Efficient Auction Schemes For Resource Allocation In Mobile Cloud Computing

In the widely distributed wireless access networks,the applications of mobile device have been developed to satisfy the requirement of mobile users(MUs).However,variety application cannot be processed well on the mobile devices due to the shortage of resources.Mobile cloud computing is introduced to alleviate the resource scarcity of mobile users.They can offload their computation-intensive or real-time works to the cloudlet nearby to improve the performance of their mobile devices.Cloudlet deployment and resource allocation for mobile users have been extensively studied in existing works for computation resource scarcity.Nevertheless,it is failure to jointly consider the two techniques.Between selfishness of cloudlet and access point(AP)are ignored.Moreover,most of related works which have employed the auction scheme only care about that one MU deal with one cloudlet,which leads to the low efficiency and utilization of resource in cloudlet.Inspired by the group-buying mechanism,a kind of three-stage auction schemes is proposed for the cloudlet placement and resource assignment.The purpose is to improve the efficiency and stability of auction schemes under according with the social welfare subject and the economic properties.Firstly,all MUs are divided into s ome small groups according to the associated APs.Then,the MUs in same group can trade with cloudlets in a group-buying way through the APs.Finally,the MUs pay for the cloudlets if they are the winners in the auction scheme.In the first auction scheme,the number m based on the capacity of each given cloudlet is generated randomly.The budget for the given cloudlet is calculated using the selection of the first m MUs based the performance price ratio.After the first scheme,the second scheme is proposed to calculate several profitable cases,and then randomly selected one from them,which can improve the revenue of the small group of MUs significantly.In the third scheme,a global way based on the second scheme is improved based on above two proposed scheme,which is utilized to match cloudlets for Aps.Through the theoretical analysis and demonstration,the proposed auction schemes can work in polynomial time.To evaluate the performance,the existing schemes are compared with our schemes in some simulation results.The results demonstrate that the proposed schemes achieve the better performance of the three kinds of entities inculding MU,AP and cloudlet.Moreover,For instance,when the number of MUs is 1000,our schemes achieve social 4.53% TACDp and 5.62%TACDpp,which is higher than the existing schemes.