-Third tuning rules. The performances are evaluated based on two standard criteria: ability to imitate the time-triggered system and computation load reduction. The results show that the performance of the proposed method able to surpass others simple event-based PI controller approaches by giving a closest response to the time-triggered system, lowest computational load and able to avoid the limit cycles occurrences. It is envisaged that the proposed method can be useful in designing a simple event-based PI controller that compatible with any type of tuning rules.