Federated learning synchronizes models through gradient transmission and aggregation. However, these gradients pose significant privacy risks, as sensitive training data is embedded within them. Existing gradient inversion attacks suffer from significantly degraded reconstruction performance when gradients are perturbed by noise-a common defense mechanism. In this paper, we introduce gradient-guided conditional diffusion models for reconstructing private images from leaked gradients, without prior knowledge of the target data distribution. Our approach leverages the inherent denoising capability of diffusion models to circumvent the partial protection offered by noise perturbation, thereby improving attack performance under such defenses. We further provide a theoretical analysis of the reconstruction error bounds and the convergence properties of the attack loss, characterizing the impact of key factors—such as noise magnitude and attacked model architecture—on reconstruction quality. Extensive experiments demonstrate our attack's superior reconstruction performance with Gaussian noise-perturbed gradients, and confirm our theoretical findings.

Text-driven Image to Video Generation (TI2V) aims to generate controllable video given the first frame and corresponding textual description. The primary challenges of this task lie in two parts: (i) how to identify the target objects and ensure the consistency between the movement trajectory and the textual description. (ii) how to improve the subjective quality of generated videos. To tackle the above challenges, we propose a new diffusion-based TI2V framework, termed TIV-Diffusion, via object-centric textual-visual alignment, intending to achieve precise control and high-quality video generation based on textual-described motion for different objects. Concretely, we enable our TIV-Diffuion model to perceive the textual-described objects and their motion trajectory by incorporating the fused textual and visual knowledge through scale-offset modulation. Moreover, to mitigate the problems of object disappearance and misaligned objects and motion, we introduce an object-centric textual-visual alignment module, which reduces the risk of misaligned objects/motion by decoupling the objects in the reference image and aligning textual features with each object individually. Based on the above innovations, our TIV-Diffusion achieves state-of-the-art high-quality video generation compared with existing TI2V methods.

Business intelligence (BI) transforms large volumes of data within modern organizations into actionable insights for informed decision-making. Recently, large language model (LLM)-based agents have streamlined the BI workflow by automatically performing task planning, reasoning, and actions in executable environments based on natural language (NL) queries. However, existing approaches primarily focus on individual BI tasks such as NL2SQL and NL2VIS. The fragmentation of tasks across different data roles and tools lead to inefficiencies and potential errors due to the iterative and collaborative nature of BI. In this paper, we introduce DataLab, a unified BI platform that integrates a one-stop LLM-based agent framework with an augmented computational notebook interface. DataLab supports various BI tasks for different data roles in data preparation, analysis, and visualization by seamlessly combining LLM assistance with user customization within a single environment. To achieve this unification, we design a domain knowledge incorporation module tailored for enterprise-specific BI tasks, an inter-agent communication mechanism to facilitate information sharing across the BI workflow, and a cell-based context management strategy to enhance context utilization efficiency in BI notebooks. Extensive experiments demonstrate that DataLab achieves state-of-the-art performance on various BI tasks across popular research benchmarks. Moreover, DataLab maintains high effectiveness and efficiency on real-world datasets from Tencent, achieving up to a 58.58% increase in accuracy and a 61.65 % reduction in token cost on enterprise-specific BI tasks.

We propose a training-free and robust solution to offer camera movement control for off-the-shelf video diffusion models. Unlike previous work, our method does not require any supervised finetuning on camera-annotated datasets or self-supervised training via data augmentation. Instead, it is plug-and-play with most pretrained video diffusion models and can generate camera-controllable videos with a single image or text prompt as input. The inspiration for our work comes from the layout prior that intermediate latents encode for the generated results, thus rearranging noisy pixels in them will cause the output content to relocate as well. As camera moving could also be seen as a type of pixel rearrangement caused by perspective change, videos can be reorganized following specific camera motion if their noisy latents change accordingly. Building on this, we propose **CamTrol**, which enables robust camera control for video diffusion models. It is achieved by a two-stage process. First, we model image layout rearrangement through explicit camera movement in 3D point cloud space. Second, we generate videos with camera motion by leveraging the layout prior of noisy latents formed by a series of rearranged images. Extensive experiments have demonstrated its superior performance in both video generation and camera motion alignment compared with other finetuned methods. Furthermore, we show the capability of CamTrol to generalize to various base models, as well as its impressive applications in scalable motion control, dealing with complicated trajectories and unsupervised 3D video generation. Videos available at https://lifedecoder.github.io/CamTrol/.

For multiple smart sensors (MSSs) operating with limited energy, a smart sensor is said to be controllable if there exists proper external signal (ES) enabling it to report the sensing data, reflecting the physical world with tolerable sensing errors. In order to improve energy-efficiency while meeting the required sensing accuracy, the controllable smart sensors are expected to be activated while the uncontrollable ones are not. However, due to the exterior noises or interior interferences (ENII) of MSSs, whether a smart sensor is or not controllable is often uncertain, which may wrongly activate uncontrollable sensors or deactivate controllable ones, not only wasting energy but also deteriorating sensing errors. Therefore, addressing the problem of assuring MSSs with uncertain controllability to report tolerable sensing errors at the minimum cost of energy emerges as an interesting and challenging issue. To address this problem, this paper first formulates MSSs as the probabilistic multiple input multiple output (PMIMO) system, with all the ESs and sensing errors as the input and output, respectively. It then discloses the optimal activation probability for each smart sensor and reveals the optimal ES corresponding to the minimum energy for MSSs experiencing uncertain controllability to report the desired sensing errors. An algorithm is proposed based on our discovered results to implement such ES. The theoretical analysis and field experiments verify its performance. Note to Practitioners—This paper addresses the challenging issue of how to guarantee MSSs experiencing uncertain controllability to report tolerable sensing errors at the minimum cost of energy. Though the insightful discovery of the optimal probability that each smart sensor is activated, and further the optimal ES corresponding to the minimum energy for enabling MSSs experiencing uncertain controllability to report the desired sensing errors, this paper facilitates MSSs with uncertain controllability to operate in an energy-efficient as well as high-accurate manner, which we believe could improve the power usage and strengthen the sensing capacity of the Internet of Things (IoT) or Cyber-Physical Systems (CPS) that depend on MSSs settings to sense the physical world. Theoretical analysis and experiments both verify that the proposed solution outperforms existing ones.

Piecewise linear neural networks (PLNNs) are proven universal approximators for continuous functions on the compact domain. For multiple PLNNs (mPLNNs) differing from each other in suffering different approximation errors (AEs) when approximating the same continuous function, activating all the PLNNs to approximate the continuous function, and then, picking up the minimum AE (MAE) from all the AEs seems to be a practical way to arrive at such MAE. Activating PLNNs has to consume energy, and more activated PLNNs provide more AEs to consider, which can maximize the probability of guaranteeing the MAE, while also needing to harvest more energy. Therefore, how to make the optimal tradeoff between energy harvested and approximation accuracy for mPLNNs arises as an interesting issue. To address this problem, this paper first deduces the objective function, with the accumulative probability of obtaining the MAE as the objective and the accumulative energy harvested as the constraint, then reveals the optimal probability that each hidden neuron (HN) in the activated PLNN should be activated, and finally uncovers the optimal activation probability for each PLNN. An algorithm based on our discovered theoretical results is proposed for mPLNNs to enjoy the maximum probability of achieving the MAE at the acceptable level of accumulative energy harvested. Theoretical analysis and experiments verify its performance. Note to Practitioners—This paper addresses the interesting issue of how to make mPLNNs approximate the continuous function with the highest approximation accuracy under the energy constraint. Through the insightful discovery of the optimal activation probability for each HN in the activated PLNN as well as the optimal activation probability for each PLNN, this paper facilitates mPLNNs to suffer the MAE in their approximations to continuous functions, with the maximum probability while maintaining the energy harvested within a given range. Because PLNNs can model the nonlinear and complex system with the arbitrary AE, our proposed approach can be effectively applied by such system to operate in an energy-efficient and universal-approximation manner, which we believe could push the development of the low-power multiple neural networks.

The paper considers the embedded system that can either compute tasks locally by itself or offload tasks to the edge server for remote computing during the running period (RP) and switch to the sleep mode to save energy once the RP ends, i.e., the idle period (IP) arrives. The tasks are stored in the cache, and the more tasks are computed during the RP, the less cache space will be occupied at the end of RP, while also leading to more energy consumption. Meanwhile, the sleep mode that the embedded system enters during the IP also influences the energy consumption. Therefore, how to make the optimal tradeoff between energy consumption and cache occupancy arises as an interesting issue. To address this issue, this paper first establishes an optimization-theoretical framework to formulate the energy consumption under the constraint of cache occupancy, then discovers the most energy-saving RP, computing mode (i.e., local or edge computing), and low-power mode. An algorithm based on our discovered theoretical results is proposed for the embedded system to minimize the energy consumption within the acceptable level of cache occupancy. Theoretical analysis and field experiments jointly verify its good performance. Note to Practitioners—This paper addresses the interesting tradeoff between energy consumption and cache occupancy in the embedded system that operates in the environments with limited available energy as well as cache space. It facilitates to improve the operation efficiency of the embedded systems in the area of Internet of Things (IoT) or Cyber-Physical Systems (CPS) that employs edge computing technology to empower embedded systems with more computing capability and sleep modes to guarantee embedded systems with more energy savings, in order to minimize the accumulative energy consumption, while maintaining the cache occupancy in terms of task data bits stored within an acceptable range. Experimental investigations show that the solution proposed here outperforms existing ones.

Generalized Laguerre neural network (GLNN) is a specialized artificial neural network that employs generalized Laguerre polynomials (GLPs) as activation functions for hidden neurons, offering distinct advantages in function approximation tasks. Statistically, the larger number of activated GLPs, the more hidden neurons utilizing these GLPs will be engaged, thereby enhancing approximation accuracy. However, this comes with a tradeoff: Each arithmetic operation inherent to GLP computations consumes energy, leading to higher energy consumption. Consequently, optimizing the activation patterns of GLPs to strike an optimal tradeoff between approximation accuracy and energy efficiency emerges as a compelling research issue. To address this issue, this paper establishes an energy-constrained approximation accuracy optimization framework. Within this framework, we reveal the optimal number of candidate GLPs serving as activation functions for hidden neurons, the optimal number of hidden neurons that should be set in the GLNN, and the optimal activation criterion for candidate GLPs, under energy constraint. We also establish the root-exponentially decay convergence of the minimum approximation error under energy constraint. An algorithm based on our discovered results is proposed for the GLNN to minimize the approximation error under the acceptable level of energy consumption. Theoretical analysis and experiments jointly demonstrate its performance.

The Laguerre Neural Network (LaNN) employs Laguerre polynomials (LPs) as the activation functions of its hidden neurons (HNs) to approximate the nonlinear functions (NFs). From the perspective of statistics, the more LPs are activated, the more LP-based activation functions (LPAFs) of the HNs will operate, and thus the smaller approximation error (APE) will be suffered by the LaNN, while leading to more energy consumption since activating any LP has to consume energy. Therefore, how to control the activation of LPs to make the optimal tradeoff between approximation accuracy and energy consumption arises as an interesting issue. To address this issue, this paper first establishes an energy-constraint probability-based approximation-accuracy optimization-theoretical (EPAO) framework, considering the probability of obtaining the minimum APE (MAPE) as the objective while the energy consumption of the LaNN as the constraint, then reveals the optimal number of HNs and thus their optimal LPAFs, and finally discloses the optimal activation probability for each LP. Based on our discovered theoretical results, a novel algorithm for LaNN to maximize the probability of obtaining the MAPE at the acceptable level of energy consumption is proposed. Theoretical analysis verifies its performance, and field experiments verify that, compared with existing methods, it improves the probability of obtaining the MAPE by 6.00% to 20.26% at the similar cost of energy. To our best knowledge, this is the initial work for LaNN towards joint optimization of its energy efficiency and approximation accuracy.

This paper studies the refactoring problem of multiple smart sensors (MSSs) with edge computing capability, where the program consisting of codes and driving MSSs distributes into all the smart sensors, the available energy for refactorings of MSSs is limited, and each smart sensor adopts binary refactoring mode, i.e., either performs refactoring locally by itself or fully offloads its codes to the edge server (ES) for edge refactoring. More energy strengthens MSSs to wipe out more code bugs (CBs), while corresponding to more energy consumption. Meanwhile. the refactoring mode (i.e., local or edge refactoring) also influences the CB ratio (CBR) and energy efficiency. Therefore, how to make the optimal tradeoff between CBR and energy consumption for such MSSs arises as an interesting issue. To address this issue, this paper first reveals a necessary and sufficient condition to judge whether the minimum CBR can be reached as well as its analytical expression, and then discloses the optimal refactoring time, refactoring computation rate, and refactoring mode, in terms of guaranteeing the minimum CBR under the energy constraint. An algorithm based on our discovered foundations is proposed for such MSSs to minimize the CBR within the acceptable level of energy consumption. Theoretical analysis, simulation and field experiments verify its performance. To our best knowledge, this is the initial work towards the optimal refactoring of MSSs. Note to Practitioners—For MSSs in practice, CBs lurking in their driving program often endanger their normal function. As a kind of behavior-preserving code transformation, the refactoring built in MSSs can help them to remove CBs. Such refactoring must consume energy and often suffers energy setback because the electricity and computing power of MSSs are usually very limited. To overcome this setback, the edge computing is introduced for MSSs to offload their codes to the ES for refactoring. However, the combination of MSSs and edge computing heavily challenges energy-saving refactoring, involving the invisible and unknown CBs and the curse of dimensionality regarding code-offloading. Accordingly, this work allows MSSs with edge computing capability to operate in a program-healthy and energy-efficient manner by making the optimal tradeoff between CBR and energy consumption, covering both theoretical results and algorithm. Specifically, a necessary and sufficient condition to judge whether the minimum CBR can be reached is disclosed, the analytical expression of that minimum CBR is derived out, the optimal refactoring time, refactoring computation rate, and refactoring mode for each smart sensor are discovered, and an effective algorithm for the practitioners to enjoy the minimum CBR while maintaining the energy consumption within a given range is further proposed. By controlling the refactoring time, refactoring computation rate, and refactoring mode given that the necessary and sufficient condition is satisfied, this work can be applicable for MSSs with edge computing capability to suffer the minimum CBR when they are employed to sense the physical world in the scenarios where their available energy is limited.

Diffusion models have attained remarkable success in the domains of image generation and editing. It is widely recognized that employing larger inversion and denoising steps in diffusion model leads to improved image reconstruction quality. However, the editing performance of diffusion models tends to be no more satisfactory even with increasing denoising steps. The deficiency in editing could be attributed to the conditional Markovian property of the editing process, where errors accumulate throughout denoising steps. To tackle this challenge, we first propose an innovative framework where a rectifier module is incorporated to modulate diffusion model weights with residual features from the original images, thereby providing compensatory information to bridge the fidelity gap. Furthermore, we introduce a novel learning paradigm aimed at minimizing error propagation during the editing process, which trains the editing procedure in a manner similar to denoising score-matching. Extensive experiments demonstrate that our proposed framework and training strategy achieve high-fidelity reconstruction and editing results across various levels of denoising steps, meanwhile exhibits exceptional performance in terms of both quantitative metric and qualitative assessments. Lastly, we explore our model's generalization though several applications like image-to-image translation and out-of-domain image editing.

For multiple smart sensors (MSSs) operating with limited energy, the external signal (ES) often serves an indispensable role in enabling them to report their sensing data, reflecting the physical world with tolerable sensing errors. However, due to interior disturbances lurking in the MSSs or exterior noises originating from practical environments, the ES may become faulty for some smart sensors and subsequently affect the sensor connections. This not only deteriorates sensing errors but also wastes energy. Therefore, addressing the problem of assuring MSSs to report tolerable sensing errors with minimum energy consumption under the faulty ES (FES) emerges as an interesting and challenging issue. To address this problem, this paper first formulates MSSs as a temporal multiple input multiple output (MIMO) system, considering all the ESs and sensing errors as the input and output, respectively. It then reveals a necessary and sufficient condition for MSSs to determine whether or not to update the sensor connections, discloses a criterion guiding MSSs in performing such updates, and finally determines the optimal ES in terms of controlling MSSs to report tolerable sensing errors with the minimum energy consumption under the FES. An algorithm based on our discovered theoretical foundations is further proposed to implement such an optimal ES. The theoretical analysis and experiments against existing methods validate its performance.Note to Practitioners—This paper addresses the interesting issue of how to ensure that MSSs report acceptable sensing errors with the minimum energy consumption under the FES. Though the insightful disclosure of criterions for MSSs to judge whether and guide how to update the sensor connections, and the optimal ES corresponding to the minimum energy supporting MSSs to report the acceptable sensing errors under the FES, this paper enables MSSs to perform operations in a high-accuracy, energy-efficiency and FES-tolerance manner, which we believe could help the Internet of Things (IoT) or Cyber-Physical Systems (CPS) that employ MSSs settings to capture the physical world to improve their power usage, data accuracy and fault tolerance. Theoretical analysis and experimental investigations both verify that the proposed solution here outperforms the existing solutions.

Piecewise linear neural network (PLNN) possesses universal approximation ability for continuous functions on the compact domain, and for a PLNN in which the hidden neuron (HN) is wireless powered through wireless power transfer (WPT) technology and activated only if its energy harvest arrives at the activation threshold, the more HNs are activated, the smaller approximation error may be enjoyed, while the more energy will be consumed. Therefore, how to control the energy supply for the PLNN to reach the minimum approximation error while keeping the acceptable level of energy harvest arises as an interesting issue. To address this issue, this paper first formulates the energy-harvesting of an individual HN as a Markov decision process (MDP) to further deduce the objective function of the PLNN approximating the given continuous function, then uncovers the optimal activation probability and expected energy harvest for each HN, and finally reveals the optimal energy supply for the HNs to satisfy the optimal energy harvest. A novel algorithm based on our discovered theoretical foundations to control the energy supply to minimize the approximation error within the acceptable level of energy harvest is proposed, and the simulations and field experiments are both given to test its performance. To our best knowledge, this is the initial work towards the joint optimization of energy efficiency and universal approximation performance of the wireless powered PLNN.Note to Practitioners—This paper addresses the interesting issue of how to guarantee wireless powered PLNN to approximate the continuous function with the minimum approximation error under the energy constraint. Through the insightful disclosure of the optimal activation probability, optimal expected energy harvest, and optimal control of energy supply for each HN, this paper facilitates the wireless powered PLNN to operate in an energy-efficiency and universal-approximation way. Since PLNN can model the nonlinear and complex system (NCS) with arbitrary accuracy on the compact domain, our proposed solution can be applied by the NCS modeling technology to enjoy high approximation accuracy while keeping the energy consumption at the acceptable level, which we believe could inspire the development of low-power neural networks (NNs).

For multiple smart sensors with limited energy supply, the relationship between the energy supply and sensor fault-free region is often unknown, and neither of the interior structure nor exterior circumstance modeling the smart sensors is easy to accurately achieve, so how to guarantee the fault-free smart sensors to harvest the most energy in the fault-free way is a very challenging topic. To address this issue, this paper first formulates the individual smart sensor as a single-input single-output (SISO) model-free system (MFS), with its energy supply and sensing error as the input and output, respectively, then makes use of Lyapunov function to deduce an upper bound of the fault-free region to realize the fault-detection of any smart sensor and disclose the relationship between the energy supply and fault-free region, and finally achieves the optimal energy supply guiding the overall energy harvest of all fault-free smart sensors to converge to the maximum with the convergence rate no larger than$\kappa$,$\kappa\in[0,1]$, while enjoying the real-time fault-detection. An algorithm based on the sound theoretical foundations is further proposed to implement the optimal energy supply. Theoretical analysis, simulations and field experiments jointly verify the performance of our method. To our best knowledge, it is the initial work towards this issue.Note to Practitioners—This paper addresses the interesting issue of how to guarantee multiple smart sensors to harvest the most energy in the free-fault way. Through the insightful disclosure of relationship between the energy supply and sensor fault-free region, and the optimal control of energy supply, this paper facilitates the overall energy harvest of all fault-free smart sensors that work under the environments, where the available energy is limited, to converge to the maximum while enjoying the real-time fault-detection, which we believe could push the development of Internet of Things (IoT) or Cyber-Physical System (CPS) that employs multiple smart sensors to sense the physical world. Simulations and field experimental investigations jointly show that the proposed solution outperforms the existing solutions.

This paper studies the edge computing system (ECS) in which caching the frequently reusable service (FRS) at the edge server (ES) is an effective way to reduce delays. The larger cache space available in the ES (we call it ES cache space) might buffer the larger scale of FRS, and subsequently decrease the delays while bringing higher cache storage costs. Meanwhile, the distribution of FRS is not always known in advance. Therefore, how much ES cache space should be supplied to make the optimal tradeoff between the delays and cache storage costs arises as an interesting issue in practice. To address this issue, this paper first formulates the problem of determining the amount of ES cache space supply as a constrained Markov decision process (CMDP), then adopts the Zipf’s distribution to estimate the probability distribution of FRS, and finally proposes an effective cache space control algorithm (CSCA) guiding the ES to determine the amount of ES cache space supply to minimize the cache storage costs while maintaining the delays at the acceptable level. Theoretical analysis, simulations and field experiments document and illustrate its performance. Note to Practitioners—This paper addresses the interesting trade-off between the delays and cache storage costs for the edge computing system that operates with limited cache storage budgets while must satisfy the required real-time performances. It helps to improve the operation efficiency of the systems with edge computing setting in the area of Internet of Things (IoT) or Cyber-Physical Systems (CPS) that employ the edge server to cache the frequently reusable services, arriving at the minimization of the accumulative cache storage costs while maintaining the accumulative delays at the acceptable level. Theoretical analysis, simulation and field experimental investigations jointly show that the solution proposed here outperforms existing solutions.

This paper considers the cloud-based multi-agent system (MAS) in which for any agent to obtain the most trustworthy data (MTD) that best matches the agents’ personalized demands on the trusted attributes of data, the agents’ data usage often comes with cost, and the better the trustworthiness of data (TOD), the higher the cost of data usage (CDU). Therefore, under the environments where the budgets for data usage are constrained, there may exist a trade-off between the TOD and CDU for the cloud-based MAS, and how to optimally make such trade-off arises as an interesting issue in practice. To address this issue, this paper first formulates the trade-off as a mathematical constrained optimization problem, then explores out the theoretical foundations, both using the fuzzy similarity computing, and finally proposes an algorithm to guarantee the agents to obtain the MTD within the acceptable level of CDU. The experiment results illustrate its effectiveness. Note to Practitioners— This paper addresses the interesting trade-off between TOD and CDU for the cloud-based MAS that operates in the cloud environments where the TOD and CDU are positively correlated with each other. By establishing the mathematical constrained optimization model and designing the algorithm for the MTD recommendation within the acceptable CDU based on the theoretical foundations using fuzzy similarity computing, it helps the MAS with cloud settings to achieve the MTD that best matches their personalized demands on the trusted attributes within the acceptable CDU, which could push the development of cloud-based MAS by benefiting from the optimal trade-off between TOD and CDU. The experiments show that the solution proposed here outperforms existing solutions.

In a Web service-based multi-agent system (MAS), with the Web server collecting data from the agents periodically, the agents first submit their data to the Web server, and then, the Web server picks up the ones that match its demands on the quality of data (QoD) seems to be a practical way to arrive at the satisfactory data (SD). More data-submissions provide the Web server with more candidates to consider, increasing the probability of the Web server guaranteeing all the SDs, while causing more network traffic. Meanwhile, the heterogeneities of network traffic usage between different agents may lead to early network traffic exhaustion for some agents. Therefore, how to make the optimal tradeoff between network traffic and satisfaction with data (SWD) while balancing the network traffic arises as an interesting issue. To address this issue, this paper first formulates the QoD-evaluation model to classify the agents, then reveals the optimal probability for every agent to determine whether to submit the data to the server, and finally proposes an algorithm to make the optimal tradeoff. Theoretical results, simulations, and field experiments jointly verify its performance.

Wireless powered edge computing system (WPECS) enhances the computing power and extends the lifetime of wireless devices (WDs). This paper studies the WPECS with multiple WDs, in which the access point (AP) provides some transmission channels which differ from each other in the channel gain, and the WD powered through the wireless power transfer (WPT) technology has some indivisible tasks and adopts binary task-offloading actions. More energy harvested strengthens the WDs with more computing power, while corresponding to more energy consumption. Therefore, how to make the optimal tradeoff between computation rate and energy harvested arises as an interesting issue. To address this issue, this paper first formulates the switch process of transmission channel as a constrained Markov decision process (CMDP), and then proposed an effective algorithm to maximize the sum of computation rates of all WDs in terms of task data bits computed, within the required level of accumulative energy harvested. Theoretical analysis, simulations and field experiments jointly document and illustrate its performance. Note to Practitioners—This paper addresses the interesting tradeoff between computation rate and energy harvested in a wireless powered edge computing system that operates in the environments with limited available energy. It helps to improve the operation efficiency of the edge computing systems in the area of Internet of Things (IoT) or Cyber-Physical Systems (CPS) that employ wireless power transfer technology to power the wireless devices through the access point over the air to maximize the sum of computations rates of all WDs in terms of task data bits computed, while keeping the accumulative energy harvested within a range. Simulations and experimental investigations show that the solution proposed here outperforms existing solutions.

This paper considers the edge computing system (ECS) in which the tasks with dependencies are offloaded and computed in sequence. Different task-offloading orderings come with different ECS memory-cache execution latency (EMCEL) which is caused by writing and reading (WR) the computed results of earlier offloaded tasks between the ECS memory and cache. Therefore, the optimal ordering to offload all the tasks while leading to the minimum EMCEL arises as an interesting issue in practice. This requires to solve a hard exponential explosion optimization problem. To address this issue, this paper first formulates the tasks and their dependencies as a direct acyclic graph (DAG), then converts the exponential explosion problem into a discrete problem that can be solved in polynomial time, and finally develops some theoretical conditions to guide to determine the optimal task-offloading orderings. A novel algorithm called OTOOA to find the optimal task-offloading orderings in polynomial time is proposed. Field experiments show that OTOOA outperforms the existing algorithms. To our best knowledge, this is the initial work towards this issue. Note to Practitioners—For the edge computing system that operates in the application scenarios in which the ECS cache is small while the size of the tasks is relatively large such that it is not allowed for multiple tasks to be processed in the ECS cache parallelly or at the same time, e.g., the execution latency-sensitive and fast big data-processing scenarios in which the multiple tasks depending on the other are offloaded and computed in sequence, this paper helps such edge computing system to improve the operation efficiency with the minimum EMCEL by finding an optimal task-offloading ordering to guide the wireless devices to offload their tasks to the ECS server. Experimental investigations show that the solution proposed here outperforms existing ones.

This paper considers the stopping-time control of multiple smart sensors which independently sample the physical parameter of interests and request the desired data from the common server with probabilities. More samples may lead to higher sensing accuracy but cost more energy. More requests may facilitate the smart sensors to obtain the desired data with higher probability while yet cost more energy. Meanwhile, the heterogeneities of smart sensors issuing the requests may lead to early energy exhaustion for some smart sensors. Therefore, how to make an optimal tradeoff between sensing accuracy and energy consumption while keeping the energy consumption balance arises as an interesting problem. To address this issue, this paper first formulates the stopping-time control policy of an individual smart sensor as a partially observable Markov decision process (POMDP), then coordinates the stopping-time control policies of multiple smart sensors within a noncooperative game (NCG), and finally proposes a noncooperative PODMP game-based algorithm to make the above tradeoff. Theoretical analysis and field experiments jointly document the performance. Note to Practitioners—This paper addresses the interesting tradeoff between sensing accuracy and energy cost while balancing the energy consumption in a sensor network that operates in the environments where the available energy is limited and the early energy exhaustion for some of the smart sensors should be avoided. It helps to improve the operation efficiency of the sensor network with multiple smart sensors and a shared server setting in the area of Internet of Things (IoT) or Cyber-Physical Systems (CPS) to minimize the estimation error, while keeping the accumulative energy cost within a range and balancing the energy consumption. Field experimental investigations show that the solution proposed here outperforms existing solutions.

In a Web service-based sampling system (WSSS) with multiple smart sensors, where the smart sensors request the sampling-commands (SCs) from the Web server with polling mechanism, and the Web server picks up the targeted estimates (TEs) from all the estimates received, higher polling frequency and more estimate-submissions enable the Web server to obtain all TEs (ATE) with larger probability, enjoying less delays, while suffering more network traffic. Therefore, how to minimize the network traffic while satisfying the acceptable delays and probability of the server obtaining ATE arises as an interesting issue. To address this issue, this paper first proposes a Markov decision process (MDP) framework to formulate the request process of WSSS, then explores out the optimal polling frequency and probability in the polynomial time for the smart sensors' requests, and finally proposes a probability threshold-based algorithm to guide the smart sensors to decide the polling frequency and whether to request the SCs. Theoretical results, simulations, and field experiments document and illustrate its performance.

This article studies the refactoring problem of smart sensors whose energy supply is limited. Refactoring helps smart sensors to wipe out code bugs (CBs) while consuming its limited energy. A smart sensor may need more than one refactoring. The more energy consumed by a refactoring, the more CBs may be removed by this refactoring, while the less energy will be left for the other refactorings (subsequently, less CBs may be reduced by the other refactorings). Therefore, how to make the optimal tradeoff for smart sensors between CB ratio (CBR) and energy consumption arises as an interesting problem. To address this problem, this article first establishes a constrained optimization-theoretical framework that can allocate the limited energy for refactorings, second explores out the optimal amount of energy supply for each refactoring that can minimize the CBR, third discovers the theoretical value, lower bound, upper bound, infimum, supremum, and convergence properties of the minimum CBR, and finally proposes an effective algorithm to minimize the CBR within the given energy level. To our best knowledge, this is the initial work toward this issue. Simulation and field experiments both document the performance.

In no-reference 360-degree image quality assessment (NR 360IQA), graph convolutional networks (GCNs), which model interactions between viewports through graphs, have achieved impressive performance. However, prevailing GCN-based NR 360IQA methods suffer from three main limitations. First, they only use high-level features of the distorted image to regress the quality score, while the human visual system scores the image based on hierarchical features. Second, they simplify complex high-order interactions between viewports in a pairwise fashion through graphs. Third, in the graph construction, they only consider the spatial location of the viewport, ignoring its content characteristics. Accordingly, to address these issues, we propose an adaptive hypergraph convolutional network for NR 360IQA, denoted as AHGCN. Specifically, we first design a multi-level viewport descriptor for extracting hierarchical representations from viewports. Then, we model interactions between viewports through hypergraphs, where each hyperedge connects two or more viewports. In the hypergraph construction, we build a location-based hyperedge and a content-based hyperedge for each viewport. Experimental results on two public 360IQA databases demonstrate that our proposed approach has a clear advantage over state-of-the-art full-reference and no-reference IQA models.

Diagnosis and treatment for Parkinson's disease rely on the evaluation of motor functions, which is expensive and time consuming when performing at clinics. It is also difficult for patients to record correct movements at home without the guidance from experienced physicians. To help patients with Parkinson’s disease get better evaluation from in-home recorded movement videos, we developed an interactive video acquisition and learning system for clinical motor assessments. The system provides real-time guidance with multi-level body keypoint tracking and analysis to patients, which guarantees correct understanding and performing of clinical tasks. We tested its effectiveness on healthy subjects, and the efficiency and usability on patient groups. Experiments showed that our system enabled high quality video recordings following clinical standards, benefiting both patients and physicians. Our system provides a novel learning-based telemedicine approach for the care of patients with Parkinson’s disease.

Learning-based image deraining methods have achieved remarkable success in the past few decades. Currently, most deraining architectures are developed by human experts, which is a laborious and error-prone process. In this paper, we present a study on employing neural architecture search (NAS) to automatically design deraining architectures, dubbed AutoDerain. Specifically, we first propose an U-shaped deraining architecture, which mainly consists of residual squeeze-and-excitation blocks (RSEBs). Then, we define a search space, where we search for the convolutional types and the use of the squeeze-and-excitation block. Considering that the differentiable architecture search is memory-intensive, we propose a memory-efficient differentiable architecture search scheme (MDARTS). In light of the success of training binary neural networks, MDARTS optimizes architecture parameters through the proximal gradient, which only consumes the same GPU memory as training a single deraining model. Experimental results demonstrate that the architecture designed by MDARTS is superior to manually designed derainers.

Recently, reinforced adaptive bitrate (ABR) algorithms have achieved remarkable success in tile-based 360-degree video streaming. However, they heavily rely on accurate viewport prediction. To alleviate this issue, we propose a hierarchical reinforcement-learning (RL) based ABR algorithm, dubbed 360HRL. Specifically, 360HRL consists of a top agent and a bottom agent. The former is used to decide whether to download a new segment for continuous playback or re-download an old segment for correcting wrong bitrate decisions caused by inaccurate viewport estimation, and the latter is used to select bitrates for tiles in the chosen segment. In addition, 360HRL adopts a two-stage training methodology. In the first stage, the bottom agent is trained under the environment where the top agent always chooses to download a new segment. In the second stage, the bottom agent is fixed and the top agent is optimized with the help of a heuristic decision rule. Experimental results demonstrate that 360HRL outperforms existing RL-based ABR algorithms across a broad of network conditions and quality of experience (QoE) objectives.

Web services integrate various components in the Internet of Things (IoT). In a Web service-based data-collection system with multiple smart sensor nodes periodically sampling and estimating the same unknown physical parameter of interest, the smart sensor nodes first submit their estimates to the Web server, and then, the server picking the one with the minimum error seems to be a practical way to arrive at a minimum error estimate (MEE). More submissions provide the Web server with more candidates to consider, which can maximize the probability of the server guaranteeing the MEE, while also leading to more network traffic. Therefore, how to make the optimal tradeoff between network traffic and sensing accuracy arises as an interesting problem. This article proposes a network traffic-dependent probability threshold policy within an intended underlying optimization-theoretical framework to address this problem. The policy is such that the smart sensor nodes submit their estimates and corresponding estimation errors (ECEEs) to the Web server within a tolerable network traffic threshold while maximizing the probability of the server delivering the MEE. Theoretical analysis, simulation, and field experiments document and illustrate its performance.Note to Practitioners—This article addresses the interesting tradeoff between sensing accuracy and network traffic demand in the Web service-based data-collection system that operates in some remote areas with limited network traffic. It helps to improve the operation efficiency of the Internet-of-Things (IoT) systems that employ Web service technology to enable the Web server to deliver minimum error estimate with maximum probability while keeping the network traffic within a given range. Our simulation and experimental investigations show that the solution developed here outperforms existing solutions.

In Internet of Things systems, Web services enable interoperable machine-to-machine communication over networks. Polling mechanism is a practical way for a Web service-based control system to enable its actuator to respond to its controller under uncertain environments where the actuator does not know exactly when the controller updates its command. Fast response demands a high polling frequency (polling mechanism handles the event in time-driven mode) which may lead to heavy network traffic. Therefore, how to make the optimal tradeoff between the network traffic and delay in a Web service-based control system becomes an interesting problem. This paper formulates the problem of finding the optimal polling frequency control policy of a Web service-based control system as a constrained Markov decision process (CMDP). The policy is such that the actuator responds to the controller within a tolerable delay threshold while minimizing network traffic. An algorithm called CMDPA is proposed to solve the problem. Simulation and field experiments show our policy performance.

This paper concerns stopping-time management of smart sensing nodes, a kind of very large scale integration system, based on the tradeoffs between sensing accuracy and power consumption, which are foundations of Internet of Things systems and cyber-physical systems. In practice, smart sensing nodes work periodically, and more samples leads to higher accuracy but more energy cost, so when to stop sampling in a periodic cycle to achieve the optimal tradeoff is an interesting issue. This paper formulates this issue as a partially observable Markov decision process (POMDP) and develops a POMDP-based Optimal Stopping-time Algorithm to make the above tradeoff. Field experiments demonstrate its performance.

For embedded systems with multiple sleep modes, it is interesting to understand how to maximize the energy saving potential by choosing the suitable sleep mode(s) during the idle period. In this paper, we establish a sufficient condition to narrow down the search space of sleep policy and propose a new algorithm: optimal-idle-threshold-policy-algorithm under more realistic setting than the existing works. Theoretical proofs and experimental results justify the benefits of our approach.

The Internet of Things (IoT) has received an increasing attention in recent years. Embedded intelligent terminal (EIT), an indispensable part of IoT, works not only as a sensor but also as a primary processor. Due to the limited power resource of EIT, it is important to study how to improve the efficiency of its power use. To tackle this problem, we propose an energy-saving algorithm, Bayesian idle time prediction (BIP). The basic idea of BIP is to explore historical information and obtain a better estimation of idle time. In this paper, we provide a theoretical analysis of BIP and compare our method with three existing algorithms [weighted idle-time-prediction (IP) algorithm, IP algorithm, and running time fixed threshold in IP algorithm] with respect to energy-saving potential, as well as system delay under a random number of tasks. Both simulation and field experiment results demonstrate the advantages of our algorithm in energy saving.

The space-filling fractal network in the human lung creates a remarkable distribution system for gas exchange. Landmark studies have illuminated how the fractal network guarantees minimum energy dissipation, slows air down with minimum hardware, maximizes the gas- exchange surface area, and creates respiratory flexibility between rest and exercise. In this paper, we investigate how the fractal architecture affects oxygen transport and exchange under varying physiological conditions, with respect to performance metrics not previously studied.We present a renormalization treatment of the diffusion-reaction equation which describes how oxygen concentrations drop in the airways as oxygen crosses the alveolar membrane system. The treatment predicts oxygen currents across the lung at different levels of exercise which agree with measured values within a few percent. The results exhibit wide-ranging adaptation to changing process parameters, including maximum oxygen uptake rate at minimum alveolar membrane permeability, the ability to rapidly switch from a low oxygen uptake rate at rest to high rates at exercise, and the ability to maintain a constant oxygen uptake rate in the event of a change in permeability or surface area. We show that alternative, less than space-filling architectures perform sub-optimally and that optimal performance of the space-filling architecture results from a competition between underexploration and overexploration of the surface by oxygen molecules.