Localization method for wireless sensor network based on compressed sensing and bp neural network

Abstract

The invention discloses a wireless sensor network positioning method based on compressed sensing and BP neural networks. The method applies compressed sensing and BP neural networks to grid-based multi-target positioning and uses the received signal strength indicator (RSSI) as a positioning parameter. The method comprises the following steps: step one, using compressed sensing to position a wireless sensor network node, and determining whether the node is at the center of a grid; and step two, for an unknown node not at the center of the grid, using BP neural networks and maximum likelihood estimation successively to calculate coordinates of the node. The BP neural networks are used to correct the error of an RSSI measurement value, and finally the maximum likelihood estimation is used to calculate the real coordinates of the unknown node. The method can position multiple targets at the same time, overcomes the defect that a conventional compressed sensing method can only do positioning at the center of the grid, uses the BP neural network to overcome the defect of poor precision of RSSI positioning, and improves the positioning precision while reducing the power consumption.

Description

technical field [0001] The invention relates to a wireless sensor network positioning method based on compressed sensing and BP neural network, which belongs to the technical field of mobile communication. Background technique [0002] Node positioning is one of the most important technologies in wireless sensor networks, because it plays a vital role in many application fields, such as target tracking, military reconnaissance, geographic environment monitoring, traffic road condition monitoring, medical and health and other fields. If the user cannot know the accurate location information of the node in advance, it will not be possible to complete the relevant application. The above location information needs to be obtained through the node location information of the wireless sensor network itself. The research on wireless sensor node positioning includes many aspects, such as anchor node density, network robustness and fault tolerance, positioning accuracy affected by th...

AI Technical Summary

Problems solved by technology

Cevher V et al. transformed the problem of target positioning into compressed sensing, but it can only be positioned in the center of the grid, and each node needs a positioning dictionary

He Fengxing et al. used the residual best matching algorithm to optimize the compressed sensing positioning method, but the positioning error is still relatively large in the case of large noise

Malioutov and others will use sparse transformation for node positioning research, but the disadvantage is that the algorithm is too complicated

Images