LAUSR

There is a prominent contradiction between the existing resolution requirements of marine monitoring data (MMD) and the cost of sensor network deployment. This article proposes an MMD acquisition and reconstruction scheme based on compressed sensing theory (CS-MMD). Firstly, the sparse characteristics of the measured MMD are analyzed and modeled as a multi-measurement vector (MMV) compressed sensing reconstruction problem. Furthermore, the operating state of the sensor is adjusted by a random sparse polynomial distribution matrix. The sensors corresponding to the non-zero elements in the matrix work with a probability of $p$ (small value), and the rest of the sensors sleep. In the reconstruction, the energy prior of the data is fully used to obtain the support set, and the MMV is randomly reduced to the SVM to simplify the support set reconstruction process. The theoretical analysis gives the conditions for accurate reconstruction. The simulation results show that CS-MMD can save a lot of acquisition resources and accurately reconstruct data, and the accuracy rate reaches 99% under the premise of saving up to 99% of sampling resources.