Research On Resource Optimal Allocation Double Auction Algorithm Based On Deep Learning

In recent years,blockchain has been deeply integrated with cutting-edge information technologies such as artificial intelligence,big data,and the Internet of Things(Io T).Among them,research on blockchain-based Internet of Things,such as privacy protection,energy and data transactions,consensus mechanism under the Internet of Things,is in full swing.However,blockchain-based Io T faces some serious challenges due to the large amount of computing resources required to solve proof-of-work problems in consensus mechanisms,but the limited computing resources of Io T devices limit the opportunities for more complex research applications.To solve this problem,edge cloud computing can be introduced to Io T,which offloads the computing tasks of Io T devices to computing resource service providers.In this context,how to efficiently and reasonably allocate computing resources from suppliers to Io T devices has become an important topic of common concern for researchers at home and abroad,and auction has become the mainstream solution for many researchers to solve the problem of resource allocation.Firstly,this thesis expounds the research background and significance of auction algorithm,and then analyzes the resource optimal allocation in detail from the three aspects of single auction algorithm,pairing double auction algorithm and iterative double auction algorithm,points out the existing problems of the existing double auction algorithm,and finally designs the corresponding solutions and experimental verification.The main contributions of this thesis are as follows:To solve the problems of imperfect incentive compatibility mechanism and malicious bidding in the process of pairing auction,this thesis proposes a pairing double auction algorithm based on optimal pairing model.The algorithm limits the number of sellers matching buyers by introducing a second elimination factor in the cyclic pairing stage between buyers and sellers,which embodies the incentive compatibility among buyers and improves the incentive compatibility mechanism of the algorithm.Furthermore,the optimal value of is obtained by the optimal pairing model based on fully connected neural network to improve the performance of the algorithm on economic benefit index.In addition,the introduction of?₂ can effectively avoid the monopolistic behavior of individual sellers and enhance sellers'participation in pairing transactions.At the same time,because the trading rules of traditional pairing double auction algorithm have the risk of malicious bidding between buyers and sellers,the algorithm proposed in this paper can effectively punish the sellers or buyers of malicious bidding by optimizing the trading rules and making the trading price directly related to the bidding between buyers and sellers.Multi-level experiments under different market scales also verify the effectiveness of the algorithm in economic benefits,seller participation index superiority.To solve the problems of low computational efficiency and unreasonable utilities distribution in the process of iterative auction,this thesis proposes an iterative double auction algorithm based on resource optimal allocation model.The initial bidding data of both buyers and sellers are used to train resources optimal allocation model based on CNN in the algorithm,then invoking the trained model quick responds to the real-time bidding data directly solving the broker optimal allocation problem to reach computing resource optimal,which significantly reduces the calculation cost,consumes less time,and improves the efficiency of the algorithm.Further,in view of problems such as unreasonable distribution of utilities,adjustment factors are introduced into the spending rule and earning rule of the iterative double auction framework to adjust the utilities of the buyers and sellers,which solves the unreasonable utilities distribution in the process of maximizing social welfare under the existing algorithm.The experimental results show that the proposed algorithm is superior to the existing iterative double auction algorithm in terms of running time,social welfare,the utilities of buyers and sellers,the utilities of broker and other indicators.