Energy-efficient Sensory Data Collection Based On Spatio-temporal Correlation In IoT Networks

The Internet of things is developing rapidly.Hundreds of millions of sensing nodes and intelligent terminals undertake the task of sensing and transmitting data.Data collection is the key to realize data analysis and intelligent application of Internet of things.However,a large number of real-time data collection is also a big challenge.The life cycle of the Internet of things is limited by the energy of the Internet of things nodes in the network.Most of the existing Internet of things devices only have limited battery life,and it is not convenient to replace the power supply.Therefore,how to achieve energy-saving data collection and reduce network energy consumption is the key and difficult point of Io T research.The data collected by the Internet of things has a wide range of dynamics,in which there are temporal and spatial trends.The effective use of the spatial-temporal relationship between data can achieve the purpose of reducing data redundancy and saving energy.Due to the limited resources of Internet of things nodes,complex computing model will bring serious or even unbearable burden to Io T nodes.It is necessary to use lightweight time correlation prediction method and spatial correlation judgment method.And in order to adapt to the dynamic changes of data,the network model and prediction model need to be constantly adjusted.In order to adjust the appropriate spatial sampling rate,the spatial correlation of sensory data is explored,and the temporal correlation of data based on data prediction method is explored for further data reduction.The main contents include the following three aspects:Firstly,based on the temporal correlation of sensing data,the sensing data of sensor nodes is predicted by using the data prediction method.In this thesis,a simple and effective data prediction model DBP is used for data prediction,and the model is further optimized and improved.In order to reduce the amount of sensor nodes sensing and transmitting data,a jump sensing strategy is adopted.When the prediction function can effectively predict the sensor data,increasing the time interval of the sensing data can further reduce the energy consumption.Secondly,based on the spatial characteristics of sensing data,this thesis takes the slope of DBP function as the standard to evaluate the data similarity between nodes,and reduces the energy consumption by reducing the sampling rate through the mutual substitution of similar node data.In addition,an optimal greedy algorithm based on node priority is proposed to select the appropriate dominating node to improve the utilization of network energy.Finally,in order to adapt to the dynamic of data,this thesis uses probabilistic wake-up strategy to determine whether the data similarity between nodes has changed.When the correlation of sensing data between nodes is detected to change,the spatial correlation of sensing data is adjusted to improve the reliability of data.In order to verify the spatiotemporal correlation data collection method(TSD)proposed in this thesis,a simulation experiment is carried out on the real perception data provided by Berkeley research laboratory.The experimental results show that the proposed TSD algorithm has good performance and achieves the goal of reducing network energy consumption on the premise of ensuring data reliability.