Amidst the diverse computational and communication capabilities of the resource-constrained edge devices (EDs), synchronous model aggregation in wireless federated learning (FL) often suffers from inefficiencies due to stragglers, which diminishes learning efficiency. Conversely, asynchronous aggregation contends with delayed model updates, leading to staleness that adversely affects convergence and overall learning performance. To address these challenges, we propose the Genetic Algorithm (GA) Tuned Semi-Asynchronous FL (GATune-SAFL) strategy, which leverages the strengths of both synchronous and asynchronous FL while mitigating their inherent drawbacks. Specifically, a GA evolves a population of staleness parameter sets in each training round to achieve an optimal balance between learning latency and accuracy. Recognizing that data heterogeneity among edge devices may introduce biases in global model updates, we also propose Dynamic Learning Rate Adjustment to further enhance model performance. Theoretically, we analyze the impact of staleness in different rounds on the convergence bound of GATune-SAFL, showing that GA-based staleness tuning significantly improves learning performance by accelerating convergence and enhancing accuracy. Experimental results on the MNIST, FMNIST, and CIFAR-10 datasets demonstrate that the proposed strategy outperforms benchmark FL algorithms in both convergence speed and learning accuracy.

Nanorobots are microscopic robots that operate at the molecular and cellular level and can potentially revolutionize fields such as medicine, manufacturing, and environmental monitoring due to their precision. However, the challenge for researchers is to analyze the data and provide a constructive recommendation framework instantly, as most nanorobots demand on-time and near-edge processing. To tackle this challenge, this research presents a novel edge-enabled intelligent data analytics framework called Transfer Learning Population Neural Network (TLPNN) to predict glucose levels and associated symptoms from invasive and non-invasive wearable devices. The TLPNN is designed to be unbiased in predicting symptoms during the initial phase but later modified based on the best-performing neural networks during the learning phase. The effectiveness of the proposed method is validated using two publicly available glucose datasets with various performance metrics. The simulation results demonstrate the effectiveness of the proposed TLPNN method over existing ones.

There is significant interest in edge computing (EC) for computational social systems to process and store data at the edge of the network. One of the key applications of EC is to analyze the large-scale social data, received from multiple sources using machine learning/deep learning (ML/DL) models with minimum delay and higher accuracy. However, traditional models are often large and require significant computational resources, posing a challenge in resource-constrained edge networks. Besides that, traditional centralized ML/DL methods including collaborative learning have limitations such as data privacy and communication overhead. Federated learning (FL) is an alternative solution to overcome some of these limitations by allowing model training across multiple decentralized devices without sharing the actual data. However, the standard FL approaches face some challenges, including extended training times due to the heterogeneous devices and the risk of single-point failure. To address these challenges, in this article, we propose a novel Gossip-assisted DL model for resource-constraint edge devices problem by enabling decentralized and serverless training while mitigating the risk of single-point failure. Besides that, we develop a lightweight model extractor for local edge devices to train a DL model with the collaboration of neighboring devices that improves knowledge discovery with higher prediction accuracy. Extensive simulation results over two publicly available large-scale datasets demonstrate the effectiveness of the proposed approach over the state-of-the-art techniques.

Federated Learning (FL) has emerged as a prominent solution for distributed learning environments, enabling collaborative model training without centralized data collection. However, FL faces significant challenges such as data heterogeneity and resource-constraint edge devices for model training and analysis, leading to accuracy degradation and bias in model performance. To address these critical issues, we propose a novel FL strategy named FedTune-SGM, designed to optimize model training in decentralized settings. In this strategy, a cloud-based model is initially trained and fine-tuned on the edge devices with additional layers tailored to the specific data characteristics. This fine-tuning process effectively mitigates the impact of data heterogeneity, enhancing the robustness and generalization capability of the model. FedTune-SGM employs a strategic weighting mechanism that ensures a balanced and equitable contribution from participating edge devices to prevent dominant influences from resource-rich devices and promote a fairer and more accurate aggregated model. Additionally, the proposed strategy integrates a Stackelberg Game model to foster an interactive and dynamic cloud-edge setup that motivates edge devices to invest more effort in model training and ensures the effectiveness of resource-constraint edge devices. Extensive experiments conducted on three diverse datasets highlight the superior performance of the proposed FedTune-SGM strategy compared to state-of-the-art FL techniques in terms of accuracy and robustness while meeting the critical challenges of data heterogeneity and resource limitations in FL environments. Through these innovations, FedTune-SGM paves the way for more reliable and efficient distributed learning systems, unlocking the full potential of FL in practical applications.

Different geospatial and geotechnical parameters change over time and can affect the residual soil stability on a slope. Thus, it is essential to analyze the stability of slopes continuously to identify the potential landslide sections. The stability of slopes is defined by the factor of safety (FOS). To track the immediate changes in soil stability, it is essential to monitor multiple environmental parameters using Internet of Things (IoT) devices for real-time decision making. Further, the relation between the environmental parameters and FOS is nonlinear which makes it a multivariate and complex problem. To motivate the above-mentioned challenges, in this article, we propose a new fusion-based bag-of-neural network (FuBoNN) model for predicting FOS using a set of IoT devices in edge networks. Besides that, for increasing the prediction accuracy of FOS, multiple laboratory data related to FOS are fused over the monitoring parameters and prepare a rich data set. The newly created fused data set is fed into the population-based neural network (NN) and the best NN is selected in each iteration that transfers its knowledge to the population. The fused data set is categorized into four class labels to simulate the stability issue of residual soil and fed to the input of the proposed FuBoNN model, which provides a 0.0003% of error in predicting the multiple categories of the FOS. The proposed work is compared to the standard machine learning models that demonstrate the efficiency of the proposed model and produce 2.5% improved prediction accuracy over the existing ones.

Nowadays, by integrating the Industrial Internet of Things (IIoT) with fog networks, companies can efficiently manage the increasing data traffic and enhance the capabilities of sensing devices. However, control of critical IIoT applications has become difficult because of the increasing demand for technology during the Industry 4.0 revolution and the use of fog computing. To address this issue, we introduce an efficient resource provisioning strategy called Fair Scheduling and Computation Co-offloading (FSCC) for executing the maximum number of tasks within the corresponding deadline while achieving network stability. Initially, we formulate the task scheduling problem as a stochastic problem and devise a novel optimization framework by exploiting the Lyapunov optimization technique. A two-phase task offloading strategy is also proposed to efficiently offload scheduled tasks to suitable computing devices in fog networks. The proposedFSCCstrategy combines devices’ current state information and a collaborative fog-cloud infrastructure for controlling network parameters and utilizing available fog resources. Experimental results demonstrate that the proposed strategy improves 15-20% end-to-end delay and deadline satisfaction over existing methods.

Malaria is a serious and fatal disease spread by female Anopheles. In past decades, the Deep Learning models played the most powerful roles for malaria parasite detection. However, concerning the privacy of patient medical records, prevents data sharing in medical institutions, resulting in the unexpected performance of Deep Neural Network models. Thankfully, Federated Learning (FL) provides the opportunity to train models locally and build the global model based on the updates received from local client models without sharing data. The traditional Federated Averaging algorithm assumes equal contributions from all participants in building the global model. Therefore, it limits the different applications by ignoring the potential domain of several different FL participants. The challenges mentioned above motivate us to propose a framework of FL using customized Weighted Averaging for Malaria Parasite Detection using Cell Microscopic (CMR) Images, collected from different medical institutions at the edge using a Convolutional Neural Network, named FedWtAvg. The proposed FedWtAvg provides the advantage of giving higher priority to participants who have large amounts of data and can build an effective global model. Extensive experiments were conducted on the CMR Images dataset to demonstrate that FedWtAvg outperforms and improves the accuracy by around 1.21 % over existing ones without data sharing and real-time continuous learning.

With the evolutionary development of the latency-sensitive industrial Internet-of-Things (IIoT) applications, delay restriction becomes a critical challenge, which can be resolved by distributing IIoT applications on nearby fog devices. Besides that, efficient service provisioning and energy optimization are confronting serious challenges with the ongoing expansion of large-scale IIoT applications. However, due to insufficient resource availability, a single fog device cannot execute large-scale applications completely. In such a scenario, a partial service provisioning strategy provides a promising outcome to enable the services on multiple fog devices or collaboration with cloud servers. By motivating this scenario, in this article, we introduce a new deep reinforcement learning (DRL)-enabled partial service provisioning strategy in the green industrial fog networks. With this strategy, multiple fog devices share the excessive workload of an application among themselves. To reflect this, a task partitioning policy is introduced to partition the requested applications into a set of independent or interdependent tasks. Furthermore, we develop an intelligent partial service provisioning strategy to utilize maximum fog resources in the network. The experimental results express the significance of the proposed strategy over the traditional baseline algorithms in terms of energy consumption and latency up to 25% and 16%, respectively.

Fault diagnostics involving the Internet-of-Things (IoT) sensors and edge devices is a challenging task due to their limited energy and computational capabilities. Another challenge concerning IoT sensors or devices is the incursion of noise when used in an industrial environment. The noisy samples affect the decision support system that could lead to financial and operational losses. This article proposes a noisy encoder using artificial intelligence of things (NEAT) architecture for fault diagnosis in IoT edge devices. NEAT combines autoencoders and Inception module to co-train the clean and noisy samples for solving the said problem. Experimental results on benchmark data sets reveal that the NEAT architecture is noise resilient in comparison to the existing works. Furthermore, we also show that the NEAT architecture has lightweight characteristics as it yields a lower number of parameters, weight storage, training, and testing times that support its real-life applicability in an Industrial IoT environment.

The seamless integration of sensors and smart communication technologies has led to the development of various supporting systems for financial technology (FinTech). The emergence of the next-generation Internet of Things (Nx-IoT) for FinTech applications enhances the customer satisfaction ratio. The main research challenge for FinTech applications is to analyze the incoming tasks at the edge of the networks with minimum delay and power consumption while increasing the prediction accuracy. Motivated by the above-mentioned challenge, in this article, we develop a ranked-based service deployment strategy and an artificial intelligence technique for financial data analysis at edge networks. Initially, a risk-based task classification strategy has been developed for classifying the incoming financial tasks and providing the importance to the risk-based task for meeting users’ satisfaction ratio. Besides that, an efficient service deployment strategy is developed using$Hall's$theorem to assign the ranked-based financial data to the suitable edge or cloud servers with minimum delay and power consumption. Finally, the standard support vector machines (SVMs) algorithm is used at edge networks for analyzing the financial data with higher accuracy. The experimental results demonstrate the effectiveness of the proposed strategy and SVM model at edge networks over the baseline algorithms and classification models, respectively.

Considering the increasing number of communicable disease cases such as COVID-19 worldwide, the early detection of the disease can prevent and limit the outbreak. Besides that, the PCR test kits are not available in most parts of the world, and there is genuine concern about their performance and reliability. To overcome this, in this paper, we develop a novel edge-centric healthcare framework integrating with wearable sensors and advanced machine learning (ML) model for timely decisions with minimum delay. Through wearable sensors, a set of features have been collected that are further preprocessed for preparing a useful dataset. However, due to limited resource capacity, analyzing the features in resource-constrained edge devices is challenging. Motivated by this, we introduce an advanced ML technique for data analysis at edge networks, namely Deep Transfer Learning (DTL). DTL transfers the knowledge from the well-trained model to a new lightweight ML model that can support the resource-constraint nature of distributed edge devices. We consider a benchmark COVID-19 dataset for validation purposes, consisting of 11 features and 2 Million sensor data. The extensive simulation results demonstrate the efficiency of the proposed DTL technique over the existing ones and achieve 99.8% accuracy while diseases prediction.

The papers in this special issue focus on advanced wearable sensor technologies for monitoring and disease prediction. The seamless integration of sensor technologies with the smart healthcare infrastructure has leveraged the sensing and communication capabilities to monitor patient’s health parameters remotely through various wearable/medical sensors. Advanced sensor technologies enable various types of smart healthcare applications, including diagnosing the symptomatic/ asymptomatic patients’ health condition, health symptoms forecasting, disease prediction and analysis, and ontologybased recommendation. Advancements in wearable sensors and communication technologies (6G/5G and-beyond) enable the design of smart healthcare frameworks and efficiently analyzing the sensing parameters. Besides that, advanced AI-enabled technologies, including machine learning and deep learning algorithms, come into play to analyze the sensed data at remote computing devices for disease prediction and diagnosis. This special issue focus on discussions and insights into the latest advancements and technologies pertaining to these technologies.

Recent advancements in computation and communication technologies and the increasing adoption of the Internet of Things (IoT) and Artificial Intelligence (AI) technologies have paved the way to tremendous developments in modern transportation systems. Driven by the massive number of connected vehicles and the stringent requirements of the public traffic management system, the transportation of data to and from the centralized cloud servers poses a great challenge. As a result, to meet the computational requirements and handle the massive amount of sensory data efficiently, the potential solution is to process/analyze the data at the edge of the network. Motivated by the challenges mentioned above, in this paper, we design a new empirically intelligent XGboost (EIXGB)-enabled logistic transportation system at the edge network for analyzing the data efficiently. Besides that, the proposed EIXGB technique intends to obtain real-time results based on the monitoring parameters of the public traffic management system with higher accuracy and minimum error. Extensive simulation results demonstrate the efficiency of the proposed EIXGB technique over the standard machine learning techniques using a set of parameters. The proposed technique achieves 87-97% accuracy over the different sets of features of a real-time dataset as per the simulation results.

With the exponential growth of the Internet of Things (IoT) devices in Maritime Transportation Systems (MTS), the centralized cloud-centric framework can hardly meet the requirements of the applications in terms of low latency and power consumption. By inventing the distributed edge-centric framework, real-time IoT applications can meet the requirements of the MTS by analyzing the tasks at the edge of the networks. However, one of the critical challenges of the edge-centric MTS is to provide security and privacy between local IoT devices and distributed edge nodes. Motivated by that, in this paper, we design a blockchain-enabled edge-centric framework for analyzing the real-time data at the edge of the networks with minimum latency and power consumption while meeting the security and privacy issue of MTS. The introduction of blockchain and smart contract in the edge-centric MTS frameworks help to validate the transactions of each block at edge nodes by estimating the lifetime, belief, and trustfulness, and mitigate various types of security threats. Further, we introduce different classification models to predict the malicious vessels over the real-time maritime dataset at a secured edge-centric MTS framework. Extensive simulation results demonstrate that the superiority of the proposed strategy with baseline approaches under various performance metrics.

In the era of the Internet of Things (IoT), fog computing has become an enticing concept for supporting delay-sensitive tasks by offering versatile and convenient computing and communication services to the end users, in conjunction with cloud services. Most of the existing research mainly draws attention to the communication delay minimization and completion time reduction in the hierarchical fog networks without giving the priority to select the suitable computing device during failure or resource unavailability of the current computing devices. By motivating the above-mentioned challenges, in this article, we propose a deadline-aware dynamic task placement (DDTP) strategy to offload and place the tasks to a suitable computing device in fog networks. In this context, we design a new federated fog framework consisting of several fog clusters in which the cluster head, termed as master fog node, acts as a fog controller that controls and manages the data distribution among the other fog nodes, termed as slave fog nodes. The proposed DDTP strategy selects the suitable computing device for each incoming task as per the deadline and ensures to meet the deadline constraints of the tasks using a dynamic task allocation policy. Finally, a dispatch-constrained offloading policy is developed to reassign the failed tasks to the available fog nodes in the network. Comprehensive simulation results depict the efficiency of the proposed strategy over the existing baseline algorithms in terms of various performance matrices.

Network data traffic has expanded exponentially over the past decade, resulting in massive congestion in heterogeneous networks. Nevertheless, it is nearly impossible to run latency-intensive applications to the end-users local processing unit due to limited system resources. Recently, fog computing has come up with a solution to reduce such data congestion by offloading some or the whole part of the task to the nearby fog nodes (FNs) or the clouds. But this offloading policy becomes more complex when the FN is unable to process the task and further offload it to another neighboring FN or the cloud. In this context, in this study, we have analyzed the offloading strategy in a hierarchical fog-cloud network consisting of several heterogeneous fog devices along with a helping fog and a centralized cloud server. We have considered the most possible practical situation where the FNs are equipped with different CPU frequencies and hence, the power consumption is also different. The total system cost is formulated as a mixed-integer nonlinear problem that aims to reduce the overall delay in the proposed network. To solve the NP-hard problem, we transform it into quadratically constrained quadratic programming (QCQP) formation and further solve it by the separable semidefinite relaxation (SDR) method. Finally, by adopting several benchmark data, we conduct comprehensive simulations to test the efficiency of the proposed offloading profile. The simulation results depict that the proposed strategy outperforms in many aspects when compared to various baseline algorithms.

Fog computing has become a promising technology to improve the performance of low-powered Industrial Internet of Things (IIoT) devices by providing flexible and convenient computing services at the edge of the network with minimum delay. However, due to the ever-increasing traffic load in the network, the traditional service provisioning strategies can pose high complexity as well as network congestion, resulting in high energy consumption. Owing to this issue, in this article, we propose a deep reinforcement learning (DRL)-based service provisioning strategy in a software-defined industrial fog network to minimize the energy consumption of the network. The service provisioning strategy is performed in the network data plane, whereas the DRL is deployed in the control plane to enhance network efficiency. The service provisioning problem is formulated as the Markov decision process (MDP) and further solved by adopting the concept of a deep$Q$network (DQN). Further, we propose a task migration policy to ensure the high availability of computing devices while meeting a single point of failure (SPOF) issue. Finally, to show the effectiveness of the proposed method, it is compared with the traditional baseline algorithms over various performance metrics.

Abstract Driven by the vision of real‐time applications and smart communication, recent years have witnessed a paradigm shift from centralized cloud computing toward distributed edge computing. The main features of edge computing are to drag the cloud services toward the network edge with dramatic reductions of latency while increasing the resource utilization of the network and computing devices. Being the natural extension of cloud computing, edge computing inherits a variety of research challenges and brings forth different new issues to solve. These challenges are dealing with solving complex optimization problems including scheduling and processing real‐time applications. Nature‐inspired meta‐heuristic (NIMH) algorithm is an overarching term in the field of an optimization problem that provides robust solutions to the NP‐complete problems, from computationally tractable approximate solutions to real‐time optimization strategies. Nowadays, different NIMH algorithms have been applied in the field of edge computing for solving various research challenges including resource placement and scheduling, communication, mobility, and edge controlling with higher efficiency. In this survey, we classify the existing NIMH into three categories based on their nature of works and included fuzzy logic and systems in the field of edge networks along with different research challenges. Further, we introduce different challenges and future directions to identify promising research works in edge computing.

The papers in this special section focus on recent advances in fuzzy-based intelligent Internet of Things and cyber-physical systems. The rapid development of real-time Internet of Things (IoT) applications including smart grids, smart city, and intelligent transport networks generate a tremendous amount of data from massively distributed sources, which require high computing and communication demand that frequently exceeds the users’ requirements. Furthermore, many emerging IoT applications including remote surgery, machine monitoring and control, fault detection, and healthcare generate delay-sensitive tasks, which require timely processing with minimum delay. Besides that, cyber-physical systems (CPS) integrate computing and communication capabilities with monitoring and control of entities in the physical world. These systems are usually composed of a set of networked agents, including sensors, actuators, control processing units, and communication devices. All the critical infrastructures are also a part of the cyber-physical ecosystem to enable smart and connected environments.

Cybertwin leverages the capabilities of networks and serves in multiple functionalities, by identifying digital records of activities of humans and things, from the Internet of Everything (IoE) applications. Cybertwin emerges as a promising solution along with next-generation communication networks, i.e., 6G technology; however, it increases additional challenges at the edge networks. Motivated by the aforementioned perspectives, in this article, we introduce a new cybertwin-driven edge framework using 6G-enabled technology with an intelligent service provisioning strategy for supporting a massive scale of IoE applications. The proposed strategy distributes the incoming tasks from IoE applications using the deep reinforcement learning technique based on their dynamic service requirements. Besides that, an artificial-intelligence-driven technique, i.e., the support vector machine (SVM) classifier model, is applied at the edge network to analyze the data and achieve high accuracy. The simulation results over the real-time financial datasets demonstrate the effectiveness of the proposed service provisioning strategy and the SVM model over the baseline algorithms in terms of various performance metrics. The proposed strategy reduces the energy consumption by 15% over the baseline algorithms, while increasing the prediction accuracy by 12% over the classification models.

The seamless integration of medical sensors and the Internet of Things (IoT) in smart healthcare has leveraged an intelligent Internet of Medical Things (IoMT) framework to detect the criticality of the patients. However, due to the limited storage capacity and computation power of the local IoT devices, patient's health data needs to transfer to remote computing devices for analysis, which can easily result in privacy leakage due to lack of control over the patient's health data and the vulnerability of the network for various types of attacks. Motivated by this, in this paper, an Empirical Intelligent Agent (EIA) based on a unique Swarm-Neural Network (Swarm-NN) method is proposed to identify attackers in the edge-centric IoMT framework. The major outcome of the proposed strategy is to identify the attacks during data transmission through a network and analyze the health data efficiently at the edge of the network with higher accuracy. The proposed Swarm-NN strategy is evaluated with a real-time secured dataset, namely the ToN-IoT dataset that collected Telemetry, Operating systems, and Network data for IoT applications and compares the performance over the standard classification models using various performance metrics. The test results demonstrate that the proposed Swarm-NN strategy achieves 99.5% accuracy over the ToN-IoT dataset.

Fog computing has emerged as a promising paradigm that borrows the user-oriented cloud services to the proximity of the Internet-of-Things (IoT) users in sixth-generation (6G) networks. Currently, service providers establish a proprietary fog architecture to prolong a specific group of IoT users by offering resources and services to the edge level. However, this sort of activity creates a service barrier and limits the development of fog services to the IoT-users. Keeping this in mind, we develop a 6G-aware fog federation model for utilizing maximum fog resources and providing demand specific services across the network while maximizing the revenue of fog service providers and guaranteeing the minimum service delay and price for IoT-users. To achieve this goal, we formulate our objective function into a mixed-integer nonlinear problem. By jointly optimizing the dynamic services cost and user demands, a noncooperative Stackelberg game interaction algorithm is formulated to schedule the fog and cloud resources distributively. Further maximizing the profit for the service providers and the seamless resource provisioning, a resource controller is initiated to manage the available fog resources. Extensive simulation analysis over 6G-aware Quality-of-Service parameters demonstrates the superiority of the proposed fog federation model and it reduces up to 15%-20% service delay and 20%-25% of service cost over the standalone fog and cloud frameworks.

Modern scientific instruments generate enormous amount of data. Typically, the data collected from the instruments are stored in one or more files that are then moved to a distant supercomputer for processing. The final results are sent back to the user. In order to make effective use of the time on expensive instruments, experimenters want to process the data as they are generated. They want to stream the data from instruments’ memory directly to a supercomputer’s memory for analysis. Since the compute nodes in a supercomputer are not connected directly to the wide area network, the data streams need to be passed through intermediate gateway nodes. As opposed to the best effort file transfers, data streaming applications require resources at a specific time for a specific period. In this paper, we present a system model for enabling data streaming through gateway nodes and an algorithm to efficiently allocate gateway node resources along with compute nodes. We evaluate the algorithm using real-world traces on the Chameleon Cloud. The results show that our system can schedule compute and gateway resources efficiently for streaming analysis.

By providing the flexible and shared computing and communication resources along with the cloud services, the fog computing became an attractive paradigm to support delay-sensitive tasks in the Internet of Things (IoT). The existing researches for offloading delay-sensitive tasks in a hierarchical fog-cloud environment mostly focused on minimizing the overall communication delay. However, a fair offloading strategy selects a suitable computing device in terms of fog node or cloud server based on the resource requirements of the task while meeting the deadline. In this article, we design a new delay-dependent priority-aware task offloading (DPTO) strategy for scheduling and processing the tasks, generated from the IoT devices to suitable computing devices. The proposed strategy assigns a priority on each task based on its deadline and assigns it to a suitable multilevel-feedback queue. This schema reduces the waiting time of the delay-sensitive tasks on the queue and minimizes the starvation problem of the low priority tasks. Moreover, the DPTO strategy selects an optimal computing device for each task based on its resource availability and transmission time from the IoT device. This strategy minimizes the overall offloading time of the tasks while meeting the deadlines. Finally, the extensive simulation results with various performance parameters show the effectiveness of the proposed strategy over the existing baseline algorithms.

Nowadays, billions of Internet-of-Things devices generate various types of delay-sensitive tasks to process within a limited time frame. By processing the tasks at the network edge using distributed fog devices can efficiently overcome the deficiency of the centralized cloud data center (CDC), i.e., long latency and network congestion. Moreover, to overcome the inefficiency of the local fog devices, i.e., limited processing and storage capabilities, we investigate the collaboration between distributed fog devices and centralized CDC, where the delay-sensitive tasks can preferably be offloaded on the local fog devices, whereas the resource-intensive tasks are offloaded on the resource-rich CDC. However, one of the challenging tasks in the fog-cloud environment is to find a suitable computing device for each real-time task by considering tradeoff between the latency and cost. To meet the above-mentioned challenge, in this article, we introduce an optimal application offloading strategy in the hierarchical fog-cloud environment using the accelerated particle swarm optimization (APSO) technique. The proposed APSO-based strategy finds an optimal computing device (i.e., fog device or cloud server) for each real-time task using multiple quality-of-service parameters, namely, cost and resource utilization (RU). The performance of the proposed algorithm is evaluated using four different real-time data sets with various performance matrices. The experimental results indicate that the proposed strategy outperforms the existing schemes in terms of average delay, computation time, RU, and average cost by 18%, 21%, 27%, and 23%, respectively.