Research On The Blockchain-based Decentralized Mobile Crowdsourcing Model

In recent years crowdsourcing as a new task allocation technology has enabled enterprises to reduce costs and improve efficiency.However,most crowdsourcing platforms currently follow a centralized management approach,i.e.,all information is stored in a central system database,which runs the risk of data leakage and is also vulnerable to various attacks such as DDo S.With the features of decentralization and traceability of blockchain,the crowdsourcing solution(BCS)built on blockchain can well solve the single point of failure problem and improve service availability.Therefore,this thesis proposes a blockchain-based mobile crowdsourcing model and focuses on the task and solution confidentiality and task recommendation aspects of the model.In this thesis,we first propose a model architecture consisting of users,smart contracts and blockchain networks,study and design a crowdsourcing mechanism on the blockchain,implement various crowdsourcing operations in traditional crowdsourcing through smart contracts,and ensure the confidentiality of crowdsourced data during the interaction by embedding the ciphertext of tasks and solutions and the interaction key into the contract.In addition,this thesis is dedicated to achieving efficient task assignment in the model,so a crowdsourcing task recommendation algorithm based on worker reputation is designed to construct a worker reputation influence model by calculating the change of task completion accuracy before and after eliminating workers,and construct a task feature model with the help of the concept of information entropy to generate crowdsourcing recommendation results by mining the similar relationship between worker performance and new task features in both models.Finally,this thesis considers the cost issues such as service time and equipment energy consumption,and builds a multi-objective optimization model based on the recommended results to maximize the algorithm efficacy.The experimental results show that the various operational overheads of the model in this thesis are user-acceptable and achieve a secure exchange of tasks and solutions.The average accuracy of the recommendation algorithm is around 80%,with varying degrees of excellence in other evaluation metrics,and the cost control model allows the algorithm to maintain a cost advantage of 10% to 29%.The research on crowdsourcing model on blockchain in this thesis improves developers’ participation in decentralized crowdsourcing,which is an important guideline for optimizing crowdsourcing services and helping innovation in the application of blockchain technology.