UAV Path Planning Based On Deep Reinforcement Learning In The Internet Of Things

The Internet of things(IoT)has developed rapidly in recent years,and has been widely used in industrial/agricultural Internet,smart earth,virtual reality/augmented reality,and other fields.The development of wireless communication technology and artificial intelligence and their application in the IoT have promoted the IoT toward the direction of intellectualization and diversification,thus expanding the scope of application of the IoT and improving the application effect.Widespread application of IoT makes wireless terminals such as sensors and intelligent devices ubiquitous.How to effectively collect data in these IoT terminals for further analysis and utilization has always been a concern of researchers.In most cases,the batteries of sensor nodes in the IoT are usually constrained by the geographical environment or the sensor itself,it is usually difficult to recharge.Data is transmitted in a multi-hop manner in traditional wireless sensor networks,which may cause the high data transmission delay and unbalanced traffic load.The communication devices in the IoT are still suffering from poor wireless connections and low data rate in some remote areas,resulting in inefficient data delivery from the sensors to the data processing/storage servers.To cope with above problems,an Unmanned Aerial Vehicle(UAV)-assisted IoT architecture is introduced in this thesis,in which UAV is utilized to achieve low-latency and seamless-coverage acquisition of the sensing data.Then,based on the deep reinforcement learning algorithm,the UAV's moving path in the process of data acquisition is reasonably planned to improve the efficiency of data collection.Finally,the blockchain technology is introduced into the UAV-assisted IoT to solve the problems of node trust and data security,and an appropriate consensus mechanism is proposed.Based on deep reinforcement learning,the path of UAV is planned to improve the probability of gaining the accounting rights so as to obtain the reward income under the proposed consensus mechanism.This thesis focuses on the architecture of the UAV-assisted IoT and the path planning of UAV in the process of data collection.First,the architecture of UAV-assisted IoT and the path planning method of UAV are proposed.Then,the blockchain technology is utilized to solve the trust of nodes and data security problems in the UAV-assisted IoT architecture proposed in the first part.And the main research works are described as follow:(1)UAV-assisted IoT and path planning of the UAVAiming at some remote areas where IoT data cannot be returned in time and effectively,as well as the problems of transmission delay and load imbalance in the process of data transmission,an architecture of UAV-assisted IoT is introduced,UAV is utilized to achieve low-latency and seamless-coverage acquisition of the sensing data.Then considering both data delay requirements and network energy consumption,a real-time flight path planning scheme of UAV in the dynamic sensor networks has been proposed based on deep reinforcement learning,to reduce power consumption while meeting the delay constraints for different data types,and improve data collection efficiency of UAV.(2)UAV-Assisted IoT Based on Blockchain and Proof of Verified DataAiming at the trust of nodes and data security problems in the UAV-assisted IoT architecture,blockchain technology is introduced into the system,formed UAV-assisted IoT based on blockchain(UIB)system.Crypto-currency is adopted in the incentive scheme to encourage UAVs to collect more sensing data,and the proof of verified data(PoVD)consensus scheme is proposed to ensure the authenticity of sensing data and to achieve consensus on the creator of the sensing data blocks.In addition,from the point of view of UAV operators,a UAV path planning scheme based on deep reinforcement learning is proposed in the UIB system to maximize the encrypted currency reward for UAV.