Efficient task offloading in dynamic fog computing environments requires adaptive decision-making under uncertainty. This paper proposes a Bayesian Deep Neural Network (BDNN)-empowered Thompson Sampling (TS) framework for context-aware task offloading, enabling intelligent resource allocation while balancing exploration and exploitation. The BDNN models the stochastic reward function by learning a posterior distribution over network weights, capturing the uncertainty in offloading decisions. At each time step, the task node samples a set of weights from the learned posterior to estimate the expected reward of each helper node, facilitating adaptive decision-making in dynamic network conditions. Experimental results demonstrate that our approach outperforms conventional heuristics and deep learning-based methods, achieving lower latency, improved resource utilization, and better offloading efficiency in fog computing environments.

Task offloading in multi-tier fog computing networks presents significant challenges due to dynamic resource availability, fluctuating network conditions, and the inherent uncertainty in task execution outcomes. To address these issues, we propose an uncertainty-aware task offloading framework based on Federated Probabilistic Deep Neural Networks (Fed-PDNNs). In our approach, edge devices independently train local PDNNs to estimate both the expected reward and predictive uncertainty for offloading tasks to candidate fog nodes. These models provide a principled way to quantify epistemic uncertainty and enable effective decision-making through Thompson Sampling. To enhance generalization and maintain data privacy, the PDNNs are periodically synchronized via federated learning at the fog tier, where fog nodes aggregate model updates from connected edge devices without requiring access to raw data. Our framework naturally supports decentralized control, adapts to heterogeneous environments, and balances exploration and exploitation during offloading. Extensive experiments on synthetic and real-world workloads demonstrate that Fed-PDNN significantly outperforms baseline methods in terms of task completion delay, offloading success rate, and robustness under dynamic network conditions, offering a scalable, intelligent, and privacy-preserving solution for next-generation edge-fog-cloud systems.

In the realm of dynamic fog networks, the imper-ative to efficiently offload delay-sensitive tasks while minimizing latency remains a formidable challenge. To address this, we propose a novel approach termed Digital Twin-empowered Contextual Bandit Learning based Matching (DT-CBLM). This framework harnesses the synergistic power of digital twins, which provide virtual representations of physical devices and environments, and contextual bandit learning techniques. By integrating digital twins into the task offloading process, DT-CBLM enables intelligent task-peer matching by considering contextual information such as device status, network conditions, and task characteristics. Through adaptive offloading strategies, DT-CBLM aims to minimize task completion latency while optimizing resource utilization and system efficiency in dynamic fog environments. Theoretical foundations, algorithmic details, and empirical evaluations showcase the efficacy of DT-CBLM in enhancing system performance.

This paper presents a novel approach utilizing Multi-Armed Bandit (MAB) for parallel computation in dynamic fog computing networks. Fog computing, vital for processing data near the network edge, faces challenges in task allocation while maintaining performance. To address this, a decentralized matching approach leveraging MAB learning algorithms is proposed. Using Thomson sampling (TS), tasks are allocated based on resource availability and network conditions. This decentralized approach empowers fog nodes to autonomously make offloading decisions, enhancing adaptability. Extensive simulations demonstrate the effectiveness of the proposed method in reducing task completion time and optimizing resource utilization in dynamic fog computing environments.

Matching theory has been efficiently applied in fog computing networks (FCNs) to design distributed task offloading algorithms in the presence of selfishness and rationals of fog nodes. Given the dynamic nature of fog computing environment, it is challenging to obtain the stable matching since the preference relations of two sides of matching game is unknown a prior. To address this challenge, this paper proposes RL-MATCH, a framework for parallel computation offloading in dynamic fog computing networks (FCNs). RL-MATCH is based on the matching theory and Thompson Sampling (TS) empowered Multi-Armed Bandit (MAB) learning to deal with the inherent challenges allowing task nodes with needed computation tasks to estimate the informed preference relations of helper nodes with available computing resource quickly and accurately. Extensive simulation results demonstrate the potential advantages of the TS based learning over the$\epsilon$-greedy and upper confidence bound (UCB) based baselines.

Matching theory has been applied to design efficient offloading solutions to the multi-task multi-helper (MTMH) problem in the fog computing networks, which is modeled as a matching game between a set of task nodes (TNs) having task computation needs and a set of helper nodes (HNs) having available computing resources. However, the uncertainty of computing resource availability of HNs as well as dynamics of QoS requirements of tasks result in the lack of preferences of TN side that mainly poses a critical challenge to obtain a stable and reliable matching outcome. To address this challenge, we apply a multi-armed bandit (MAB) learning using Thomson sampling (TS) mechanism to acquire better exploitation and exploration trade-off, allowing TNs to match with their corresponding HNs efficiently. Based on these, this paper proposes online learning based matching (OLM) algorithm for decentralized task offloading to reduce the offloading delay in Fog-enabled IoT Systems. Extensive simulation results demonstrate the potential advantages of the TS-type algorithm over the $\epsilon$-greedy and UCB based offloading algorithms.

Fog computing technology has developed to support delay-sensitive applications by offering shared and adaptable communication, computation, and storage resources along the cloud-to-things continuum in Internet of Things (loT) and cyber-physical systems (CPS). In order to minimize the delay of every task, task nodes (TNs) with computation-intensive and delay-sensitive tasks should have efficient strategies to select the optimal helper nodes (HN s) having spare computation resources for task offloading operations. However, the dynamic nature of fog computing environment characterized by the time varying change of HN computing resources as well as various types of tasks with different quality of service (QoS requirements) impose as inherent challenges for designing the efficient task offloading algorithms. To deal with these challenges, we apply the principle of multi-armed bandit (MAB) learning method to efficiently learn the uncertainty of fog computing environment. In particularly, we use Thomson sampling (TS) technique to acquire better exploitation and exploration trade-off, allowing TNs to select their corresponding HN s efficiently. Extensive simulation results demonstrate the potential advantages of the TS-type algorithm over the$\epsilon-\mathbf{greedy}$and U CB based offloading algorithms.

Fog computing systems (FCS) have been widely integrated in the IoT-based applications aiming to improve the quality of services (QoS) such as low response service delay by performing the task computation nearby the task generation sources (i.e., IoT devices) on behalf of remote cloud servers. However, to achieve the objective of delay reduction remains challenging for offloading strategies due to the resource limitation of fog devices. In addition, a high rate of task requests combined with heavy tasks (i.e., large task size) may cause a high imbalance of workload distribution among the heterogeneous fog devices. To cope with the situation, this paper proposes a dynamic task offloading (DTO) approach, which is based on the resource states of fog devices to derive the task offloading policy dynamically. Accordingly, a task can be executed by either a single fog or multiple fog devices through parallel computation of subtasks to reduce the task execution delay. Through the extensive simulation analysis, the proposed approaches show potential advantages in reducing the average delay significantly in the systems with high rate of service requests and heterogeneous fog environment compared with the existing solutions.

In the IoT-based systems, the fog computing allows the fog nodes to offload and process tasks requested from IoT-enabled devices in a distributed manner instead of the centralized cloud servers to reduce the response delay. However, achieving such a benefit is still challenging in the systems with high rate of requests, which imply long queues of tasks in the fog nodes, thus exposing probably an inefficiency in terms of latency to offload the tasks. In addition, a complicated heterogeneous degree in the fog environment introduces an additional issue that many of single fogs can not process heavy tasks due to lack of available resources or limited computing capabilities. To cope with the situation, this article introduces FRATO (Fog Resource aware Adaptive Task Offloading) - a framework for the IoT-fog-cloud systems to offer the minimal service provisioning delay through an adaptive task offloading mechanism. Fundamentally, FRATO is based on the fog resource to select flexibly the optimal offloading policy, which in particular includes a collaborative task offloading solution based on the data fragment concept. In addition, two distributed fog resource allocation algorithms, namely TPRA and MaxRU are developed to deploy the optimized offloading solutions efficiently in cases of resource competition. Through the extensive simulation analysis, the FRATO-based service provisioning approaches show potential advantages in reducing the average delay significantly in the systems with high rate of service requests and heterogeneous fog environment compared with the existing solutions.

This study designs and presents routing protocols multi‐channel, multi‐hop cognitive radio sensor network (CRSN), which are based on spectrum sensing, spectrum‐availability analysis and per‐hop delay analysis. In order to cope with the spectrum dynamics in the CRSNs, a framework is proposed to estimate the remaining duration of spectrum availability which is based on the expected idle length of channels, sensing period and communication history on the channel. Such methodology allows cognitive users to mitigate communication interruptions caused by arrivals of primary users, thus to improve the transmission performance over each hop. In particular, in combination with the per‐hop delay analysis, the routing paths from sources to the sink can be established such as they provide the shortest delay opportunistically or shorter delay but reliable communication along the selected paths. Extensive simulation results show that the proposed routing algorithms potentially improve the network performances in terms of throughput, delay and energy consumption.

The Physical Internet (PI, or π) paradigm has been developed to be a global logistics system that aims to move, handle, store, and transport logistics products in a sustainable and efficient way. To achieve the goal, the PI requires a high-level interconnectivity in the physical, informational, and operational aspects enabled by an interconnected network of intermodal hubs, collaborative protocols, and standardized, modular, and smart containers. In this context, PI is a key player poised to benefit from the Internet-of-Things (IoT) revolution since it potentially provides an end-to-end visibility of the PI objects, operations, and systems through ubiquitous information exchange. This article is to investigate opportunities of application of the IoT technology in the PI vision. In addition, an IoT ecosystem (π-IoT) encompassing key enabling IoT technologies, building blocks, and a service-oriented architecture (SoA) is proposed as a potential component for accelerating the implementation of PI. The major challenges regarding the deployment of IoT into the emerging logistics concept are also discussed intensively for further research.

This paper proposes a fast and reliable routing protocol (FaReli) for multi-channel, multi-hop cognitive wireless sensor networks (CRSNs). Being aware of the spectrum dynamics in the CRSN, a framework is proposed to estimate the remaining duration of spectrum availability which is based on the expected idle length of channels, sensing period, and communication history on the channel. Specially, in combination with the per hop delay analysis, the routing path from sources to the sink can be established such as it provides a shorter delay and reliable communication. Simulation results are conducted to demonstrate that the proposed routing algorithm based on spectrum and delay analysis potentially improve the network performances in terms of throughput, delay.

This paper proposes a routing scheme that enhances energy consumption and end-to-end delay for large-scale Industrial Internet of Things (IIoT) systems based on IEEE 802.15.4a MAC. In the current IIoT, a larger-scale and complex deployment has been a noticeable obstacle for minimizing power consumption and routing on real-time. Thus, the proposed algorithm is targeted at large-scale systems where data are aggregated through different clusters on their way to the sink. Moreover, a hierarchical system framework is employed to promote scalability of IIoT elements. By estimating the residual energy and hop counts for each path, the data can be forwarded to the destination through the optimal path. Simulation results show that the scheme can reduce the energy consumption and end-to-end delay effectively.

Wireless Nano-Sensor networks (WNSN) consist of nanosensors equipped with nanotransceivers and nanoantennas to operate in Terahertz frequency band (0.1–10THz). Due to the peculiarities of this communication channel (such as, very short range of transmission (under lm), high interference, high path loss) and limited capabilities of nano-nodes (such as, computing, sensing, memory, energy), the existing ranging techniques designed for traditional wireless sensor networks are not longer used in the WNSNs. In this paper, two ranging algorithms based on hop-counting methods are developed to estimate the location of every nanosensor within certain area and distance between nodes in the networks. The first technique uses flooding mechanism to forward the packets to all nodes in the networks and count number of hops between two measured nodes. To overcome the problems of high overhead, duplication packets, and waste of consumption energy, the second algorithm based on clusters is developed. In this way, all sensornodes are grouped into different clusters. Cluster heads will communicate together and count the number of hops. The performance of the algorithms are analyzed in terms of estimated distance, delay by taking also account the energy constrains of nanosensors. The simulation results show that, by being aware of the limitations of nanosensors, the proposed protocols are able to support WNSNs with very high density in ranging and localizing.