Research On Opportunity Transfer Mechanism For Internet Of Things

Driven by new information technologies such as edge computing,big data and artificial intelligence,the Internet of Things has been widely used in areas such as emergency monitoring,environmental monitoring,intelligent transportation,smart agriculture and geological disaster warning.In the Internet of things there are a large number of fixed and mobile data acquisition node(such as smart phones,sports bracelet portable intelligent equipment and vehicles,etc.)at the same time,in some applications of mobile nodes and data acquisition nodes have the chance to meet a lot of,part of the movement of the mobile node path has a certain regularity,data flow to also have a certain direction.The mobile node uses the chance of meeting with the fixed data acquisition node to help it transmit data,which is called opportunistic transmission.Making full use of opportunistic transmission can alleviate the problem of data overflow and missing caused by delayed data transmission due to insufficient capacity of data acquisition nodes.Aiming at maximizing the total amount of opportunity-transmitted data,the thesis designs opportunity transmission mechanisms for centralized decision-making and autonomous intelligent decisionmaking for different lot application scenarios,which are described as follows:The opportunity transmission mechanism of centralized decision making is studied in view of the application scenarios in which a large number of data acquisition requirements are temporarily generated in urban emergencies,the storage and transmission capacity of data acquisition sensing devices is limited,and mobile devices with data carrying capacity,such as smart phones and vehicles,have the chance to meet data collection points.The central control server makes centralized and unified decisions based on the state information of each mobile node and data acquisition node.Aiming at maximizing the total amount of opportunistic data transfer,the optimization model is established and the data carrying strategy of mobile opportunistic transfer node is obtained by solving the optimization model based on genetic algorithm.Simulation results show that the opportunistic transmission strategy based on genetic algorithm can get the maximum amount of opportunistic transmission data and is superior to greedy algorithm.For such as environment monitoring application scenarios in remote areas,monitoring the data only through data acquisition nodes by wireless transmission to the control center,data acquisition node's ability is limited,and periodic cruising tasks of unmanned aerial vehicle and monitoring the data acquisition nodes have chance encounter,for this kind of scenario study independent intelligent decision-making opportunity transport mechanism.During flight,UAV makes autonomous decisions according to its own state,and states of data acquisition node and other uav.In the thesis,the decision-making process of multiple UAVs participating in opportunistic transmission is modeled as the Markov decision process of multiple agents.Each UAV adopts the improved S-MADDPG algorithm designed in the thesis to learn the optimal strategy for maximizing the total amount of opportunistic transmission data.Simulation results show that S-MADDPG algorithm can converge to get the maximum amount of opportunistic data and is superior to MADDPG algorithm.