Node Selfishness And Incentive Mechanism In Vehicular Communication Networks

With the rapid development of communication technologies and Internet of things,the vehicular network as the deep integration of information communication and automobile transportation becomes one of the most promising networking technology in Internet of thing paradigm.The vehicular networks carry the increasingly personalized and intelligent transportation application needs,which are often manifested as delay sensitive and computation-intensive tasks.Therefore,the vehicular networks are required to have stronger communication capabilities and higher robustness.Obviously,centralized or single-hop communication cannot meet the current application requirements.Multi-vehicle cooperative communication technology will be the key to facilitate the intelligent transportation system in the future.The collaborative communication technology of vehicular networks needs to provide highly reliable information transmission services for various traffic applications including security applications.However,most of the existing researches based on multivehicle cooperative communication assume that the network nodes are definitely cooperative,while in reality,the nodes of the vehicular network(vehicles,roadside units,portable devices carried by pedestrians,etc.)belong to different interest groups.For the purpose of saving limited resources,nodes are unwilling to provide free network services such as relay,thus showing selfishness.Such selfish behavior will seriously affect the QoS(Quality of Service,QoS)and QoE(Quality of Experience,QoE)of intelligent transportation applications.Therefore,the blind assumption that the nodes are completely cooperative will greatly reduce the reliability of multi-vehicle cooperative communication,which will cause heavy losses to the applications of vehicular networks,especially safety applications.To the end of this,this thesis is devoted to investigating the selfishness and incentive mechanism of nodes in vehicular communication networks.Comparing to other wireless mobile networks,vehicular networks have the characteristics of high mobility,dynamic network topology and weak wireless links.Due to these characteristics,the vehicular network is extremely complex,which brings great challenges to the research of node selfishness.This thesis focuses on the following topics:how harmful are selfish nodes?how to effectively detect selfish nodes?how to incentivize selfish nodes to improve the network performance?Specifically speaking,the main work of this thesis is summarized as follows.(1)The impact of node selfishness on the communication performance of MANETs(Mobile Ad-hoc Networks)is quantitatively analyzed.Mobile Ad-hoc Network(MANET)has a wider application range.Most of communication technologies in vehicular networks are inherited from MANET.Therefore,the influence of node selfishness on MANET network performance is quantified in this thesis.We define the dynamic selfish node based on energy consumption by taking the residual energy of mobile node as the basis to trigger selfish behavior.The influence of selfish nodes on network performance is quantitatively discussed and analyzed from the point of view of node mobility model,density,proportion and various combinations.Furthermore,the impact of node selfishness on the communication performance of VANETs(Vehicular Ad-hoc Networks)is quantitatively analyzed.Comparing to MANETs,the mobility model of nodes is more organized and the speed of nodes is faster in VANETs.Therefore,it is worthwhile to quantitatively analyze the impact of selfish nodes on network performance in VANETs.According to the nature of selfishness,static and dynamic selfish nodes are defined,and their impacts on network performance are evaluated in the network.The node mobility models are classified into single-route and multi-route separate patterns for more comprehensive analysis.Through extensive computer simulations,the impacts of static or dynamic selfish nodes are compared with single-route and multi-route mobility models.The experimental results can be used as the basis to design detection,incentive of selfish nodes and selfishness-aware routing algorithm in VANETs.(2)A fuzzy logic based selfish node detection method is proposed in vehicular communication networks.Aiming at the problem that it is difficult to identify selfish nodes in the environment which is full of uncertainties caused by the fast moving speed of vehicles,frequent changes of network topology and intermittent communication links,the detection method carefully concludes the selfishness of nodes based on the joint consideration of the mobility,link quality,MAC and network layer behavior of nodes with the help of fuzzy logic theory and avoids blind judgments.Furthermore,a perception mechanism is introduced in the detection scheme.The mechanism percepts the moving direction of vehicles and the probability of ACK reception to assist the final selfishness decision.Fair and effective selfish node detection is the basis of accurate incentive and the guarantee to improve the incentive efficiency.(3)A reinforcement learning based incentive mechanism for selfish nodes in vehicular communication networks is proposed.In order to deal with the inefficiency issues of conventional incentive mechanisms caused by the various uncertainties in vehicular communication networks,the proposed mechanism exploits fuzzy logic estimation for the internal and external influential factors of nodes,and further stimulates selfish nodes with the integration of game theory rewarding strategy and Q-learning incentive mechanism.The mechanism improves incentive efficiency by avoiding wrong rewards and wrong punishments,and further increases the communication performance of the network.We conducted extensive computer simulations to evaluate the efficiency and effectiveness of proposed methods and algorithms.The simulation platform we used for the experiments is the integration of OMNeT++,INET Framework,SUMO,and Veins.The simulation results indicate that the mobility model and proportion of energy consumption based selfish nodes have more impacts on the communication performance in MANETs in terms of average packet loss rate and average throughput;In MANETs,the route discovery process of the routing protocol spends more time with increasing number of selfish nodes and diversity of network topology,and further lengthens the average round-trip delay of the network;static selfish nodes have more negative impacts on the network performance than dynamic ones in both MANETs and VANETs;Comparing with multi-route mobility model,the selfish nodes with single-route mobility model have more significant impacts on communication performance.Furthermore,the experimental results verified that the proposed selfish node detection method is fairer(low false positive)and more accurate(high true positive)than the other baseline approaches.For instance,the proposed detection method improves the precision by 40%,when selfishness is 80%in highway scenarios.Additionally,the results show that the proposed incentive mechanism is able to effectively stimulate selfish nodes in vehicular networks with acceptable communication overheads by avoiding false punishments and unfair rewards.