Research On IoT Resource Optimization And Scheduling Based On Blockchain

With the continuous development of IoT applications and services,mobile terminal devices have become a very important part of the interconnected world.The connection of these devices will generate a large amount of data or tasks that need to be stored,transmitted,or calculated.However,how to handle these tasks quickly and safely faces many challenges:First,IoT devices have resource constraints in terms of processing power,battery power,and storage space,making it difficult to independently perform complex and heavy computing tasks on these devices;Secondly,it is necessary to ensure the security and authenticity of the collected data.Blockchain technology can reliably store data,but the direct transmission of raw data still makes it difficult to ensure user privacy;Thirdly,many applications have a need for trusted storage and monitoring of data,and it is necessary to design a system model,platform,and deployment mode for blockchain credit data management.Based on the above requirements,this paper takes MEC computing task unloading as the basic scenario,integrates multiple technologies such as blockchain,edge computing,DRL,federated learning,etc.,designs schemes and develops related systems,and successfully utilize resources such as computing,communication,and caching to provide low latency offload decisions,resource allocation policies,privacy protection policies,and resource monitoring services.The main work of this article is as follows:1.An IoT task offloading optimization scheme based on DQN is proposed,integrating MEC,controller,and blockchain into the system.Due to the high dimensional complexity of the system,by fully considering the adjustable factors of the system,the consensus and unloading process is designed as MDP,and DQN algorithm is introduced to dynamically select unloading actions(including verification node selection,block size selection,task unloading server selection,and block interval selection).To accelerate optimization,this paper designs a new reward function based on the size and computational complexity of the task.In order to reduce the impact of blockchain on overall performance,the system considers the consensus delay and throughput of blockchain,ensuring data security and integrity while achieving the optimal overall performance of the system.Simulation experiments show that the scheme designed in this paper performs better in terms of task offload quantity and blockchain throughput compared to the scheme without optimization.2.In order to transfer data value while meeting legal compliance requirements,a resource scheduling scheme for IoT privacy security is proposed,which can meet data privacy requirements and avoid direct sharing of data.Add whether to accept the transmission of raw data to the status list of each calculation task.The federated learning system is added to the system.The device data in each unit is trained against the sub model on its corresponding MEC server,and the sub model parameters are aggregated on the cloud server to generate a global model.In order to prevent single point crashes of central servers,a decentralized federated learning framework assisted by blockchain is proposed to provide a traceable and reliable task offloading and model training environment.In addition,optimization of blockchain is also considered.Simulation experiments show that compared with other schemes,the scheme designed in this paper achieves higher reward values in almost all regions,and has faster convergence speed.3.Based on the need for reliable monitoring and management of IoT data in practical applications,a data credit resource monitoring system for blockchain applications is designed and developed to systematically manage monitoring node resources and store credit files.Based on domestic blockchain,functions such as online certificate storage,query and supervision are implemented.The user interface can monitor the system in real time,such as viewing system parameters such as the number of nodes,message processing capacity,transaction volume,and file count.It can observe service performance,distributed storage performance,CPU usage of various nodes,and chain monitoring.