Research On Incentive Mechanism In Crowdsensing Based On Incomplete Auction

Crowdsensing is an emerging Internet sensing model that uses mobile intelligent devices carried by ordinary users to collect data for completing large-scale and complex sensing tasks.Compared with traditional sensing methods,Crowdsensing has many advantages,such as wide sensing range and low sensing cost,so it has vast potential for application and vital research significance.With the rapid development and expansion of coverage of mobile intelligent devices,crowdsensing applications are becoming diversified,such as weather monitoring,indoor positioning,traffic information perception and public bicycle scheduling,etc.However,participants complete the sensing tasks at the cost of resource consumption and even face the risk of potential privacy leakage.Few users are willing to participate for free,so a reasonable incentive mechanism is necessary to motivate users to participate in the sensing system.Simultaneously,due to the uncertainty of the Internet environment and the variability of the scale of crowdsensing systems,how to motivate users while ensuring the efficient and stable operation of the system has become a current research focus.This thesis mainly focuses on the incentive mechanism in crowdsensing based on auction.With the goal of the crowdsensing system stability and the fairness of users,an incentive mechanism based on incomplete auction is designed considering the individual behavior of users and the scale characteristics of the system platform.Accordingly,when facing complicated and changeable sensing scenarios,a stable incentive effect can still be obtained,further ensuring the robust and sustainable development of the crowdsensing system.The main work of this thesis is as follows.1)Aiming at the problem that low-quality bids in the system lead to the decline of system utility,an incentive mechanism model based on incomplete auction is proposed.The model controls the quality of winning bids by adjusting the reserve price of tasks to reduce the impact of low-quality bids on system utility.A winner determination algorithm and a payment calculation algorithm adapted to the incomplete auction model are designed by combining greedy algorithm and VCG mechanisms.Finally,the rationality of the incomplete auction model is verified by simulation experiments in terms of task allocation rate and system utility.The model and theoretical foundation are laid for the subsequent work.2)Targeting the false-name attack and monopoly behavior in the process of crowdsensing incentive,a malicious robust incentive mechanism(MR-IM)is designed based on the incomplete auction model.The mechanism imports the concept of user scarcity to distinguish the quality of user bids to solve the pure monopoly problem in a fine-grained manner.It sets a budget-sensitive reserve price and payment cap to prevent the occurrence of oligopoly while ensuring individual rationality.Then it adopts a critical price payment mechanism to ensure the robustness against false-name attack.Finally,it is verified that the proposed MR-IM can resist the above malicious behaviors through theoretical derivation and experimental simulation,thus improving the robustness of the crowdsensing system.3)For the trade-off among user fairness,task allocation rate and system cost in crowdsensing reality scenarios and diverse choices of allocations,an incomplete auction model based on multi-objective optimization is constructed.A multi-round linear weight optimization algorithm(MLWO)is proposed for the multi-objective optimization problem,and a multi-level ordering greedy strategy with fairness priority is designed to solve the problem of winner determination and task allocation.In the meantime,the ENSGA-Ⅱ algorithm is designed according to the characteristics of the crowdsensing incentive model,and the result of the MLWO algorithm is brought into the ENSGA-Ⅱ algorithm as the initial solution to achieve adequate Pareto-optimal allocation.Finally,the effectiveness and high efficiency of the proposed model and method are verified by experimental simulations.