Search

CN-121980255-A - Rainfall and surface runoff effect decoupling method and device based on wireless channel state information space-time spectrum and storage medium

CN121980255ACN 121980255 ACN121980255 ACN 121980255ACN-121980255-A

Abstract

The invention discloses a rainfall and surface runoff effect decoupling method based on a wireless channel state information space-time map, and relates to the field of wireless environment sensing and flood monitoring. The method comprises the steps of obtaining multi-subcarrier Channel State Information (CSI) of a commercial wireless link, carrying out outlier rejection, amplitude calibration and phase purification on the CSI, constructing a CSI space-time spectrum comprising an amplitude channel and a phase channel according to time and subcarrier indexes within a preset time window, inputting the space-time spectrum into a double-decoder convolution self-encoder model, extracting mixed characteristics by a shared encoder, reconstructing a smooth background component by a rainfall decoder, reconstructing a sparse texture component by a runoff decoder, realizing the decoupling of rainfall attenuation and surface water multipath effect, realizing rainfall intensity estimation and surface water detection/depth estimation based on the decoupling component, and balancing the decoupled channel estimation to a receiving end so as to improve communication reliability. The method does not need prior environmental geometric or physical models, and can improve the environmental perception precision and reduce the communication error rate under the scene of the composite effect.

Inventors

  • TANG YIGUI
  • LONG YIN

Assignees

  • 西南科技大学

Dates

Publication Date
20260505
Application Date
20260302

Claims (12)

  1. 1. The rainfall and surface runoff effect decoupling method based on the wireless channel state information space-time spectrum is characterized by comprising the following steps of: Step one, obtaining a Channel State Information (CSI) data stream comprising a plurality of sub-carriers through a wireless transceiving link; Preprocessing the CSI data stream, wherein the preprocessing at least comprises outlier rejection, amplitude calibration and phase purification to obtain purified CSI; Thirdly, arranging the purified CSI according to time and subcarrier indexes in a preset time window, constructing a two-dimensional CSI space-time spectrum, and respectively taking the amplitude and the phase as different channels to form a multi-channel spectrum; Inputting the multi-channel spectrum into a double-decoder convolution self-encoder model, obtaining potential representation by a shared encoder, outputting a background spectrum I_rain representing rainfall effect by a rainfall decoder, and outputting a texture spectrum I_ runoff representing surface runoff effect by a runoff decoder; And fifthly, decoupling rainfall and surface runoff effect based on the I_rain and the I_ runoff, and outputting the decoupled channel estimation I_rain+I_ runoff or corresponding environmental parameters.
  2. 2. The decoupling method of claim 1, wherein the wireless transceiver link is a wireless link of Wi-Fi, cellular communication, or other OFDM regime, the CSI comprising at least amplitude information and/or phase information for a plurality of subcarriers.
  3. 3. The decoupling method of claim 1, wherein the outlier rejection employs a median-based Hampel filter or an equivalent robust statistical method to reject amplitude outliers caused by transient interference.
  4. 4. The decoupling method of claim 1, wherein the amplitude calibration comprises normalizing or scaling the amplitudes of all subcarriers within the same CSI frame to suppress synchronous step changes introduced by the receive automatic gain control AGC.
  5. 5. The decoupling method of claim 1, wherein the phase cleaning comprises linearly fitting the original phase of each subcarrier in each CSI frame, estimating and removing phase offset and linear terms introduced by carrier frequency offset CFO and sampling frequency offset SFO, and obtaining a relative phase sequence.
  6. 6. The decoupling method of claim 1, wherein the construction of the CSI spatiotemporal map comprises: Forming a matrix by continuously collecting the CSI frames in a time window T according to time sequence, wherein the vertical axis of the matrix corresponds to time and the horizontal axis of the matrix corresponds to an effective subcarrier index; the phase value is wrapped to (-pi, pi) and then mapped into a pixel value according to P_pixel=255× (phase+pi)/(2 pi); After converting the amplitude value to the decibel domain a_db=20log10|h|, mapping the amplitude value to A pixel value according to a_pixel=255× (a_db-a_min)/(a_max-a_min), wherein a_min and a_max are global minimum and global maximum adopted by the training set and the test set in A unified way.
  7. 7. The decoupling method of claim 1, wherein the dual decoder convolutional self-encoder model comprises a shared encoder, a rain decoder, and a runoff decoder; The shared encoder is composed of at least one set of convolution layer, activation layer and pooling layer, and is used for extracting the mixed characteristics of the multi-channel atlas and generating potential representation; The rainfall decoder is used for reconstructing the low-frequency smooth background component and does not adopt jump connection from the encoder; the radial flow decoder is used for reconstructing high-frequency sparse texture components and comprises jump connection with a corresponding layer of an encoder to enhance detail reconstruction capability.
  8. 8. The decoupling method of claim 1, wherein the training of the double-decoder convolutional self-encoder model employs a complex loss function: L_total=W_rec·L_rec+W_smooth·L_smooth+ W_sparse·L_sparse; wherein, L_rec is a reconstruction loss, which is used for constraining the sum of I_rain and I_ runoff to reconstruct an input map; L_smooth is a smoothness loss applied to I_rain to penalize high frequency variations; l_sparse is a loss of sparsity applied to i_ runoff to encourage sparse representation of texture components.
  9. 9. The decoupling method of claim 1, wherein the overall attenuation characteristics are extracted based on the background spectrum I_rain and input into a regression model to output rainfall intensity, and the energy characteristics are extracted based on the texture spectrum I_ runoff and passed through a threshold discrimination or regression model to output surface water presence and/or water depth.
  10. 10. The decoupling method of claim 1, wherein the decoupled channel estimate i_rain+i_ runoff is used for channel estimation and equalization at the receiving end to reduce the communication error rate in a composite rainfall and water-logging scenario.
  11. 11. A rainfall and surface runoff effect decoupling device, comprising: the CSI acquisition module is used for acquiring the CSI data stream of the wireless link; the preprocessing module is used for carrying out outlier rejection, amplitude calibration and phase purification on the CSI; the map construction module is used for constructing a multi-channel CSI space-time map in a preset time window; the decoupling model module is used for outputting a background map I_rain and a texture map I_ runoff from the encoder based on double-decoder convolution; and the output module is used for outputting the decoupled channel estimation and/or the environment parameters.
  12. 12. An electronic device and/or computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of any of claims 1 to 10.

Description

Rainfall and surface runoff effect decoupling method and device based on wireless channel state information space-time spectrum and storage medium Technical Field The invention relates to the technical fields of wireless environment sensing, flood monitoring, wireless channel state information processing and deep learning signal processing, in particular to a rainfall and surface runoff effect decoupling method, device and storage medium based on a wireless Channel State Information (CSI) space-time spectrum. Background The flood disaster has the characteristics of strong burst, rapid evolution, wide influence range and the like, and the establishment of a monitoring and early warning system which has low cost and high density and can continuously work in a disaster environment is an important technical requirement for disaster reduction and prevention. In the prior art, flood monitoring is dependent on special sensors or hydrologic stations such as rain gauges, water level gauges and the like, and has the defects of high construction and maintenance cost, limited coverage density, and easy occurrence of power supply or communication interruption under extreme weather and flood impact, so that data loss at key moments is caused. In recent years, development of opportunistic environment awareness by using wireless signals in a communication network has become a new direction. Compared with the coarse granularity index of RSSI, the CSI can provide fine granularity information such as multi-subcarrier amplitude/phase and the like, and has higher environmental sensitivity. However, in a real flood or heavy rainfall scene, the wireless channel is often influenced by integral attenuation caused by rainfall and multipath reflection/selective fading formed by surface water at the same time, and the two effects are aliased on the CSI to cause 'homogeneous and heterogeneous' characteristic blurring, so that the existing sensing method based on single characteristic is difficult to accurately estimate the rainfall and water accumulation states at the same time, and the communication receiving end is difficult to perform effective channel compensation. Disclosure of Invention In view of the above problems, the present invention aims to provide a method, an apparatus, and a storage medium for decoupling rainfall and surface runoff effects based on CSI space-time maps and double decoder self-encoders, so as to separate rainfall attenuation components from ponding multipath texture components, thereby improving estimation accuracy of flood-related environmental parameters and enhancing communication reliability in severe environments. The technical scheme adopted by the invention is that the rainfall and surface runoff effect decoupling method based on the wireless channel state information space-time spectrum comprises the steps of CSI acquisition, CSI preprocessing, CSI space-time spectrum construction, double-decoder convolution self-encoder decoupling and parameter estimation/channel equalization based on decoupling components. The invention further provides a rainfall and surface runoff effect decoupling device, which comprises a CSI acquisition module, a preprocessing module, a map construction module, a decoupling model module and an output module, wherein the device can be deployed in a server, an edge computing node or a router/gateway and other equipment. Further, the invention also provides an electronic device and/or a computer readable storage medium, on which a computer program for executing the above method is stored. Compared with the prior art, the invention has the advantages that: (1) Reconstructing one-dimensional CSI time sequence data into a two-dimensional space-time map, so that low-frequency background change caused by rainfall and high-frequency sparse textures caused by ponding are structurally distinguishable, and providing a good representation for automatic decoupling; (2) By adopting a shared encoder and double decoder structure and combining background smoothness constraint and texture sparsity constraint, end-to-end decoupling can be realized under the condition of lacking a pure label, and dependence on prior physical models and scene geometric information is reduced; (3) The decoupled components are respectively used for rainfall intensity estimation and ponding detection/depth estimation, so that the perception precision in a composite scene can be improved; (4) The decoupled cleaner channel estimation can be used for equalization of a receiving end, so that the link robustness in a severe environment is improved, and the error rate is reduced. Drawings In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present application, and other drawings