Task Latency Research Articles

A joint strategy for service deployment and task offloading in satellite–terrestrial IoT

In recent years, low earth orbit satellite constellations, which are an important component of 6G, have been considered as a potential solution to achieve seamless network services for remote areas. Service deployment based on network function virtualization (NFV) has become an essential trend among satellite networks to enable flexible network services. However, current satellite–terrestrial IoT task offloading schemes rarely consider NFV-based satellite service deployment, which limits the performance of satellite networks. In this study, we address this problem by proposing an optimization problem that jointly considers service deployment and task offloading. To solve such a problem with many coupling decision variables, we decouple the problem using a two-stage approach. We propose a deep reinforcement learning-based service deployment policy to solve the service deployment subproblem and an alternating direction multiplier method-based distributed approach to solve the task offloading subproblem, which with the aim of minimizing the task latency and energy consumption of IoT devices. Simulation experiments demonstrate that our scheme can obtain near-optimal solution and can be adapted to large-scale satellite–terrestrial IoT network scenarios.

A green, secure, and deep intelligent method for dynamic IoT-edge-cloud offloading scenarios

To fulfill people's expectations for smart and user-friendly Internet of Things (IoT) applications, the quantity of processing is fast expanding, and task latency constraints are becoming extremely rigorous. On the other hand, the limited battery capacity of IoT objects severely affects the user experience. Energy Harvesting (EH) technology enables green energy to offer a continuous energy supply for IoT objects. It provides a solid assurance for the proper functioning of resource-constrained IoT objects when combined with the maturation of edge platforms and the development of parallel computing. The Markov Decision Process (MDP) and Deep Learning (DL) are used in this work to solve dynamic online/offline IoT-edge offloading scenarios. The suggested system may be used in both offline and online contexts and meets the user's quality of service expectations. Also, we investigate a blockchain scenario in which edge and cloud could work toward task offloading to address the tradeoff between limited processing power and high latency while ensuring data integrity during the offloading process. We provide a double Q-learning solution to the MDP issue that maximizes the acceptable offline offloading methods. During exploration, Transfer Learning (TL) is employed to quicken convergence by reducing pointless exploration. Although the recently promoted Deep Q-Network (DQN) may address this space complexity issue by replacing the huge Q-table in standard Q-learning with a Deep Neural Network (DNN), its learning speed may still be insufficient for IoT apps. In light of this, our work introduces a novel learning algorithm known as deep Post-Decision State (PDS)-learning, which combines the PDS-learning approach with the classic DQN. The system component in the proposed system can be dynamically chosen and modified to decrease object energy usage and delay. On average, the proposed technique outperforms multiple benchmarks in terms of delay by 4.5%, job failure rate by 5.7%, cost by 4.6%, computational overhead by 6.1%, and energy consumption by 3.9%.

Genetic algorithm for delay efficient computation offloading in dispersed computing

As a promising computing paradigm, dispersed computing uses all devices with computing, communication, and storage capabilities as ubiquitous network computing points (NCPs) to achieve low-latency, efficient collaborative computation for tasks, which greatly supports the newly arising computationally intensive applications. However, how to fully utilize the limited resources of NCPs to reduce task latency remains a challenging problem. Considering the heterogeneous communication modes and computing capabilities of NCPs, a distributed multi-hop computing task offloading framework based on an improved genetic algorithm is proposed to address this challenge, where tasks can be recursively offloaded among NCPs. The algorithm reduces the solution space dimension by designing filter chains to filter schedulable nodes before the initialization phase, and improves the population initialization, crossover operator to speed up the convergence, and avoids the risk of overconsumption of resources due to circular scheduling. Compared with the D2D-Fogging and Min–Min algorithms, the experimental results show that our scheduling algorithm improves the global resource utilization by 14% and 8%, and reduces the average task computation delay by 39.18% and 28.21%.

Metformin mitigates amyloid β1-40-induced cognitive decline via attenuation of oxidative/nitrosative stress and neuroinflammation.

Metformin is an antidiabetic medicine widely used for management of type 2 diabetes with neuroprotective effects and promising potential to attenuate cognitive impairment. The efficacy of metformin in attenuation of Alzheimer's disease (AD) pathology has not been well-documented. Thus, this study was designed to assess protective effect of metformin against Aβ1-40-instigared cognitive impairment. After intra-CA1 microinjection of aggregated Aβ1-40, rats received oral metformin (50 and/or 200mg/kg/day) for two weeks. Cognition function was analyzed in various behavioral tasks besides measurement of hippocampal oxidative stress, apoptosis, and inflammation along with H&E staining and 3-nitrotyrosine (3-NT) immunohistochemistry. Obtained data showed significant improvement of discrimination score in novel object recognition test, higher alternation score in Y maze, greater latency in passive avoidance task, and lower working and reference memory errors in radial arm maze in metformin-treated Aβ-injured group. Moreover, metformin treatment attenuated hippocampal levels of nitrite, MDA, protein carbonyl, ROS, TNFα, GFAP, DNA fragmentation intensity, caspase 3 activity, AChE activity, and increased SOD activity and level of IL-10 as an anti-inflammatory factor. In addition, metformin treatment was associated with lower CA1 neuronal loss and it also decreased intensity of 3-NT immunoreactivity as an indicator of nitrosative stress. Taken together, obtained findings showed neuroprotective and anti-dementia property of metformin in male rats and this may have potential benefit in attenuation of cognitive decline and related complications in patients with neurodegenerative disorders such as AD besides diabetes mellitus.

Joint Communication and Computation Offloading for Ultra-Reliable and Low-Latency With Multi-Tier Computing

In this paper, we study joint communication and computation offloading (JCCO) for hierarchical edge-cloud systems with ultra-reliable and low latency communications (URLLC). We aim to minimize the end-to-end (e2e) latency of computational tasks among multiple industrial Internet of Things (IIoT) devices by jointly optimizing offloading probabilities, processing rates, user association policies and power control subject to their service delay and energy consumption requirements as well as queueing stability conditions. The formulated JCCO problem belongs to a difficult class of mixed-integer non-convex optimization problem, making it computationally intractable. In addition, a strong coupling between binary and continuous variables and the large size of hierarchical edge-cloud systems make the problem even more challenging to solve optimally. To address these challenges, we first decompose the original problem into two subproblems based on the unique structure of the underlying problem and leverage the alternating optimization (AO) approach to solve them in an iterative fashion by developing newly convex approximate functions. To speed up optimal user association searching, we incorporate a penalty function into the objective function to resolve uncertainties of a binary nature. Two sub-optimal designs for given user association policies based on channel conditions and random user associations are also investigated to serve as state-of-the-art benchmarks. Numerical results are provided to demonstrate the effectiveness of the proposed algorithms in terms of the e2e latency and convergence speed.

A multi-aerial base station assisted joint computation offloading algorithm based on D3QN in edge VANETs

With the rapid development of Internet of Vehicle technology, vehicular edge computing is gradually becoming a key technology for task offloading and resource scheduling. However, the application of traditional base station and vehicle edge network processing introduces large latency, energy consumption and QoS degradation issues for sensitive intensive tasks. To address these issues, we propose a Multi-Aerial Base Station Assisted Joint Computation Offload algorithm based on D3QN in Edge VANETs (MAJVD3). The use of SDN Controllers senses global information and enables efficient resource scheduling. Firstly, a non-convex optimization problem is formulated and proved as an NP-hard problem by combining optimization variables, metrics and environmental constraints. Secondly, we propose to use ϵ-greedy MAJVD3 algorithm for interactive exploration with a dynamic environment to find the optimal solution to non-convex optimization problem. Finally, the network performance, applicability and neural network learning are evaluated separately, and the simulation results shows the effectiveness and convergence of the proposed algorithm. Compared with the existing baseline algorithms, the proposed MAJVD3 algorithm improves the network utility by 4.8%, 100% and 149.6%, respectively. The task latency is reduced by 3%, 30.4% and 30.7%, respectively, and the energy consumption is reduced by 4.8%, 42.3% and 45.7%, respectively.

Energy-Efficient Task Offloading of Edge-Aided Maritime UAV Systems

This paper considers the autonomous detecting and tracking task of unmanned aerial vehicles (UAVs) in the maritime environment. Due to the high computational complexity of the image-processing task, the strict latency requirement of the tracking task, and the shortage of batteries and computational capability of UAVs, the UAVs can offload the computing-intensive task to the edge computing server (ECS) to reduce the task latency and energy consumption. Even though, the large latency and energy consumption brought by some high-definition image-processing task still make some visual tracking difficult to perform. We first investigate the impact of video frame resolution on computing task size and detection accuracy, and further on task latency and energy consumption. In addition, we optimize the transmission power, the local CPU frequency, the offloading rate of UAVs, and the bandwidth allocation between UAVs. We model the UAVs tracking system as an energy consumption optimization problem with the constraint of task latency. Although the proposed optimization problem is complex, we transform the optimization problem into several subproblems with decoupling and rigorous mathematical proof. By combining the convex optimization with the genetic algorithm, an offloading algorithm with low complexity and good performance is obtained, which is verified by comparing with some benchmark offloading strategies in the numerical results.

Deep reinforcement learning‐based joint optimization of computation offloading and resource allocation in F‐RAN

AbstractThe fog radio access network (F‐RAN) has been regarded as a promising wireless access network architecture in the fifth generation (5G) and beyond systems to satisfy the increasing requirements for low‐latency and high‐throughput services by providing fog computing. However, because the cloud computing centre and fog computing‐enabled access points (F‐APs) in the F‐RAN have different computation and communication capabilities, it is crucial to make an efficient computation offloading and resource allocation strategy that can fully exploit the potential of the F‐RAN system. In this paper, the authors investigate a decentralized low‐complexity deep reinforcement learning (DRL)‐based framework for joint computation task offloading and resource allocation in the F‐RAN, which supports assistive computing‐enabled tasks offloading between F‐APs. Considering the constraints of task latency, wireless transmission rate, transmission power, and computational resource capacity, the authors formulate the system processing efficiency maximization problem by jointly optimizing offloading mode selection, channel allocation, power control, and computation resource allocation in the F‐RAN. To solve this non‐linear and non‐convex problem, the authors propose a federated DRL‐based computation offloading and resource allocation algorithm to improve the task processing efficiency and ensure privacy in the system, which can significantly reduce the computing complexity and signalling overhead of the training process compared with the centralized learning‐based method. Specifically, each local F‐AP agent consists of dueling deep Q‐network (DDQN) and deep deterministic policy gradient (DDPG) networks to appropriately deal with discrete and continuous valuable action spaces, respectively. Finally, the simulation results show that the proposed federated DRL algorithm can achieve significant performance improvements in terms of system processing efficiency and task latency compared with other benchmarks.

Adaptive resource optimization for edge inference with goal-oriented communications

Goal-oriented communications represent an emerging paradigm for efficient and reliable learning at the wireless edge, where only the information relevant for the specific learning task is transmitted to perform inference and/or training. The aim of this paper is to introduce a novel system design and algorithmic framework to enable goal-oriented communications. Specifically, inspired by the information bottleneck principle and targeting an image classification task, we dynamically change the size of the data to be transmitted by exploiting banks of convolutional encoders at the device in order to extract meaningful and parsimonious data features in a totally adaptive and goal-oriented fashion. Exploiting knowledge of the system conditions, such as the channel state and the computation load, such features are dynamically transmitted to an edge server that takes the final decision, based on a proper convolutional classifier. Hinging on Lyapunov stochastic optimization, we devise a novel algorithmic framework that dynamically and jointly optimizes communication, computation, and the convolutional encoder classifier, in order to strike a desired trade-off between energy, latency, and accuracy of the edge learning task. Several simulation results illustrate the effectiveness of the proposed strategy for edge learning with goal-oriented communications.

An Energy-Efficient Method for Recurrent Neural Network Inference in Edge Cloud Computing

Recurrent neural networks (RNNs) are widely used to process sequence-related tasks such as natural language processing. Edge cloud computing systems are in an asymmetric structure, where task managers allocate tasks to the asymmetric edge and cloud computing systems based on computation requirements. In such a computing system, cloud servers have no energy limitations, since they have unlimited energy resources. Edge computing systems, however, are resource-constrained, and the energy consumption is thus expensive, which requires an energy-efficient method for RNN job processing. In this paper, we propose a low-overhead, energy-aware runtime manager to process tasks in edge cloud computing. The RNN task latency is defined as the quality of service (QoS) requirement. Based on the QoS requirements, the runtime manager dynamically assigns RNN inference tasks to edge and cloud computing systems and performs energy optimization on edge systems using dynamic voltage and frequency scaling (DVFS) techniques. Experimental results on a real edge cloud system indicate that in edge systems, our method can reduce the energy up to 45% compared with the state-of-the-art approach.

Multisignal Cooperative Processing Method for Internet of Vehicles Based on Data-Driven Edge Computing Method

With the rapid development of Internet of Vehicles applications, more and more data are generated. How to effectively distribute content in the Internet of Vehicles to meet the service quality requirements of users has become one of the industry pain points in the field of smart cars and autonomous driving. In order to solve the shortage of local computing resources of vehicles, a vehicle edge network is proposed, which uses data-driven edge computing to offload vehicle tasks to a mobile edge computing server to reduce overall network energy consumption and meet task latency requirements. In addition, in order to reduce the end-to-end delay, caching technology is adopted at the network edge, which can reduce the content transmission delay. This paper focuses on the problem of computing offloading in the data-driven edge computing method in the context of the Internet of Vehicles. Simulation experiments have proved its superiority compared with the traditional offloading method, and the delay and energy consumption are better than the traditional method. First, the basic concepts of the Internet of Vehicles and MEC are introduced; second, the TOAI algorithm flow chart and the computing tasks of the offloading work of the Internet of Vehicles are introduced; then, based on MEC and partial offloading, the task offloading problem is modeled and solved; the problem of unloading collaborative content under networking is solved, and the simulation results are analyzed and verified. The simulation experiment not only shows that the proposed algorithm optimizes the efficiency of the task under the average unit but also shows the effectiveness of this method, which lays a foundation for the engineering implementation of the algorithm. The experimental results show that the average task completion rate is increased by 0.58%, and the average unit task energy consumption is increased by 0.32%, which improves the practicability of the system.

Impaired Aversive Memory Formation in GPR37L1KO Mice.

GPR37L1 is an orphan G-protein-coupled receptor, which is implicated in neurological disorders, but its normal physiological role is poorly understood. Its close homologue, GPR37, is implicated in Parkinson's disease and affective disorders. In this study, we set out to characterize adult and middle-aged global GPR37L1 knock-out (KO) mice regarding emotional behaviors. Our results showed that GPR37L1KO animals, except adult GPR37L1KO males, exhibited impaired retention of aversive memory formation as assessed by the shorter retention latency in a passive avoidance task. Interestingly, the viral-mediated deletion of GPR37L1 in conditional knockout mice in the hippocampus of middle-aged mice also showed impaired retention in passive avoidance tasks, similar to what was observed in global GPR37L1KO mice, suggesting that hippocampal GPR37L1 is involved in aversive learning processes. We also observed that middle-aged GPR37L1KO male and female mice exhibited a higher body weight than their wild-type counterparts. Adult and middle-aged GPR37L1KO female mice exhibited a reduced level of serum corticosterone and middle-aged GPR37L1KO females showed a reduced level of epinephrine in the dorsal hippocampus in the aftermath of passive avoidance task, with no such effects observed in GPR37L1KO male mice, suggesting that lack of GPR37L1 influences behavior and biochemical readouts in age- and sex-specific manners.

ATTENUATION OF OXIDATIVE STRESS AND NEUROTOXICITY BY MK-801 (DIZOCILPINE) ON DIPENTYLPHTHALATE-INDUCED COGNITIVE DYSFUNCTION IN MICE

Objectives: The aim of our research is to study the effect of dipentylphthalate (DPeP), a plasticizer on cognition and various oxidative stress markers in mice, and to explore the modulatory effects of MK-801. Methods: In the present study, experimental mice were orally treated with two doses (33 and 100 mg/kg) of DPeP for 28 days. Cognitive functions were assessed using spatial navigation task on Morris water maze (MWM) and step-down latency (SDL) in passive avoidance apparatus. Oxidative stress was assessed by examining the levels of malondialdehyde (MDA), glutathione (GSH), ferric-reducing antioxidant power (FRAP), and 8-hydroxy-deoxyguanosine (8-OH-dG) levels in whole brain of mice. Results: DPeP exposure led to a statistically significant increase of latency in spatial navigation task and significant decline in the SDL in passive avoidance apparatus when compared to the control groups. Oxidative stress markers showed a significant increase following DPeP administration as seen with rise in levels of MDA, 8-OH-dG, and a fall in GSH and FRAP levels. Conclusion: The present data suggest that DPeP could adversely affect learning and memory functions in mice by an oxidative stress-mediated neuronal damage and pre-administration of MK-801 has the potential to attenuate these effects.

Toward Mobility-Aware Computation Offloading and Resource Allocation in End–Edge–Cloud Orchestrated Computing

Mobile devices (MDs) have undergone a booming development, yet are still capacity limited in computation and energy resources and thus could face troubles when serving computation-intensive and delay-sensitive applications. Mobile-edge computing (MEC) has been proposed to accommodate MDs with both satisfactory latency and acceptable resources, by offloading MDs’ tasks to near-deployed edge servers (ESs). Whereas, offloading tasks solely to ESs are difficult to meet distinct requirements of various applications, which leads to the emergence of end–edge–cloud orchestrated computing (EECOC). However, studies on EECOC are still insufficient, most existing of which do not consider MDs’ dynamical movements partly due to the intractability of associating optimal ESs with moving MDs. To address the issue, a novel deep reinforcement learning (DRL)-based mobility-aware (MA) EECOC scheduling approach is proposed in this article. With the goal of minimizing maximal task latency, we first formulate and transform the optimization problem into a Markov decision problem (MDP). Then, we enable DRL with elaborately designed reward functions and integrate it with NoisyNet to obtain near-optimal solutions. Furthermore, a MA component based on ConvLSTM is developed to extract MDs’ temporal–spatial distribution features and predict their movements, which are further utilized to facilitate the decision making of computation-offloading and resource-allocation actions. Extensive experimental results indicate the promising performance improvements of our approach against the state-of-the-art approaches in various scenarios.

On-demand inference acceleration for directed acyclic graph neural networks over edge-cloud collaboration

As the development of deep neural network (DNN) tasks that feature a directed acyclic graph (DAG) surges in modern industries, simultaneously meeting the latency and accuracy requirements of these key DNN tasks has remained elusive. The general feasible solution is to find an optimal DNN partition that combines cloud with edge computing to implement edge–cloud collaboration for DNN inference. Nevertheless, the dynamicity of network conditions and the uncertain availability of cloud computing resources pose formidable obstacles to this collaborative inference. In this study, we formulate the optimization of DNN partition as a minimum cut problem in DAG derived from DNN and then propose an early-exit DAG-DNN inference (EDDI) framework that supports synergistically on-demand inference acceleration for DAG DNN. This framework introduces two novel components: (1) Evaluator that derives an approximately optimal solution for constructing DNN into appropriate DAG, assisting in solving the optimization problem; and (2) Optimizer that enables collaborative optimization of the early exit and DNN partition strategy at run time to improve performance while meeting user-defined latency requirements. Quantitative evaluations show that EDDI outperforms state-of-the-art schemes by 10.6%, 3.3%, and 3%, on average, in terms of model accuracy, inference latency, and throughput under diverse latency constraints, respectively. Meanwhile, latency speedup ratio increases by an average of 8% and 4% under varying network conditions and cloud server capacities, respectively.

Physically active undergraduates perform better on executive-related oculomotor control: Evidence from the antisaccade task and pupillometry.

Previous studies have shown that exercise can improve executive function in young and older adults. However, it remains controversial whether a sufficient amount of physical activity leads to higher-level executive function. To examine the effect of physical activity on executive function, we used eye-tracking technology and the antisaccade task in 41 young undergraduates with various levels of physical activity. Moreover, we also investigated their differences in cognitive ability by examining their pupil size during the antisaccade task. Eye-tracking results showed that physically active individuals showed shorter saccade latency and higher accuracy in the antisaccade task than their physically inactive counterparts. Furthermore, the former showed larger pupil size during the preparatory period of antisaccade. These findings suggest that individuals with higher-level physical activity have higher-level executive function. The larger pupil sizes of physically active individuals may imply that their locus coeruleus-norepinephrine system and executive-related prefrontal cortex are more active, which contributes to their higher-level cognitive ability.

Efficient Matching-Based Parallel Task Offloading in IoT Networks.

Fog computing is one of the major components of future 6G networks. It can provide fast computing of different application-related tasks and improve system reliability due to better decision-making. Parallel offloading, in which a task is split into several sub-tasks and transmitted to different fog nodes for parallel computation, is a promising concept in task offloading. Parallel offloading suffers from challenges such as sub-task splitting and mapping of sub-tasks to the fog nodes. In this paper, we propose a novel many-to-one matching-based algorithm for the allocation of sub-tasks to fog nodes. We develop preference profiles for IoT nodes and fog nodes to reduce the task computation delay. We also propose a technique to address the externalities problem in the matching algorithm that is caused by the dynamic preference profiles. Furthermore, a detailed evaluation of the proposed technique is presented to show the benefits of each feature of the algorithm. Simulation results show that the proposed matching-based offloading technique outperforms other available techniques from the literature and improves task latency by 52% at high task loads.

Computational access point selection based on resource allocation optimization to reduce the edge computing latency

Mobile Devices (MD) and the Internet of Things (IoT) transfer compute tasks to the client through Computational Access Point (CAP) through the extensive implementation of Local Area Networks (LAN). If all IoT or mobile devices choose the same ports of entry to offload their work, then the calculation of offloading leads to an increase in operating costs. In a network of different CAP to Edge Computing (EC), this research proposes a resource allocation optimization technique and a computation offloading strategy that aims to reduce network costs by recommending the best transmit power distribution, bandwidth assignment, and computation task planning. The optimization problem of NP-hard solved by the proposed technique separates from sub-problems of resource allocation and offloading technique. An algorithm called Dynamic Weighed Quantum Arithmetic Optimization Algorithm (DWQAOA) was proposed for this offloading problem NP-hard. To achieve the ideal solution, a transition probability introduced the optimal Pareto connection during the optimization process. Numerous simulation results show that the edge task latency could be reduced from 3 to 25% relative to competing solutions.

MHDNNL

Task optimization scheduling is one of the key concerns of both cloud service providers (CSPs) and cloud users. The CSPs hope to reduce the energy consumption of executing tasks to save costs, while the users are more concerned about shorter task completion time. In cloud computing, multi-queue and multi-cluster (MQMC) is a common resource configuration mode, and batch is a common task commission mode. The task scheduling (TS) in these modes is a multi-objective optimization (MOO) problem, and it is difficult to get the optimal solution. Therefore, the authors proposed a MOO scheduling algorithm for this model based on multiple heterogeneous deep neural networks learning (MHDNNL). The proposed algorithm adopts a collaborative exploration mechanism to generate the samples and use the memory replay mechanism to train. Experimental results show that the proposed algorithm outperforms the benchmark algorithms in minimizing energy consumption and task latency.

The Autism Biomarkers Consortium for Clinical Trials: Initial Evaluation of a Battery of Candidate EEG Biomarkers.

Numerous candidate EEG biomarkers have been put forward for use in clinical research on autism spectrum disorder (ASD), but biomarker development has been hindered by limited attention to the psychometric properties of derived variables, inconsistent results across small studies, and variable methodology. The authors evaluated the basic psychometric properties of a battery of EEG assays for their potential suitability as biomarkers in clinical trials. This was a large, multisite, naturalistic study in 6- to 11-year-old children who either had an ASD diagnosis (N=280) or were typically developing (N=119). The authors evaluated an EEG battery composed of well-studied assays of resting-state activity, face perception (faces task), biological motion perception, and visual evoked potentials (VEPs). Biomarker psychometrics were evaluated in terms of acquisition rates, construct performance, and 6-week stability. Preliminary evaluation of use was explored through group discrimination and phenotypic correlations. Three assays (resting state, faces task, and VEP) show promise in terms of acquisition rates and construct performance. Six-week stability values in the ASD group were moderate (intraclass correlations ≥0.66) for the faces task latency of the P1 and N170, the VEP amplitude of N1 and P1, and resting alpha power. Group discrimination and phenotype correlations were primarily observed for the faces task P1 and N170. In the context of a large-scale, rigorous evaluation of candidate EEG biomarkers for use in ASD clinical trials, neural response to faces emerged as a promising biomarker for continued evaluation. Resting-state activity and VEP yielded mixed results. The study's biological motion perception assay failed to display construct performance. The results provide information about EEG biomarker performance that is relevant for the next stage of biomarker development efforts focused on context of use.