In the Internet of Vehicles (IoV), federated learning (FL) provides a privacy-preserving solution by aggregating local models without sharing data. Traditional supervised learning requires image data with labels, but data labeling involves significant manual effort. Federated self-supervised learning (FSSL) utilizes self-supervised learning (SSL) for local training in FL, eliminating the need for labels while protecting privacy. Compared to other SSL methods, Momentum Contrast (MoCo) reduces the demand for computing resources and storage space by creating a dictionary. However, using MoCo in FSSL requires uploading the local dictionary from vehicles to base station (BS), which poses a risk of privacy leakage. Simplified contrast (SimCo) addresses the privacy leakage issue in MoCo-based FSSL by using dual temperature instead of a dictionary to control sample distribution. Additionally, considering the negative impact of motion blur on model aggregation, and based on SimCo, we propose a motion blur-resistant FSSL method, referred to as BFSSL. Furthermore, we address energy consumption and delay in the BFSSL process by proposing a deep reinforcement learning (DRL)-based resource allocation scheme, called DRL-BFSSL. In this scheme, BS allocates the central processing unit (CPU) frequency and transmission power of vehicles to minimize energy consumption and latency, while aggregating received models based on the motion blur level. Simulation results validate the effectiveness of our proposed aggregation and resource allocation methods.

Federated learning (FL) can protect the privacy of the vehicles in vehicle edge computing (VEC) to a certain extent through sharing the gradients of vehicles’ local models instead of the local data. The gradients of vehicles’ local models are usually large for the vehicular artificial intelligence (AI) applications, thus transmitting such large gradients would cause large per-round latency. Gradient quantization has been proposed as one effective approach to reduce the per-round latency in FL enabled VEC through compressing gradients and reducing the number of bits, i.e., the quantization level, to transmit gradients. The selection of quantization level and thresholds determines the quantization error (QE), which further affects the model accuracy and training time. To do so, the total training time and QE become two key metrics for the FL enabled VEC. It is critical to jointly optimize the total training time and QE for the FL enabled VEC. However, the time-varying channel condition causes more challenges to solve this problem. In this article, we propose a distributed deep reinforcement learning (DRL)-based quantization level allocation scheme to optimize the long-term reward in terms of the total training time and QE. Extensive simulations identify the optimal weighted factors between the total training time and QE, and demonstrate the feasibility and effectiveness of the proposed scheme.

Edge caching is a promising solution for next-generation networks by empowering caching units in small-cell base stations (SBSs), which allows user equipments (UEs) to fetch users’ requested contents that have been pre-cached in SBSs. It is crucial for SBSs to predict accurate popular contents through learning while protecting users’ personal information. Traditional federated learning (FL) can protect users’ privacy but the data discrepancies among UEs can lead to a degradation in model quality. Therefore, it is necessary to train personalized local models for each UE to predict popular contents accurately. In addition, the cached contents can be shared among adjacent SBSs in next-generation networks, thus caching predicted popular contents in different SBSs may affect the cost to fetch contents. Hence, it is critical to determine where the popular contents are cached cooperatively. To address these issues, we propose a cooperative edge caching scheme based on elastic federated and multi-agent deep reinforcement learning (CEFMR) to optimize the cost in the network. We first propose an elastic FL algorithm to train the personalized model for each UE, where adversarial autoencoder (AAE) model is adopted for training to improve the prediction accuracy, then a popular content prediction algorithm is proposed to predict the popular contents for each SBS based on the trained AAE model. Finally, we propose a multi-agent deep reinforcement learning (MADRL) based algorithm to decide where the predicted popular contents are collaboratively cached among SBSs. Our experimental results demonstrate the superiority of our proposed scheme to existing baseline caching schemes.

Vehicles in platoons need to process many tasks to support various real-time vehicular applications. When a task arrives at a vehicle, the vehicle may not process the task due to its limited computation resource. In this case, it usually requests to offload the task to other vehicles in the platoon for processing. However, when the computation resources of all the vehicles in the platoon are insufficient, the task cannot be processed in time through offloading to the other vehicles in the platoon. Vehicular fog computing (VFC)-assisted platoon can solve this problem through offloading the task to the VFC which is formed by the vehicles driving near the platoon. Offloading delay is an important performance metric, which is impacted by both the offloading strategy for deciding where the task is offloaded and the number of the allocated vehicles in VFC to process the task. Thus, it is critical to propose an offloading strategy to minimize the offloading delay. In the VFC-assisted platoon system, vehicles usually adopt the IEEE 802.11p distributed coordination function (DCF) mechanism while having various computation resources. Moreover, when vehicles arrive and depart the VFC randomly, their tasks also arrive at and depart the system randomly. In this paper, we propose a semi-Markov decision process (SMDP) based offloading strategy while considering these factors to obtain the maximal long-term reward reflecting the offloading delay. Our research provides a robust strategy for task offloading in VFC systems, its effectiveness is demonstrated through simulation experiments and comparison with benchmark strategies.

Vehicular networks enable vehicles support some real-time applications through training data. Due to the limited computing capability of vehicles, vehicles usually transmit data to a road side unit (RSU) deployed along the road to process data collaboratively. However, vehicles are usually reluctant to share data with each other due to the inevitable data privacy. For the traditional federated learning (FL), vehicles train the data locally to obtain a local model and then upload the local model to the RSU to update the global model through aggregation, thus the data privacy can be protected through sharing model instead of raw data. The traditional FL requires to update the global model synchronously, i.e., the RSU needs to wait for all vehicles to upload local models to update the global model. However, vehicles may usually drive out of the coverage of the marked RSU before they obtain their local models through training, which reduces the accuracy of the global model. In this paper, a mobility-aware vehicular asynchronous federated learning (AFL) is proposed to solve this problem, where the RSU updates the global model once it receives a local model from a vehicle where the mobility of vehicles, amount of data and computing capability are taken into account. Simulation experiments validate that our scheme outperforms the conventional AFL scheme.

Fog nodes in the vicinity of IoT devices are promising to provision low-latency services by offloading tasks from IoT devices to them. Mobile IoT is composed by mobile IoT devices, such as vehicles, wearable devices, and smartphones. Owing to the time-varying channel conditions, traffic loads, and computing loads, it is challenging to improve the Quality of Service (QoS) of mobile IoT devices. As task delay consists of both the transmission delay and computing delay, we investigate the resource allocation (i.e., including both radio resource and computation resource) in both the wireless channel and fog node to minimize the delay of all tasks while their QoS constraints are satisfied. We formulate the resource allocation problem into an integer nonlinear problem, where both the radio resource and computation resource are taken into account. As IoT tasks are dynamic, the resource allocation for different tasks are coupled with each other and the future information is impractical to be obtained. Therefore, we design an online reinforcement learning algorithm to make the suboptimal decision in real time based on the system’s experience replay data. The performance of the designed algorithm has been demonstrated by extensive simulation results.

The performance of federated learning systems is bottlenecked by communication costs and training variance. The communication overhead problem is usually addressed by three communication-reduction techniques, namely, model compression, partial device participation, and periodic aggregation, at the cost of increased training variance. Different from traditional distributed learning systems, federated learning suffers from data heterogeneity (since the devices sample their data from possibly different distributions), which induces additional variance among devices during training. Various variance-reduced training algorithms have been introduced to combat the effects of data heterogeneity, while they usually cost additional communication resources to deliver necessary control information. Additionally, data privacy remains a critical issue in FL and, thus, there have been attempts at bringing Differential Privacy to this framework as a mediator between utility and privacy requirements. This article investigates the tradeoffs between communication costs and training variance under a resource-constrained federated system theoretically and experimentally, and studies how communication reduction techniques interplay in a differentially private setting. The results provide important insights into designing practical privacy-aware federated learning systems.

Multi unmanned aerial vehicle (UAV) network is a promising solution to providing wireless coverage to ground users in challenging rural areas (such as Internet of Things (IoT) devices in farmlands), where the traditional cellular networks are sparse or unavailable. A key challenge in such networks is the 3-D placement of all UAV base stations (BSs) such that the formed multi-UAV network: 1) utilizes a minimum number of UAVs while ensuring—2) backhaul connectivity directly (or via other UAVs) to the nearby terrestrial BS; and 3) wireless coverage to all ground users in the area of operation. This joint backhaul-and-coverage-aware drone deployment (BoaRD) problem is largely unaddressed in the literature and, thus, is the focus of this article. We first formulate the BoaRD problem as integer linear programming (ILP). However, the problem is NP-hard and, therefore, we propose a low complexity algorithm with a provable performance guarantee to solve the problem efficiently. Our simulation study shows that the Proposed algorithm performs very close to that of the Optimal algorithm (solved using ILP solver) for smaller scenarios, where the area size and the number of users are relatively small. For larger scenarios, where the area size and the number of users are relatively large, the proposed algorithm greatly outperforms the baseline approaches—Backhaul-aware Greedy and random algorithm, respectively, by up to 17% and 95% in utilizing fewer UAVs while ensuring 100% ground-user coverage and backhaul connectivity for all deployed UAVs across all considered simulation setting.

Platooning strategy is an important part of autonomous driving technology. Due to the limited resource of autonomous vehicles in platoons, mobile-edge computing (MEC) is usually used to assist vehicles in platoons to obtain useful information, increasing its safety. Specifically, vehicles usually adopt the IEEE 802.11 distributed coordination function (DCF) mechanism to transmit large amount of data to the base station (BS) through vehicle-to-infrastructure (V2I) communications, where the useful information can be extracted by the edge server connected to the BS and then sent back to the vehicles to make correct decisions in time. However, vehicles may be moving on different lanes with different velocities, which incurs the unfair access due to the characteristics of platoons, i.e., vehicles on different lanes transmit different amount of data to the BS when they pass through the coverage of the BS, which also results in the different amount of useful information received by various vehicles. Moreover, age of information (AoI) is an important performance metric to measure the freshness of the data. Large average age of data implies not receiving the useful information in time. It is necessary to design an access scheme to jointly optimize the fairness and data freshness. In this article, we formulate a joint optimization problem in the MEC-assisted V2I networks and present a multiobjective optimization scheme to solve the problem through adjusting the minimum contention window under the IEEE 802.11 DCF mode according to the velocities of vehicles. The effectiveness of the scheme has been demonstrated by simulation.

With the development of Internet of Vehicles, the platooning strategy has been widely studied as the potential approach to ensure the safety of autonomous driving. Vehicles in the form of platoon adopt 802.11p to exchange messages through vehicle-to-vehicle (V2V) communications. When multiple platoons arrive at an intersection, the leader vehicle of each platoon adjusts its movement characteristics to ensure that it can cross the intersection, and thus the following vehicles have to adjust their movement characteristics accordingly. In this case, the time-varying connectivity among vehicles leads to the significant nonstationary performance change in platooning communications, which may incur safety issues. In this article, we construct the time-dependent model to evaluate the platooning communication performance at the intersection based on the initial movement characteristics. We first consider the movement behaviors of vehicles at the intersection, including turning, accelerating, decelerating, and stopping as well as the periodic change of traffic lights to construct a movement model and then establish a hearing network to reflect the time-varying connectivity among vehicles. Afterward, we adopt the pointwise stationary fluid flow approximation (PSFFA) to model the nonstationary behavior of the transmission queue. Then, we consider four access categories (ACs) and continuous backoff freezing of 802.11p to construct the models to describe the time-dependent access process of 802.11p. Finally, based on the time-dependent model, the packet transmission delay (PTD) and packet delivery ratio (PDR) are derived. The accuracy of our proposed model is verified by comparing the simulation results with analytical results.

Vehicular edge computing (VEC) is envisioned as a promising approach to process the explosive computation tasks of vehicular user (VU). In the VEC system, each VU allocates power to process partial tasks through offloading and the remaining tasks through local execution. During the offloading, each VU adopts the multi-input multi-output and non-orthogonal multiple access (MIMO-NOMA) channel to improve the channel spectrum efficiency and capacity. However, the channel condition is uncertain due to the channel interference among VUs caused by the MIMO-NOMA channel and the time-varying path loss caused by the mobility of each VU. In addition, the task arrival of each VU is stochastic in the real world. The stochastic task arrival and uncertain channel condition affect greatly on the power consumption and latency of tasks for each VU. It is critical to design an optimal power allocation scheme considering the stochastic task arrival and channel variation to optimize the long-term reward, including the power consumption and latency in the MIMO-NOMA VEC. Different from the traditional centralized deep reinforcement learning (DRL)-based scheme, this article constructs a decentralized DRL framework to formulate the power allocation optimization problem, where the local observations are selected as the state. The deep deterministic policy gradient (DDPG) algorithm is adopted to learn the optimal power allocation scheme based on the decentralized DRL framework. Simulation results demonstrate that our proposed power allocation scheme outperforms the existing schemes.

Vehicular fog computing (VFC) is envisioned as a promising solution to process the explosive tasks in autonomous vehicular networks. In the VFC system, task offloading is the key technique to process the computation-intensive tasks efficiently. In the task offloading, the task is transmitted to the VFC system according to the 802.11p standard and processed by the computation resources in the VFC system. The delay of task offloading, consisting of the transmission delay and computing delay, is extremely critical especially for some delay-sensitive applications. Furthermore, the long-term reward of the system (i.e., jointly considers the transmission delay, computing delay, available resources, and diversity of vehicles and tasks) becomes a significantly important issue for providers. Thus, in this article, we propose an optimal task offloading scheme to maximize the long-term reward of the system where 802.11p is employed as the transmission protocol for the communications between vehicles. Specifically, a task offloading problem based on a semi-Markov decision process (SMDP) is formulated. To solve this problem, we utilize an iterative algorithm based on the Bellman equation to approach the desired solution. The performance of the proposed scheme has been demonstrated by extensive numerical results.

With the development of 5G, the Internet of Vehicles (IoV) evolves to be one important component of the Internet of Things (IoT), where vehicles and public infrastructure communicate with each other through a IEEE 802.11p EDCA mechanism to support four access categories (ACs) to access a channel. Due to the mobility of the vehicles, the network topology is time varying and thus incurs a dynamic network performance. There are many works on the stationary performance of 802.11p EDCA and some on real-time performance, but existing work does not consider real-time performance under extreme highway scenario. In this paper, we consider four ACs defined in the 802.11p EDCA mechanism to evaluate the limit of the real-time network performance in an extreme highway scenario, i.e., all vehicles keep the minimum safety distance between each other. The performance of the model has been demonstrated through simulations. It is found that some ACs can meet real-time requirements while others cannot in the extreme scenario.

Various emerging vehicular applications such as autonomous driving and safety early warning are used to improve the traffic safety and ensure passenger comfort. The completion of these applications necessitates significant computational resources to perform enormous latency-sensitive/nonlatency-sensitive and computation-intensive tasks. It is hard for vehicles to satisfy the computation requirements of these applications due to the limit computational capability of the on-board computer. To solve the problem, many works have proposed some efficient task offloading schemes in computing paradigms such as mobile fog computing (MFC) for the vehicular network. In the MFC, vehicles adopt the IEEE 802.11p protocol to transmit tasks. According to the IEEE 802.11p, tasks can be divided into high priority and low priority according to the delay requirements. However, no existing task offloading work takes into account the different priorities of tasks transmitted by different access categories (ACs) of IEEE 802.11p. In this paper, we propose an efficient task offloading strategy to maximize the long-term expected system reward in terms of reducing the executing time of tasks. Specifically, we jointly consider the impact of priorities of tasks transmitted by different ACs, mobility of vehicles, and the arrival/departure of computing tasks, and then transform the offloading problem into a semi-Markov decision process (SMDP) model. Afterwards, we adopt the relative value iterative algorithm to solve the SMDP model to find the optimal task offloading strategy. Finally, we evaluate the performance of the proposed scheme by extensive experiments. Numerical results indicate that the proposed offloading strategy performs well compared to the greedy algorithm.

Edge computing enables data collected by Internet of Things (IoT) devices to be stored in and processed by local fog nodes as well as allows IoT users to access IoT applications via these nodes at the same time. In this case, the communications latency critically affects the response time of IoT user requests. Owing to the dynamic distribution of IoT users [i.e., user equipments (UEs)], drone base station (DBS), which can be flexibly deployed for hotspot areas, can potentially improve the wireless latency of IoT users by mitigating the heavy traffic loads of macro BSs. Drone-based communications poses two major challenges: 1) the DBS should be deployed in suitable areas with heavy traffic demands to serve more UEs and 2) the traffic loads in the network should be allocated among macro BSs and DBSs to avoid instigating traffic congestions. Therefore, we propose a traffic load balancing scheme in such drone-assisted fog network to minimize the wireless latency of IoT users. In the scheme, we divide the problem into two subproblems and design two algorithms to optimize the DBS placement and user association, respectively. Extensive simulations have been set up to validate the performance of the proposed scheme.

Empowered by computing resources at the network edge, data sensed from Internet of Things (IoT) devices can be processed and stored in their nearby cloudlets to reduce the traffic load in the core network, while various IoT applications can be run in cloudlets to reduce the response time between IoT users (e.g., user equipment in mobile networks) and cloudlets. Considering the spatial and temporal dynamics of each application's workloads among cloudlets, the workload allocation among cloudlets for each IoT application affects the response time of the application's requests. While assigning IoT users' requests to their nearby cloudlets can minimize the network delay, the computing delay of a type of requests may be unbearable if the corresponding virtual machine of the application in a cloudlet is overloaded. To solve this problem, we design an application aware workload allocation scheme for edge computing-based IoT to minimize the response time of IoT application requests by deciding the destination cloudlets for each IoT user's different types of requests and the amount of computing resources allocated for each application in each cloudlet. In this scheme, both the network delay and computing delay are taken into account, i.e., IoT users' requests are more likely assigned to closer and lightly loaded cloudlets. Meanwhile, the scheme will dynamically adjust computing resources of different applications in each cloudlet based on their workloads, thus reducing the computing delay of all requests in the cloudlet. The performance of the proposed scheme has been validated by extensive simulations.

As accessing computing resources from the remote cloud for big data processing inherently incurs high end-to-end (E2E) delay for mobile users, cloudlets, which are deployed at the edge of networks, can potentially mitigate this problem. Although load offloading in cloudlet networks has been proposed, placing the cloudlets to minimize the deployment cost of cloudlet providers and E2E delay of user requests has not been addressed so far. The locations and number of cloudlets and their servers have a crucial impact on both the deployment cost and E2E delay of user requests. Therefore, in this paper, we propose the Cost Aware cloudlet PlAcement in moBiLe Edge computing strategy (CAPABLE) to optimize the tradeoff between the deployment cost and E2E delay. When cloudlets are already placed in the network, we also design a load allocation scheme to minimize the E2E delay of user requests by assigning the workload of each region to the suitable cloudlets. The performance of CAPABLE is demonstrated by extensive simulation results.

Greening information and communications technology is becoming an environmental and economic sine qua non, and has attracted much research attention. For a cellular network, base stations (BSs) incur more than 50% of the energy consumption of the whole network. Therefore, BSs can be powered by green energy to reduce its on-grid power consumption. Meanwhile, the throughput has always been a critical issue in cellular networks. Since the throughput and energy consumption mutually affect each other, saving on-grid power is at the cost of sacrificing a certain amount of throughput. In this paper, we propose a Throughput Aware and Green Energy aware user association (TAGE) scheme in heterogeneous cellular networks (HCNs) to optimize the trade-off between the throughput and on-grid power consumption. Meanwhile, we employ an energy-throughput coefficient α to control the energy-throughput tradeoff. The simulation results verify that TAGE improves the effective throughput and saves a significant amount of on-grid power for HCNs.

Greening information and communications technology is becoming an environmental and economic sine qua non, and has attracted much research attention. For a cellular network, the base stations (BSs) cost more than 50% of the energy consumption of the whole network. Therefore, BSs can be powered by green energy to reduce its on-grid power consumption. In this paper, we propose a greeN Energy Aware user associaTion (NEAT) scheme in the two-tier green heterogeneous network, that enables a BS depleting of green energy to offload its traffic load to other BSs with excessive green energy. Since the Macro BS (MBS) and Pico BS (PBS) employ different partitions of the licensed spectrum, we also consider the bandwidth allocation and adjust the two spectrum partitions dynamically. However, in the NEAT scheme, achieving the optimal user association in terms of minimizing the on-grid power consumption of BSs, is NP-hard. Therefore, we propose a heuristic NEAT algorithm to approximate the optimal solution with low computational complexity. Finally, the performance and viability of the algorithm are substantiated by simulation results.

We propose a Green Cloudlet Network (GCN) architecture to provide seamless Mobile Cloud Computing (MCC) services to User Equipments (UEs) with low latency in which each cloudlet is powered by both green and brown energy. Fully utilizing green energy can significantly reduce the operational cost of cloudlet providers. However, owing to the spatial dynamics of energy demand and green energy generation, the energy gap among different cloudlets in the network is unbalanced, i.e., some cloudlets' energy demands can be fully provided by green energy but others need to utilize on-grid energy (i.e., brown energy) to satisfy their energy demands. We propose a Green-energy awarE Avatar migRation (GEAR) strategy to minimize the on-grid energy consumption in GCN by redistributing the energy demands via Avatar migration among cloudlets according to cloudlets' green energy generation. Furthermore, GEAR ensures the Service Level Agreement (SLA) in terms of the maximum Avatar propagation delay by avoiding Avatars hosted in the remote cloudlets. We formulate the GEAR strategy as a mixed integer linear programming problem, which is NP-hard, and thus apply the Branch and Bound search to find its sub-optimal solution. Simulation results demonstrate that GEAR can save on-grid energy consumption significantly as compared to the Follow me AvataR (FAR) migration strategy, which aims to minimize the propagation delay between an UE and its Avatar.