arXiv subjects

Yuanqing Xia

Publications and source records attributed to Yuanqing Xia.

At least 37 records · Page 2Linked to original sources

arXiv2024

Arena: A Patch-of-Interest ViT Inference Acceleration System for Edge-Assisted Video Analytics

The advent of edge computing has made real-time intelligent video analytics feasible. Previous works, based on traditional model architecture (e.g., CNN, RNN, etc.), employ various strategies to filter out non-region-of-interest content to minimize bandwidth and computation consumption but show inferior performance in adverse environments. Recently, visual foundation models based on transformers have shown great performance in adverse environments due to their amazing generalization capability. However, they require a large amount of computation power, which limits their applications in real-time intelligent video analytics. In this paper, we find visual foundation models like Vision Transformer (ViT) also have a dedicated acceleration mechanism for video analytics. To this end, we introduce Arena, an end-to-end edge-assisted video inference acceleration system based on ViT. We leverage the capability of ViT that can be accelerated through token pruning by only offloading and feeding Patches-of-Interest to the downstream models. Additionally, we design an adaptive keyframe inference switching algorithm tailored to different videos, capable of adapting to the current video content to jointly optimize accuracy and bandwidth. Through extensive experiments, our findings reveal that Arena can boost inference speeds by up to 1.58$\times$ and 1.82$\times$ on average while consuming only 47\% and 31\% of the bandwidth, respectively, all with high inference accuracy.

cs.MM↗

arXiv2024

Cooperative Online Learning for Multi-Agent System Control via Gaussian Processes with Event-Triggered Mechanism: Extended Version

In the realm of the cooperative control of multi-agent systems (MASs) with unknown dynamics, Gaussian process (GP) regression is widely used to infer the uncertainties due to its modeling flexibility of nonlinear functions and the existence of a theoretical prediction error bound. Online learning, which involves incorporating newly acquired training data into Gaussian process models, promises to improve control performance by enhancing predictions during the operation. Therefore, this paper investigates the online cooperative learning algorithm for MAS control. Moreover, an event-triggered data selection mechanism, inspired by the analysis of a centralized event-trigger, is introduced to reduce the model update frequency and enhance the data efficiency. With the proposed learning-based control, the practical convergence of the MAS is validated with guaranteed tracking performance via the Lynaponve theory. Furthermore, the exclusion of the Zeno behavior for individual agents is shown. Finally, the effectiveness of the proposed event-triggered online learning method is demonstrated in simulations.

eess.SY↗

arXiv2024

Minimal Sensor Placement for Generic State and Unknown Input Observability

This paper addresses the problem of selecting the minimum number of dedicated sensors to achieve observability in the presence of unknown inputs, namely, the state and input observability, for linear time-invariant systems. We assume that the only available information is the zero-nonzero structure of system matrices, and approach this problem within a structured system model. We revisit the concept of state and input observability for structured systems, providing refined necessary and sufficient conditions for placing dedicated sensors via the Dulmage-Mendelsohn decomposition. Based on these conditions, we prove that determining the minimum number of dedicated sensors to achieve generic state and input observability is NP-hard, which contrasts sharply with the polynomial-time complexity of the corresponding problem with known inputs. We also demonstrate that this problem is hard to approximate within a factor of $(1-o(1)){\rm{log}}(n)$, where $n$ is the state dimension. Notwithstanding, we propose nontrivial upper and lower bounds that can be computed in polynomial time, which confine the optimal value of this problem to an interval with length being the number of inputs. We further present a special case for which the exact optimal value can be determined in polynomial time. Additionally, we propose a two-stage algorithm to solve this problem approximately. Each stage of the algorithm is either optimal or suboptimal and can be completed in polynomial time.

eess.SY↗

arXiv2024

Learning-based Parameterized Barrier Function for Safety-Critical Control of Unknown Systems

With the increasing complexity of real-world systems and varying environmental uncertainties, it is difficult to build an accurate dynamic model, which poses challenges especially for safety-critical control. In this paper, a learning-based control policy is proposed to ensure the safety of systems with unknown disturbances through control barrier functions (CBFs). First, the disturbance is predicted by Gaussian process (GP) regression, whose prediction performance is guaranteed by a deterministic error bound. Then, a novel control strategy using GP-based parameterized high-order control barrier functions (GP-P-HOCBFs) is proposed via a shrunk original safe set based on the prediction error bound. In comparison to existing methods that involve adding strict robust safety terms to the HOCBF condition, the proposed method offers more flexibility to deal with the conservatism and the feasibility of solving quadratic problems within the CBF framework. Finally, the effectiveness of the proposed method is demonstrated by simulations on Franka Emika manipulator.

eess.SY↗

arXiv2024

Privacy and Security Trade-off in Interconnected Systems with Known or Unknown Privacy Noise Covariance

This paper is concerned with the security problem for interconnected systems, where each subsystem is required to detect local attacks using locally available information and the information received from its neighboring subsystems. Moreover, we consider that there exists an additional eavesdropper being able to infer the private information by eavesdropping transmitted data between subsystems. Then, a privacy-preserving method is employed by adding privacy noise to transmitted data, and the privacy level is measured by mutual information. Nevertheless, adding privacy noise to transmitted data may affect the detection performance metrics such as detection probability and false alarm probability. Thus, we theoretically analyze the trade-off between the privacy and the detection performance. An optimization problem with maximizing both the degree of privacy preservation and the detection probability is established to obtain the covariance of the privacy noise. In addition, the attack detector of each subsystem may not obtain all information about the privacy noise. We further theoretically analyze the trade-off between the privacy and the false alarm probability when the attack detector has no knowledge of the privacy noise covariance. An optimization problem with maximizing the degree of privacy preservation with guaranteeing a bound of false alarm distortion level is established to obtain {\color{black}{the covariance of the privacy noise}}. Moreover, to analyze the effect of the privacy noise on the detection probability, we consider that each subsystem can estimate the unknown privacy noise covariance by the secondary data. Based on the estimated covariance, we construct another attack detector and analyze how the privacy noise affects its detection performance. Finally, a numerical example is provided to verify the effectiveness of theoretical results.

eess.SY↗

arXiv2024

Generalized Cactus and Structural Controllability of Switched Linear Continuous-Time Systems

This paper explores the structural controllability of switched linear continuous-time systems. It first identifies a gap in the proof for a pivotal criterion for the structural controllability of switched linear systems in the literature. To address this void, we develop novel graph-theoretic concepts, such as multi-layer dynamic graphs, generalized stems/buds, and generalized cacti, and based on them, provide a comprehensive proof for this criterion. Our approach also induces a new, generalized cactus based graph-theoretic criterion for structural controllability. This not only extends Lin's cactus-based graph-theoretic condition to switched systems for the first time, but also provides a lower bound for the generic dimension of controllable subspaces.

eess.SY↗

arXiv2024

Egret: Reinforcement Mechanism for Sequential Computation Offloading in Edge Computing

As an emerging computing paradigm, edge computing offers computing resources closer to the data sources, helping to improve the service quality of many real-time applications. A crucial problem is designing a rational pricing mechanism to maximize the revenue of the edge computing service provider (ECSP). However, prior works have considerable limitations: clients are static and are required to disclose their preferences, which is impractical in reality. However, previous works assume user privacy information to be known or consider the number of users in edge scenarios to be static. To address this issue, we propose a novel sequential computation offloading mechanism, where the ECSP posts prices of computing resources with different configurations to clients in turn. Clients independently choose which computing resources to purchase and how to offload based on their prices. Then Egret, a deep reinforcement learning-based approach that achieves maximum revenue, is proposed. Egret determines the optimal price and visiting orders online without considering clients' preferences. Experimental results show that the revenue of ECSP in Egret is only 1.29\% lower than Oracle and 23.43\% better than the state-of-the-art when the client arrives dynamically.

cs.DC↗

arXiv2024

Tangram: High-resolution Video Analytics on Serverless Platform with SLO-aware Batching

Cloud-edge collaborative computing paradigm is a promising solution to high-resolution video analytics systems. The key lies in reducing redundant data and managing fluctuating inference workloads effectively. Previous work has focused on extracting regions of interest (RoIs) from videos and transmitting them to the cloud for processing. However, a naive Infrastructure as a Service (IaaS) resource configuration falls short in handling highly fluctuating workloads, leading to violations of Service Level Objectives (SLOs) and inefficient resource utilization. Besides, these methods neglect the potential benefits of RoIs batching to leverage parallel processing. In this work, we introduce Tangram, an efficient serverless cloud-edge video analytics system fully optimized for both communication and computation. Tangram adaptively aligns the RoIs into patches and transmits them to the scheduler in the cloud. The system employs a unique ``stitching'' method to batch the patches with various sizes from the edge cameras. Additionally, we develop an online SLO-aware batching algorithm that judiciously determines the optimal invoking time of the serverless function. Experiments on our prototype reveal that Tangram can reduce bandwidth consumption and computation cost up to 74.30\% and 66.35\%, respectively, while maintaining SLO violations within 5\% and the accuracy loss negligible.

cs.DC↗

arXiv2024

Distributed Adaptive Gradient Algorithm with Gradient Tracking for Stochastic Non-Convex Optimization

This paper considers a distributed stochastic non-convex optimization problem, where the nodes in a network cooperatively minimize a sum of $L$-smooth local cost functions with sparse gradients. By adaptively adjusting the stepsizes according to the historical (possibly sparse) gradients, a distributed adaptive gradient algorithm is proposed, in which a gradient tracking estimator is used to handle the heterogeneity between different local cost functions. We establish an upper bound on the optimality gap, which indicates that our proposed algorithm can reach a first-order stationary solution dependent on the upper bound on the variance of the stochastic gradients. Finally, numerical examples are presented to illustrate the effectiveness of the algorithm.

math.OC↗

arXiv2024

Perturbation-Tolerant Structural Controllability for Linear Systems

This paper proposes a notion termed perturbation-tolerant structural controllability (PTSC) to study the generic property of controllability preservation/resilience of structured linear systems under structured perturbations. A structured system is said to be PTSC with respect to a perturbation structure if for almost all controllable realizations of this system, there are no complex-valued perturbations obeying the zero/nonzero pattern prescribed by the perturbation structure that can make the perturbed systems uncontrollable. We prove a generic property in this notion, that for almost all controllable realizations of a structured system, either there exist such structured perturbations rendering the systems uncontrollable, or there are no such perturbations. We present a decomposition-based necessary and sufficient condition for the PTSC of single-input linear systems, whose verification has polynomial time complexity. We then discuss some intuitive graph-theoretic conditions for PTSC. As an application, our results can serve as some feasibility conditions for the conventional structured controllability radius problems from a generic viewpoint.

math.OC↗

arXiv2024

KCES: A Workflow Containerization Scheduling Scheme Under Cloud-Edge Collaboration Framework

As more IoT applications gradually move towards the cloud-edge collaborative mode, the containerized scheduling of workflows extends from the cloud to the edge. However, given the high delay of the communication network, loose coupling of structure, and resource heterogeneity between cloud and edge, workflow containerization scheduling in the cloud-edge scenarios faces the difficulty of resource coordination and application collaboration management. To address these two issues, we propose a KubeEdge-Cloud-Edge-Scheduling scheme named KCES, a workflow containerization scheduling scheme for the KubeEdge cloud-edge framework. The KCES includes a cloud-edge workflow scheduling engine for KubeEdge and workflow scheduling strategies for task horizontal roaming and vertical offloading. Considering the scheduling optimization of cloud-edge workflows, this paper proposes a cloud-edge workflow scheduling model and cloud-edge node model and designs a cloud-edge workflow scheduling engine to maximize cloud-edge resource utilization under the constraint of workflow task delay. A cloud-edge resource hybrid management technology is used to design the cloud-edge resource evaluation and resource allocation algorithms to achieve cloud-edge resource collaboration. Based on the ideas of distributed functional roles and the hierarchical division of computing power, the horizontal roaming among the edges and vertical offloading strategies between the cloud and edges for workflow tasks are designed to realize the cloud-edge application collaboration. Through a customized IoT application workflow instance, experimental results show that KCES is superior to the baseline in total workflow duration, average workflow duration, and resource usage and has the capabilities of horizontal roaming and vertical offloading for workflow tasks.

cs.DC↗

arXiv2023

Security Defense of Large Scale Networks Under False Data Injection Attacks: An Attack Detection Scheduling Approach

In large-scale networks, communication links between nodes are easily injected with false data by adversaries. This paper proposes a novel security defense strategy from the perspective of attack detection scheduling to ensure the security of the network. Based on the proposed strategy, each sensor can directly exclude suspicious sensors from its neighboring set. First, the problem of selecting suspicious sensors is formulated as a combinatorial optimization problem, which is non-deterministic polynomial-time hard (NP-hard). To solve this problem, the original function is transformed into a submodular function. Then, we propose an attack detection scheduling algorithm based on the sequential submodular optimization theory, which incorporates \emph{expert problem} to better utilize historical information to guide the sensor selection task at the current moment. For different attack strategies, theoretical results show that the average optimization rate of the proposed algorithm has a lower bound, and the error expectation is bounded. In addition, under two kinds of insecurity conditions, the proposed algorithm can guarantee the security of the entire network from the perspective of the augmented estimation error. Finally, the effectiveness of the developed method is verified by the numerical simulation and practical experiment.

eess.SY↗

arXiv2023

Arena: A Learning-based Synchronization Scheme for Hierarchical Federated Learning--Technical Report

Federated learning (FL) enables collaborative model training among distributed devices without data sharing, but existing FL suffers from poor scalability because of global model synchronization. To address this issue, hierarchical federated learning (HFL) has been recently proposed to let edge servers aggregate models of devices in proximity, while synchronizing via the cloud periodically. However, a critical open challenge about how to make a good synchronization scheme (when devices and edges should be synchronized) is still unsolved. Devices are heterogeneous in computing and communication capability, and their data could be non-IID. No existing work can well synchronize various roles (\textit{e.g.}, devices and edges) in HFL to guarantee high learning efficiency and accuracy. In this paper, we propose a learning-based synchronization scheme for HFL systems. By collecting data such as edge models, CPU usage, communication time, \textit{etc}., we design a deep reinforcement learning-based approach to decide the frequencies of cloud aggregation and edge aggregation, respectively. The proposed scheme well considers device heterogeneity, non-IID data and device mobility, to maximize the training model accuracy while minimizing the energy overhead. Meanwhile, the convergence bound of the proposed synchronization scheme has been analyzed. And we build an HFL testbed and conduct the experiments with real data obtained from Raspberry Pi and Alibaba Cloud. Extensive experiments under various settings are conducted to confirm the effectiveness of \textit{Arena}.

cs.DC↗

arXiv2023

On real structured controllability/stabilizability/stability radius: Complexity and unified rank-relaxation based methods

This paper addresses the real structured controllability, stabilizability, and stability radii (RSCR, RSSZR, and RSSR, respectively) of linear systems, which involve determining the distance (in terms of matrix norms) between a (possibly large-scale) system and its nearest uncontrollable, unstabilizable, and unstable systems, respectively, with a prescribed affine structure. This paper makes two main contributions. First, by demonstrating that determining the feasibilities of RSCR and RSSZR is NP-hard when the perturbations have a general affine parameterization, we prove that computing these radii is NP-hard. Additionally, we prove the NP-hardness of a problem related to the RSSR. These hardness results are independent of the matrix norm used. Second, we develop unified rank-relaxation based algorithms for these problems, which can handle both the Frobenius norm and the $2$-norm based problems and share the same framework for the RSCR, RSSZR, and RSSR problems. These algorithms utilize the low-rank structure of the original problems and relax the corresponding rank constraints with a regularized truncated nuclear norm term. Moreover, a modified version of these algorithms can find local optima with performance specifications on the perturbations, under appropriate conditions. Finally, simulations suggest that the proposed methods, despite being in a simple framework, can find local optima as good as several existing methods.

eess.SY↗

arXiv2023

Observability Blocking for Functional Privacy of Linear Dynamic Networks

This paper addresses the problem of determining the minimum set of state variables in a network that need to be blocked from direct measurements in order to protect functional privacy with respect to {\emph{any}} output matrices. The goal is to prevent adversarial observers or eavesdroppers from inferring a linear functional of states, either vector-wise or entry-wise. We prove that both problems are NP-hard. However, by assuming a reasonable constant bound on the geometric multiplicities of the system's eigenvalues, we present an exact algorithm with polynomial time complexity for the vector-wise functional privacy protection problem. Based on this algorithm, we then provide a greedy algorithm for the entry-wise privacy protection problem. Our approach is based on relating these problems to functional observability and leveraging a PBH-like criterion for functional observability. Finally, we provide an example to demonstrate the effectiveness of our proposed approach.

eess.SY↗

arXiv2023

On the Reachability and Controllability of Temporal Continuous-Time Linear Networks: A Generic Analysis

Temporal networks are a class of time-varying networks, which change their topology according to a given time-ordered sequence of static networks (known as subsystems). This paper investigates the reachability and controllability of temporal continuous-time linear networks from a generic viewpoint, where only the zero-nonzero patterns of subsystem matrices are known. We demonstrate that the reachability and controllability on a single temporal sequence are generic properties with respect to the parameters of subsystem matrices and the time durations of subsystems. We then give explicit expressions for the minimal subspace that contains the reachable set across all possible temporal sequences (called overall reachable set). It is found that verifying the structural reachability/controllability and structural overall reachability are at least as hard as the structural target controllability verification problem of a single system, implying that finding verifiable conditions for them is hard. Graph-theoretic lower and upper bounds are provided for the generic dimensions of the reachable subspace on a single temporal sequence and of the minimal subspace that contains the overall reachable set. These bounds extend classical concepts in structured system theory, including the dynamic graph and the cactus, to temporal networks, and can be efficiently calculated using graph-theoretic algorithms. Finally, applications of the results to the structural controllability of switched linear systems are discussed.

eess.SY↗

arXiv2023

Encrypted Data-driven Predictive Cloud Control with Disturbance Observer

In data-driven predictive cloud control tasks, the privacy of data stored and used in cloud services could be leaked to malicious attackers or curious eavesdroppers. Homomorphic encryption technique could be used to protect data privacy while allowing computation. However, extra errors are introduced by the homomorphic encryption extension to ensure the privacy-preserving properties, and the real number truncation also brings uncertainty. Also, process and measure noise existed in system input and output may bring disturbance. In this work, a data-driven predictive cloud controller is developed based on homomorphic encryption to protect the cloud data privacy. Besides, a disturbance observer is introduced to estimate and compensate the encrypted control signal sequence computed in the cloud. The privacy of data is guaranteed by encryption and experiment results show the effect of our cloud-edge cooperative design.

eess.SY↗

arXiv2023

Adaptive Resource Allocation for Workflow Containerization on Kubernetes

In a cloud-native era, the Kubernetes-based workflow engine enables workflow containerized execution through the inherent abilities of Kubernetes. However, when encountering continuous workflow requests and unexpected resource request spikes, the engine is limited to the current workflow load information for resource allocation, which lacks the agility and predictability of resource allocation, resulting in over and under-provisioning resources. This mechanism seriously hinders workflow execution efficiency and leads to high resource waste. To overcome these drawbacks, we propose an adaptive resource allocation scheme named ARAS for the Kubernetes-based workflow engines. Considering potential future workflow task requests within the current task pod's lifecycle, the ARAS uses a resource scaling strategy to allocate resources in response to high-concurrency workflow scenarios. The ARAS offers resource discovery, resource evaluation, and allocation functionalities and serves as a key component for our tailored workflow engine (KubeAdaptor). By integrating the ARAS into KubeAdaptor for workflow containerized execution, we demonstrate the practical abilities of KubeAdaptor and the advantages of our ARAS. Compared with the baseline algorithm, experimental evaluation under three distinct workflow arrival patterns shows that ARAS gains time-saving of 9.8% to 40.92% in the average total duration of all workflows, time-saving of 26.4% to 79.86% in the average duration of individual workflow, and an increase of 1% to 16% in CPU and memory resource usage rate.

cs.DC↗