Multi-node cooperated sensor network missing value reconstruction method

Abstract

The present invention relates to a multi-node cooperated sensor network missing value reconstruction method. The method includes the steps of acquiring an opinion of a neighbor node A for a missing value of a target node O according to a spatial correlation of the target node O and the neighbor node A, wherein a judgment for the missing value of the target node O is x''(On+1), and the opinion is {b,u,a}; similarly, calculating judgments and corresponding opinions of other neighbor nodes of the target node O for the missing value; acquiring a final opinion and a corresponding judgment set by comprehensively considering the judgments and corresponding opinions of multiple neighbor nodes of the target node O; and taking expectation ei as an occurrence probability of judgment xi, performing weight-based combination on the expectation ei, and restoring the missing value x''(On+1) of the target node O. According to the method, the sensor node is fully used to sense time correlation characteristics and space correlation characteristics of data, the opinion of each neighbor node for the missing value of the target node is quantified objectively, and opinions of multiple neighbor nodes are converged accurately, therefore, errors of the restored missing value are reduced, a user does not interfere much, the robustness is higher, and the restoration accuracy rate is higher.

Description

technical field [0001] The invention relates to the technical field of wireless sensor networks, in particular to a multi-node cooperative sensor network missing value reconstruction method. Background technique [0002] The wireless sensor network is composed of a large number of cheap micro-sensor nodes deployed in the monitoring area, and forms a multi-hop self-organizing network system through wireless communication. Object information, and sent to observers. The nodes within the communication range of sensor nodes are 1-hop neighbor nodes, and the neighbor nodes are 1-hop neighbor nodes by default. The most common way to measure neighbor nodes is whether the physical distance between nodes is within the communication range. Due to noise, collisions, and unreliable connections, wireless sensor networks often lose some of the data sensed by sensor nodes. [0003] In order to solve this problem, people often use the characteristics of temporal correlation and spatial cor...

AI Technical Summary

Problems solved by technology

This is mainly reflected in two aspects: first, the author tries to use a simple model to directly describe the temporal correlation and spatial correlation, which leads to obvious defects in his method, for example: the autoregressive model does not distinguish between the neighbor node and the current node. The correlation of the correlation, increasing the consideration of this aspect will greatly increase the amount of calculation, and if this problem is not solved, it will lead to an increase in the error of missing value restoration; secondly, the author tries to use the method of weighted averaging to combine the missing values ​​generated by two different methods Fusion, this method relies heavily on the setting of weights, and the setting of weights depends on the user's subjective expectations of missing values, and it is easy to introduce error values, which increases the error of the restored missing values

In general, when the existing work comprehensively examines the methods of multiple missing values, there is no reasonable data support, which may easily lead to large errors between the restored missing values ​​and the real values.

Images