Research On Resource Sharing Mechanism Based On Multi-user Collaboration In Mobile Cloud Computing Environment

In mobile cloud computing environment,mobile users can offload the computationally intensive tasks to the remote cloud computing data center through the wireless network and Internet,and complete the tasks by the resource-rich cloud computing data center.However,these cloud computing data center are typically far from the mobile devices,and the way of migrating computational tasks to the remote cloud can increase the network load and lead to the data transmission latency.To solve the problems of the traditional mobile cloud computing,the framework of mobile ad hoc cloud has been proposed,where the neighboring mobile devices are pooled together to meet the needs of users with resource limitation.The mobile ad hoc cloud is largely dependent on user's participation.Due to the rational and selfish nature of each user,users typically do not voluntarily share their local free resources.Thus,to implement the mobile ad hoc cloud,it needs to design an efficient mechanism to incentive the resource sharing.This dissertation proposes two different scenarios of the mobile ad hoc cloud:1)there is one supplying user;2)there are multiple supplying users.For two different scenarios,an effective mechanism is proposed to incentive the resource sharing,respectively.The main work of this dissertation includes:1.For the scenario where there is one supplying user and multiple demanding users,this dissertation studies how to design an effective pricing mechanism to coordinate the demand of multiple users for limited resources.In the scenario where supplying user does not know the utility functions of multiple demanding users,this dissertation proposes a prediction-based pricing scheme,and theoretically analyzes the difference of the social welfare achieved by the proposed prediction-based pricing scheme and the optimal social welfare and the impact of prediction error.2.For the scenario where there are multiple supplying users and multiple demanding users,this dissertation studies how to design an effective resource allocation mechanism to match multiple supplying users and demanding users.Combined with the social relationship among the users,this dissertation formulates the problem of resource sharing as a one-to-many matching game,and then proposes a fully distributed algorithm based on classical Gale-Shapley(GS)algorithm that all the users can benefit from the resource sharing and can converge to a stable matching in a finite number of iterations.3.This dissertation presents the detailed numerical simulation and analysis of the proposed algorithms.For the scenario of the mobile ad hoc cloud where there is one supplying user,numerical results validate that the prediction-based pricing scheme is efficient and can provide better performance with better prediction.For the scenario of the mobile ad hoc cloud where there are multiple supplying users,the fully distributed resource sharing algorithm based on GS is validated,and numerical results show that the social welfare achieved by the algorithm is close to the maximum social welfare.