CN-122028206-A - Intelligent super-surface auxiliary communication resource scheduling method based on attention mechanism

CN122028206ACN 122028206 ACN122028206 ACN 122028206ACN-122028206-A

Abstract

The invention discloses an intelligent subsurface auxiliary communication resource scheduling method based on an attention mechanism, which relates to the technical field of wireless communication and intelligent subsurface auxiliary communication resource scheduling and is used for solving the problems that the resource allocation and reflection control are not enough in cooperation and the link quality, interference suppression and service delay requirements are difficult to consider in an intelligent subsurface auxiliary communication scene; the method comprises the steps of firstly collecting multisource observation data of a base station, a user terminal and an intelligent super-surface, completing channel estimation, service state acquisition and link feedback construction, then extracting interference association relations between users and reflection coupling relations between the users and the intelligent super-surface based on an attention mechanism to form joint characterization, generating candidate scheduling results of time-frequency resources, transmitting power, wave beams and intelligent super-surface reflection parameters according to the joint characterization, correcting the candidate results by combining actual execution conditions, and finally outputting a resource scheduling scheme matched with an actual propagation state.

Inventors

SUN DEZHONG
WANG YULIANG
LI CHENGCHAO
CHEN CHANG

Assignees

上海密阳通信科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. The intelligent super-surface auxiliary communication resource scheduling method based on the attention mechanism is characterized by comprising the following steps of: constructing first observation data, second observation data and third observation data, establishing a direct channel, an incident channel and a reflection channel, and constructing a user state vector and a super-surface state vector; performing attention extraction based on the user state vector and the super-surface state vector, and constructing user terminal joint characterization and intelligent super-surface joint characterization; The method comprises the steps of respectively constructing a candidate time-frequency distribution result, a candidate transmitting power distribution result, a candidate beam distribution result, a candidate phase control quantity, a candidate reflection coefficient control quantity and a candidate starting control quantity based on user terminal joint characterization and intelligent super-surface joint characterization, mapping the candidate phase control quantity into a discrete phase value, mapping the candidate reflection coefficient control quantity into an actual reflection coefficient value, mapping the candidate starting control quantity into an actual starting state, constructing an executing reflection matrix and an executing starting state matrix, constructing a pre-executing reflection component and an executing reflection component based on the candidate reflection matrix, the executing starting state matrix, an incident channel and the reflection channel, calculating a reflection offset, constructing an updating equivalent channel based on the executing reflection component, determining a competitive user terminal set and calculating an interference migration quantity based on the candidate time-frequency distribution result, and further constructing a delay compression quantity based on a service queue length and a service delay constraint in service state information, constructing an offset state vector based on the reflection offset, the interference migration quantity and the delay compression quantity, and executing mismatch judgment and path correction according to the offset state vector, and determining a final resource scheduling result; and based on the final resource scheduling result, executing data transmission of the current scheduling period, constructing execution feedback information, and updating the scheduling state of the next scheduling period.
2. The intelligent super-surface auxiliary communication resource scheduling method based on the attention mechanism of claim 1, wherein the method is characterized by comprising the steps of receiving uplink pilot sequences sent by user terminals through a base station, constructing first observation data, switching reflection states according to training reflection sequences by controlling the intelligent super-surface and receiving uplink pilot responses at each training moment, constructing second observation data, receiving the uplink pilot through an active unit and uploading a receiving result under an active-passive mixed structure, and constructing third observation data.
3. The intelligent subsurface auxiliary communication resource scheduling method based on the attention mechanism according to claim 2, wherein channel estimation is performed based on first observation data, second observation data and third observation data, a direct channel, an incident channel from a base station to an intelligent subsurface and a reflection channel from the intelligent subsurface to a user terminal are established, an equivalent channel is constructed based on the direct channel, the incident channel, the reflection channel and a current reflection matrix, interference association quantity between the user terminals is calculated based on the equivalent channel, reflection coupling quantity between the user terminals and the intelligent subsurface is calculated based on the incident channel, the reflection channel and the current reflection matrix, and further channel enhancement characteristics, service state information, link feedback information, interference association quantity statistics and reflection coupling quantity statistics are spliced to construct a user state vector, and an intelligent subsurface control state and reflection coupling quantity statistics are spliced to construct a subsurface state vector.
4. The intelligent subsurface auxiliary communication resource scheduling method based on the attention mechanism according to claim 3 is characterized by comprising the steps of respectively carrying out linear mapping on a user state vector and a subsurface state vector to construct an inquiry vector, a key vector and a value vector, establishing a first attention weight based on the inquiry vector of a user terminal, the key vectors and interference association amounts of other user terminals, constructing a first aggregation feature representing a competition relationship based on weighted aggregation, establishing a second attention weight based on the inquiry vector of the user terminal, the key vector of the intelligent subsurface and reflection coupling amounts, constructing a second aggregation feature representing a reflection action relationship based on weighted aggregation, fusing the user state vector, the first aggregation feature and the second aggregation feature to construct a joint representation vector, and constructing a user terminal joint representation and an intelligent subsurface joint representation through multi-layer attention extraction and interlayer aggregation.
5. The intelligent subsurface auxiliary communication resource scheduling method based on the attention mechanism, as set forth in claim 4, wherein the method comprises performing mapping calculation based on user terminal joint characterization and intelligent subsurface joint characterization, respectively, constructing a resource priority vector and a reflection control vector, generating a candidate time-frequency allocation result, a candidate transmission power allocation result and a candidate beam allocation result based on the resource priority vector, constructing a candidate phase control amount, a candidate reflection coefficient control amount and a candidate enabling control amount based on reflection control vector splitting, and the candidate time-frequency allocation result, the candidate transmission power allocation result and the candidate beam allocation result jointly form a candidate resource scheduling result.
6. The intelligent subsurface auxiliary communication resource scheduling method based on the attention mechanism according to claim 5, wherein after the candidate resource scheduling result and the candidate intelligent subsurface control result are generated, performing a floor mapping process for the candidate intelligent subsurface control result, mapping the candidate phase control amount to an actually executable discrete phase value, mapping the candidate reflection coefficient control amount to an actually executable reflection coefficient value, and mapping the candidate enabling control amount to an actually enabling state, thereby constructing an executing reflection matrix and an executing enabling state matrix, and constructing pre-executing reflection components corresponding to each user terminal based on the incident channel and the reflection channel obtained by channel estimation and the candidate reflection matrix formed by the candidate phase control amount and the candidate reflection coefficient control result, and constructing executing reflection components corresponding to each user terminal based on the executing reflection matrix and the executing enabling state matrix.
7. The intelligent super-surface auxiliary communication resource scheduling method based on the attention mechanism, as set forth in claim 6, is characterized by calculating a reflection offset by pre-executing a reflection component and executing the reflection component, updating an equivalent channel by executing the reflection component and a direct channel configuration, determining a competitive user terminal set by combining and updating the equivalent channel and a candidate equivalent channel formed by the direct channel and the pre-executing the reflection component by a candidate time-frequency allocation result, calculating an interference migration amount, constructing a delay compression amount by a service queue length and a service delay constraint in service state information, further constructing an offset state vector by splicing the reflection offset, the interference migration amount, the delay compression amount, a reflection coupling amount statistical value and an interference association amount statistical value, and judging a reflection dominant mismatch condition, a competition dominant mismatch condition and a delay dominant mismatch condition according to the offset state vector.
8. The intelligent superficially assisted communication resource scheduling method based on an attention mechanism of claim 7, wherein under a condition of reflection dominant mismatch, a selection order of adjacent discrete phase values is readjusted for a high reflection coupling reflection unit and an enabling state of a boundary reflection unit is redetermined to reconstruct an execution reflection matrix and an execution enabling state matrix, under a condition of contention dominant mismatch, resource group stripping, candidate time-frequency resource remapping and candidate beam replacement are executed for a competing user terminal with the greatest contribution of interference migration, and under a condition of time-delay dominant mismatch, a transmission power duty ratio promotion or a time resource or a frequency resource addition is executed for a corresponding user terminal.
9. The intelligent super-surface auxiliary communication resource scheduling method based on the attention mechanism of claim 8, further introducing reflection recovery gain and competition slow-release gain when the reflection dominant mismatch condition and the competition dominant mismatch condition are simultaneously satisfied, performing sequential path selection on local reflection reconstruction processing and competition relation rearrangement processing, recalculating and updating equivalent channels, receiving signal-to-noise ratios and instantaneous transmission rates after path splitting correction, and reserving corrected candidate resource scheduling results as new reference results when mismatch residues still exist, and performing offset state vector construction, mismatch condition judgment and corresponding path correction again only on user terminals with mismatch residues until the corrected candidate time-frequency allocation results, candidate transmission power allocation results, candidate beam allocation results and execution reflection matrix and execution enabling state matrix are determined as final resource scheduling results.
10. The intelligent subsurface auxiliary communication resource scheduling method based on the attention mechanism of claim 9 is characterized by comprising the steps of taking a corrected time-frequency distribution result, a corrected transmitting power distribution result, a corrected beam distribution result, an execution reflection matrix and an execution starting state matrix which are reserved after mismatching judgment and path splitting correction as execution results of a current scheduling period, updating current reflection state records of the intelligent subsurface based on phase values and reflection coefficient values of reflection units corresponding to the execution reflection matrix and starting values of reflection units corresponding to the execution starting state matrix, acquiring a receiving signal-to-noise ratio, actual throughput and actual transmission delay of each user terminal in a data transmission process of the current scheduling period, constructing execution feedback information, updating service state information of the next scheduling period based on the length of a residual service queue after the current scheduling period is ended, the newly arrived service volume and the residual delay margin of unfinished service, and writing the updated intelligent subsurface state records, execution feedback information and service state information into scheduling states of the next scheduling period.

Description

Intelligent super-surface auxiliary communication resource scheduling method based on attention mechanism Technical Field The invention relates to the technical field of wireless communication and intelligent super-surface auxiliary communication resource scheduling, in particular to an intelligent super-surface auxiliary communication resource scheduling method based on an attention mechanism. Background With the continuous improvement of the requirements of the wireless communication system on spectrum efficiency, transmission rate, link reliability and service delay, the intelligent super surface gradually becomes an important technical means for improving communication performance due to the capability of reconstructing the wireless propagation environment. Under the intelligent super-surface auxiliary communication scene, the user link state is usually optimized by combining the communication parameters of the base station side and the intelligent super-surface reflection parameters so as to improve the coverage effect and the overall transmission capacity of the system. In the prior art, communication resource allocation and intelligent super-surface reflection control are often processed separately, and a unified modeling and collaborative adjustment mechanism is lacked for interference relation among multiple users, user service demand difference and link influence caused by intelligent super-surface reflection state change. Particularly, under the conditions of discrete phase control, reflection coefficient constraint or active and passive mixed structures, candidate control results are easy to deviate from an ideal scheduling state when actually performing landing, so that reflection path gain change, inter-user interference enhancement and link performance fluctuation are caused, and system throughput, transmission reliability and service delay performance are affected. The present invention proposes a solution to the above-mentioned problems. Disclosure of Invention In order to overcome the above-mentioned drawbacks of the prior art, embodiments of the present invention provide an intelligent subsurface assisted communication resource scheduling method based on an attention mechanism, so as to solve the problems set forth in the background art. In order to achieve the above purpose, the present invention provides the following technical solutions: In a preferred embodiment, the method comprises: constructing first observation data, second observation data and third observation data, establishing a direct channel, an incident channel and a reflection channel, and constructing a user state vector and a super-surface state vector; performing attention extraction based on the user state vector and the super-surface state vector, and constructing user terminal joint characterization and intelligent super-surface joint characterization; The method comprises the steps of respectively constructing a candidate time-frequency distribution result, a candidate transmitting power distribution result, a candidate beam distribution result, a candidate phase control quantity, a candidate reflection coefficient control quantity and a candidate starting control quantity based on user terminal joint characterization and intelligent super-surface joint characterization, mapping the candidate phase control quantity into a discrete phase value, mapping the candidate reflection coefficient control quantity into an actual reflection coefficient value, mapping the candidate starting control quantity into an actual starting state, constructing an executing reflection matrix and an executing starting state matrix, constructing a pre-executing reflection component and an executing reflection component based on the candidate reflection matrix, the executing starting state matrix, an incident channel and the reflection channel, calculating a reflection offset, constructing an updating equivalent channel based on the executing reflection component, determining a competitive user terminal set and calculating an interference migration quantity based on the candidate time-frequency distribution result, and further constructing a delay compression quantity based on a service queue length and a service delay constraint in service state information, constructing an offset state vector based on the reflection offset, the interference migration quantity and the delay compression quantity, and executing mismatch judgment and path correction according to the offset state vector, and determining a final resource scheduling result; and based on the final resource scheduling result, executing data transmission of the current scheduling period, constructing execution feedback information, and updating the scheduling state of the next scheduling period. In a preferred embodiment, the base station receives an uplink pilot sequence sent by each user terminal to construct first observation data, the intelligent super-surface is controlled to s