Guest Editorial

Future vehicular Internet-of-Things (IoT) systems feature a large number of devices and multi-access environments where different types of communication,computing,and storage resources must be efficiently utilized.At the same time,novel services such as cooperative autonomous driving and intelligent transportation systems (ITS),that demand unprecedented high accuracy,ultra-low latency,and large bandwidth,are emerging.These services also have an extreme variance in user requirements and resource demands with respect to time,location,and context.Hence,current research is no longer confined to improving reliable communication and system operation in the presence of highly mobile vehicles,which has been the main focus in the past.It is therefore important to empower future vehicular IoT systems with advanced features,such as real-time reactive and proactive cooperation and coordination among different agents (or decision makers),including vehicles,roadside units,base stations,pedestrians,and other entities.

Recently,artificial intelligence (AI) based approaches have been attracting great interest in empowering computer systems.Some collaborative learning approaches,such as federated learning and multi-agent systems,have been used to reduce network traffic and improve the learning efficiency of some smartphone applications.In vehicular IoT systems,collaborative intelligence can be achieved via an efficient collaboration among heterogeneous entities,including vehicles,edges,and the cloud.

This feature topic focuses on the technical challenges and the synergistic effect of collaboration among heterogeneous entities and AI in enabling intelligent perception of the environment,intelligent networking,and intelligent processing of big data in vehicular IoT systems.We were successful at attracting 22 high-quality submissions.All of he submitted papers were evaluated according to the standard reviewing process of China Communications.Following a rigorous peer review process,12 papers were accepted in this special issue.

The accepted papers cover a wide range of topics for enabling collaborative intelligence in vehicular IoT applications,including intelligent perception,radio resource allocation,routing protocols,data sharing,task offloading,and security enhancement.We hope this special issue will open up many exciting and critical future research activities in related fields.

The first paper,“V2I based Environment Perception for Autonomous Vehicles at Intersections” by Duan et al.,proposes a novel approach for collaborative perception about complex road environments while driving.In this approach,a vehicle-road collaborative system is built through vehicle-to-infrastructure (V2I) communications at intersections.Sensors are deployed on roadside to sense the traffic environment around intersections in real-time,and the object detection results are sent back to the autonomous vehicles via V2I links.Compared with the traditional perception methods,this approach uses roadside sensors to assist the surrounding autonomous vehicles for achieving perception enhancement.Therefore,the perception range of the autonomous vehicles in an intersection environment is improved,resulting in a better perception result.

The article by Santa et al.,“Machine Learning-Based Radio Access Technology Selection in the Internet of Moving Things,” proposes a machine learning-based approach for selecting the most adequate transmission interface to transmit a certain message in a multi-access Internet-of-moving-things(IoMT) environment where the cellular,vehicular WiFi,and low power wide area network technologies coexist.The authors explore different machine learning algorithms by taking into account the computing and energy constraints of IoMT end-devices and traffic types.After comparing Naive Bayes,multi-layer perceptron,decision tree and k-nearest neighbor algorithms,the authors employ a decision tree-based approach,and evaluate the approach by using real experiments based on Arduino Uno devices.The results demonstrate the efficiency of the proposal,including energy saving in communication tasks and quality-of-service guarantees for urgent messages.

In the article entitled “A Joint Power and Bandwidth Allocation Method based on Deep Reinforcement Learning for V2V Communications Network in 5G,” Hu et al.propose a new radio resource allocation mechanism based on proximal policy optimization for vehicle-to-vehicle (V2V) communications.In this mechanism,each vehicle is an agent,and it learns the optimal policy for selecting the optimal sub-band and the transmission power level by interacting with its network environment.Based on a deep reinforcement learning,a feasible mapping between states and actions is realized,which enables each agent to continuously improve its strategy and achieve strategy optimization based on past experiences.Simulation results indicate that the proposed mechanism can meet the latency and data rate constraints under the premise of minimizing the interference to V2I communications.

The article by Liao et al.,“CSI Intelligent Feedback for Massive MIMO Systems in V2I Scenarios,”discusses the characteristics of massive multi-input multi-output (MIMO) channels in high-speed V2I scenarios,and proposes a deep learning-based channel state information (CSI) feedback network model.The network model learns the channel characteristics in the V2I scenario at vehicle users,compresses the CSI,and sends it over the channel.The roadside base station receives the data and learns the compressed data characteristics,and then restores the original CSI.Simulation results show that the proposed model achieves a higher accuracy and faster training speed as compared with existing deep learning-based CSI feedback algorithms.

The article by Wang et al.,“Better Platooning Toward Autonomous Driving: Inter-Vehicle Communications with Directional Antenna,” introduces the use of inter-vehicle communications with directional antenna for vehicle platooning.The authors provide an analysis of the relationship between the platoon’s safety and its communication quality.They find that the dominant factor affecting the platoon’ s safety is the beacon delay.Since the beacon delay is closely related to the packet loss rate,the authors propose to utilize directional antennas to reduce the interference of inter-platoon communications.Through extensive simulations,they evaluate the effects of packet delay and inter-vehicle distance in both normal driving and braking scenarios,and verify the usefulness of directional antennas in platooning.

The article by Liu et al.,“Reinforcement Learning Based Dynamic Spectrum Access in Cognitive Internet of Vehicles,” discusses a spectrum access problem for cognitive Internet of vehicles (IoV),and proposes a reinforcement learning-based dynamic spectrum access scheme to improve transmission performance in the licensed spectrum,while reducing harmful interference to primary users.A Q-learning based spectrum access algorithm is employed to enable vehicle users to select the optimal channel,bandwidth and transmission power intelligently under the dynamic change of the primary users’ spectrum states.Simulation results show that the proposed scheme can improve the spectral efficiency and throughput significantly as compared with existing spectrum access schemes.

The article by Bilal et al.,“MADCR: Mobility Aware Dynamic Cluster based Routing Protocol in Internet of Vehicles,” considers the routing problem in IoV,and proposes a mobility-aware dynamic clustering-based routing protocol to maximize the route lifetime and reduce the end-to-end delay.The protocol consists of a cluster formation and cluster head(CH) selection processes where each CH vehicle is selected by using the mayfly optimization algorithm.The CH vehicle collects vehicle data and forwards them to an RSU.Computer simulations are used to prove that the proposed protocol can achieve a lower end-to-end delay and a higher packet delivery ratio than existing baseline protocols.

The article by Huo et al.,“CHRT: Clustering-based Hybrid Re-routing System for Traffic Congestion Avoidance,” proposes CHRT,a clustering-based hybrid re-routing system which dynamically re-routes vehicles based on real-time traffic information.CHRT employs a multi-layer hybrid architecture that combines the advantages of centralized and decentralized control.The central traffic management center accesses the global view of traffic,and the distributed part is composed of vehicles divided into clusters to reduce latency and communiinformation to avoid secondary congestion.Furthermore,to plan the optimal routes for vehicles while alleviating global traffic congestion,a multi-metric re-routing algorithm is also used.Through extensive simulations based on the SUMO traffic simulator,the authors show the advantage of CHRT over existing baselines in terms of the vehicle traveling time,fuel consumption,and CO2 emissions.

The article by Hawbani et al.,“A Novel Artificial Bee Colony and Blockchain-Based Secure Clustering Routing Scheme for FANET,” proposes a clustering-based secure routing scheme for flying ad hoc networks,which aims to solve the routing and data security problem.Each update in the routing table is recorded as a blockchain transaction.A lightweight consensus scheme is also proposed to enable efficient blockchain transactions.The CH selection is based on residual energy,online time,reputation,blockchain transactions,mobility,and connectivity by using an improved artificial bee colony optimization (IABC) algorithm.Simulation results demonstrate that the proposed scheme outperforms existing baseline approaches in terms of the packet delivery ratio,end-to-end delay,throughput,resilience against security attacks,and block processing time.

While data sharing in IoV can enable personalized services for users,the privacy and efficiency of data sharing need to be considered.Federated learning allows data exchange among multiple vehicles without violating data privacy.The article by Yuan et al.,“A Federated Bidirectional Connection Broad Learning Scheme for Secure Data Sharing in Internet of Vehicles,” proposes a federated bidirectional connection broad learning scheme (FeBBLS) to solve the data sharing issues.The authors adopt the bidirectional connection broad learning system (BiBLS)to process data at vehicular nodes.The central server aggregates the collected parameters from vehicular nodes.Moreover,they also use transfer learning to solve unbalanced data issues.Simulation results show that FeBBLS can improve the efficiency of data sharing while protecting user privacy.

The article by Zhou et al.,“Deep Reinforcement Learning-Based URLLC-Aware Task Offloading in Collaborative Vehicular Networks,” considers the optimization of task offloading with ultra-reliable and low-latency communications (URLLC) in highly dynamic collaborative vehicular networks,and proposes a deep reinforcement learning-based URLLC-aware task offloading algorithm to maximize the throughput of vehicle users while satisfying the URLLC constraints in a best-effort way.The authors exploit Q-learning to optimize the task offloading strategy based solely on the observed performance,and deep neural networks to approximate the Q-function to cope with the problem of dimensionality curse.Compared with existing task offloading algorithms,this approach achieves superior performance in terms of throughput,convergence,latency,and reliability.

Last but not the least,the article by Liu et al.,“Game Theoretical Secure Wireless Communication for UAV-assisted Vehicular Internet of Things,” discusses the security threat of jamming attacks in unmanned aerial vehicle (UAV) assisted vehicular IoT,and proposes a game theory-based secure data transmission scheme.The authors exploit the offensive and defensive games to model the interactions between the normal UAVs and the jammers.The strategy of the normal UAV is to determine whether to transmit data,while that of the jammer is whether to interfere.The authors formulate two optimization problems,i.e.,maximizing the utilities of both UAVs and jammers.Then,they exploit the backward induction method to analyze the proposed countermeasures,and finally solve the optimal solution of the optimization problem.Simulation results show that the proposed scheme can improve the wireless communications performance under the attacks of jammers as compared to existing baselines.

In conclusion,the Guest Editors of this feature topic would like to thank all the authors for their valuable contributions,and the anonymous reviewers for their constructive comments and suggestions.We also would like to acknowledge the guidance from Ms.Fan and the editorial team of China Communications.