Research On Mobile Cloud Service Offloading Mechanism In Vehicular Networks

The rapid development of the information and communication industry has promot-ed the overall structure of society and the way people live,which has also brought the new opportunities to the transportation industry.The intelligent transportation system(ITS)aiming at building the"people-vehicle-road-cloud"synergy is proposed at this time.Mo-bile cloud service offloading is an effective way to extend the computation capability and storage capacity of the vehicle terminals,which is the core technology to ensure the ef-ficient and stable execution of the intelligent transportation applications in the future.In order to guarantee the effectiveness of the mobile cloud service offloading and improve the accident avoidance rate,road smoothness,and intelligent opration level of highway transportation system,it is of great significance to study the mobile cloud service offload-ing mechanism for vehicular networks.In terms of the practical and detailed vehicular network scenario,this dissertation is initialized with a clear interpreting for the impact of random mobility on wireless com-munication and mobile cloud service offloading,and systematically construct the system design of mobile cloud service offloading mechanism in dynamic environment.The in-novation of the dissertation mainly includes:1)The dissertation proposes a vehicle mobility-aware content offloading mechanism for the infrastructure-less vehicular network to meet both high offloading fairness and the huge content throughput.Based on the comprehensi've consideration of various influencing factors such as vehicle density,mobility,and wireless channel model,a method for predicting the wireless link duration between user vehicles and mobile cloud servers is proposed to quantify the impact of content throughput and multi-user offloading fairness.The simulation results verify that the proposed content offloading mechanism improves the offloading throughput while guaranteeing the fairness of multi-user offloading.2)The dissertation proposes a vehicle relaying-based computation offloading mechanis-m to avoid the offloading performance degradation caused by the short link duration between the user vehicle and the single mobile cloud server.Firstly,with the consid-eration of vehicle mobility and task execution logic sequence,a mobile cloud server classification algorithm based on link duration characteristics is proposed.Then,the computation relaying conditions of multiple mobile cloud servers under the dynamic network topology are constructed to achieve the cooperative computation offloading in vehicular networks.Finally,two heuristic computation offloading mechanisms are proposed for the single mobile cloud server scenario and the multiple mobile cloud server scenario,respectively.The simulation results show that the proposed compu-tational offloading mechanism effectively reduces the user's vehicle energy consump-tion and application completion time.3)The dissertation proposes a computation offloading mechanism under the condition of diverse task logic structures and the dynamic random vehicle mobility model to make the offloading decision for each task.The Markov decision process(MDP)is used to describe the random variation process of vehicle motion state.And the linear taskgraph offloading problem is formulated as the MDP problem with the objective of minimizing the average completion time of the application.The complexity for solv-ing the MDP problem directly,however,is prohibitively high due to the large state space of the MDP,which leads to a large dimension of the state transition probability matrix.Based on local information,a heuristic solution is then proposed.By consid-ering all remaining tasks to be executed at the same cloudlet,the offloading decision can be obtained by solving the MDP problem with much lower complexity;and the offloading decision is updated immediately based on the current system state when the execution of one task is completed.Then,two heuristic computation offloading mechanisms are proposed to offload the concurrent taskgraph and the tree taskgraph,respectively.Theoretical derivation and simulation results show that the proposed computation offloading mechanism can improve system offloading performance with the low computational complexity.4)Considering the random network topology change and the variable computation re-sources in vehicular networks,it is difficult to achieve the real-time complete net-work topology information,which brings great challenges to the design of the com-putation offloading mechanism.The dissertation proposes an incomplete network information-based computation offloading mechanism to meet the various task de-lay requirements.To describe the variability feature of the available resources and the application-aware delay requirements,a mathematical model for mobile cloud computing resource allocation and offloading decision is established based on semi-Markov decision process(SMDP).Moreover,considering the offloading immediate reward and cost,the computation offloading mechanism for the multi-user and multi-server is obtained.Simulation results show that the proposed offloading mechanism can achieve the maximum expected long-term reward of the system compared to the conventional baseline policies.