Research On Rate Adaptation Algorithm For Backscatter Network Based On Effective Node

The Internet of Things(IoT)technology aims to realize the true interconnection of everything and provide important support for information infrastructure such as smart city,smart agriculture and smart medical care.IoT technology is a key research direction of the new generation of digital technology innovation breakthroughs.Backscatter network,as a key technology to connect a large number of IoT sensing devices,has attracted wide attention worldwide because of its advantages of low cost,low power consumption,easy networking,easy maintenance and so on.The backscatter node can not only absorb energy from the electromagnetic wave emitted by the RF source,but also modulate the sent information by adjusting the antenna "on" and "off" to absorb and reflect the electromagnetic wave.Backscatter technology provides more possibilities for the application of wireless sensing,motion tracking and other intelligent scenarios.With the continuous expansion of the scale of the Internet of Things industry,the amount of calculation and data of backscatter devices has increased significantly,and the transmission environment has become increasingly complex.Due to the intense competition between backscatter nodes for finite time slots and the unpredictability of channel conditions,the individual throughput of nodes is very limited,so the importance of a fast information collection method is obviously increased.Extracting a subset of tags can collect information from certain target tags instead of all tags.In this study,the target tags in a specific scene are called effective nodes.In order to improve the throughput of the effective nodes that carry important information,this paper proposes a rate adaptation method,RAEN,which is suitable for effective nodes.The research is carried out from two aspects:effective node extraction design and rate adaptation algorithm design.Firstly,this paper designs the extraction method of effective nodes.According to the phase distribution characteristics of effective nodes,an effective node recognition strategy based on the updatable Gaussian Model(GM)is designed.When selecting the identified effective nodes,the batch selection of effective nodes is realized through the method of bit vector in nodes,which improves the efficiency of Select command.Secondly,a rate adaptation algorithm for effective nodes is designed,including a new trigger and rate decision module.The trigger eliminates the design idea of continuous monitoring the changes of the external environment,and determines whether to predict the rate of the next stage based on the number of inventory cycles,which saves the detection cost.The rate decision module takes the Received Signal Strength Indicator(RSSI)and packet loss rate as indicators,designs a reasonable packet loss rate estimation method,and adopts the strategy of random forest combined with ensemble learning to select the overall network rate.By using this method,the optimal rate in the next stage can be predicted more fairly and the throughput of effective nodes can be significantly improved.Finally,this paper analyzes multiple sets of data collected by commercial readers and multiple RFID tags,and evaluates the design results combined with simulation experiments.The results show that the design can facilitate the effective nodes to collect information and improve the throughput of backscatter network.