Design And Implementation Of MAC Layer In Passive Wireless Sensor Networks Based On Data Aggregation

Applying sensor networks in industrial sites to realize real-time monitoring of industrial equipment,ensure safe operation of equipment,and realize industrial intelligence and automation is an inevitable trend in the development of Internet of Things technology.Relying on the project of "Research and Application of Key Technologies for Passive Wireless Sensor Networks" in Beijing,this paper analyzes the vibration data of locomotive axles and transmits data back via industrial sensor network to realize the monitoring and early warning of the running status of vehicles in the scene of rail transit.The wireless sensor network is deployed on the rail locomotive,and the mechanical energy of the locomotive axle vibration is converted into electrical energy through the Energy Harvesting module to supply energy for the sensor nodes.Due to the slow rate of power collection and unstable power supply,the high-speed data stream collected by acceleration sensors cannot be effectively transmitted in time in low-speed and low-power passive wireless sensor networks.When the scale of network is large,data transmission through multi-hop networks will lead to a strong funnel effect,causing the aggregation node to be overloaded,the energy consumption is intensified,and even the network is paralyzed,which seriously affects the stability of the network and the validity of the data.In the case of limited energy in the passive wireless sensor network,the amount of data collection is large and the transmission is difficult,the node resources are limited and the complex calculation and mass storage cannot be performed,and the funnel effect is obvious and the packet loss rate is serious under the multi-hop network.This paper proposes a fault diagnosis method based on machine learning for data compression to reduce the amount of data transmission,and proposes a MAC layer time slot allocation scheme based on data aggregation for wireless sensor networks to further reduce network traffic and reduce network energy consumption.For the high-speed vibration data stream,the vibration data is extracted and the machine learning model is used to diagnose the equipment condition by the limited storage resources and computing power of sensor nodes at the node end.When a node transmits data,only the diagnosis result is transmitted,and the original data is only transmitted back when necessary.The solution achieves fault diagnosis and reduces the data transmission capacity of more than 95%of the nodes,which reduces the energy consumption of the nodes and prolongs the survival time of the nodes.To solve the problem of data transmission in wireless sensor networks with limited energy,a TDMA-based MAC layer data slot allocation scheme is proposed,which designs superframes and allocates time slots for the network.The sensor nodes are clustered according to the physical deployment location,and different channels are occupied by different clusters to perform intra-cluster data interaction,thereby avoiding collision conflicts between the clusters.The cluster head is selected through the cluster head election algorithm,and the cluster head nodes aggregate and transmit the diagnostic states transmitted by the nodes in the cluster,thus further reducing the amount of data transmitted by the nodes.Through the time slot occupation algorithm,cluster head nodes allocate shared time slots for nodes that need additional transmission time slots,thus ensuring the flexibility of network time slot allocation.At the end of the paper,the function of passive wireless sensor network based on data aggregation is tested,and the function of sensor network is verified through packet grabbing analysis of network nodes.Experiments show that the design in this paper not only guarantees the diagnosis accuracy,but also reduces the amount of network data,reduces the transmission energy consumption and prolongs the service life of nodes.