Multi-carrier adaptive base station signal transmission method

By receiving measurement feedback from user equipment in the base station to form a channel state information matrix, determining the candidate data subcarrier set and modulation scheme, and performing amplitude limiting, water injection, and power allocation, the problem of inconsistent reception quality and channel variation in downlink multi-carrier transmission in the base station is solved, and reliable data transmission and efficient resource allocation are achieved.

CN123123094APending Publication Date: 2026-11-17BEIJING YUNMI TECHNOLOGY DEVELOPMENT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611491145.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-23
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In downlink multicarrier transmission at base stations with limited power and time-varying channels, the reception quality is inconsistent after subcarrier modulation selection and allocation. When configuration is not feasible or the channel changes, there is a lack of clear update and stop processing. The overhead of independent configuration of adjacent resources is large. The water-filling solution process is long. Furthermore, the modulation de-level does not comprehensively reflect power release, equivalent noise and channel time-domain fluctuations.

Method used

By receiving measurement feedback from user equipment, a channel state information matrix is ​​formed to determine the candidate data subcarrier set, cyclic prefix length, and candidate modulation scheme. Based on the signal-to-noise ratio (SNR) to verify the modulation scheme, amplitude limiting and power allocation are performed. The configuration table is periodically updated, and infeasible configurations are prioritized for downgrading. Power release and channel fluctuations are combined using the downgrading priority decision factor to form a closed-loop control.

Benefits of technology

It enables reliable data transmission during channel changes, reduces the risk of disconnect between modulation selection and reception quality after allocation, reduces the overhead of independent configuration of adjacent resources, and improves channel adaptability and control efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123123094A_ABST
    Figure CN123123094A_ABST
Patent Text Reader

Abstract

This invention relates to the field of downlink multi-carrier transmission technology for base stations, and discloses a base station signal transmission method based on multi-carrier adaptive transmission. The base station establishes a channel state information matrix based on downlink pilot measurement feedback from user equipment, determining candidate data subcarriers, cyclic prefixes, and candidate modulation schemes. The modulation threshold is converted into a lower bound of transmit power. Within the total data transmit power budget and the subcarrier power upper limit, amplitude limiting and modulation order reduction are performed, and subcarrier aggregation blocks are formed according to coherent bandwidth. Discrete amplitude reduction breakpoints and multi-dimensional de-order decision are used for collaborative solution. A cyclic prefix orthogonal frequency division multiplexing (OFDM) signal is generated based on the verified configuration. The configuration for the next frame is updated when the channel state changes, thereby improving reliable loading capability and reducing solution and control overhead.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of downlink multi-carrier signal transmission technology in wireless communication, and specifically to a base station signal transmission method based on multi-carrier adaptive transmission. Background Technology

[0002] Multicarrier transmission enables frequency-selective channels to be equalized and detected at the receiver by carrying modulation symbols on different subcarriers. Base stations typically select transmission resources, modulation schemes, and transmit power based on channel quality feedback from user equipment. A cyclic prefix is ​​used to absorb effective multipath delay and transmission / reception timing errors; insufficient prefixes can cause inter-symbol interference, while excessively long prefixes reduce the proportion of effective data. Therefore, subcarrier selection, modulation, power, and cyclic prefixes are not independent configuration parameters and need to be coordinated and determined under the same channel measurement reference and the same effective time.

[0003] The modulation order determines the number of bits carried by a single constellation symbol. The received signal-to-noise ratio (SNR) required to meet the target error rate varies with the modulation scheme, coding configuration, and receiver performance. Power allocation directly alters the SNR of each subcarrier. If the modulation scheme is selected first according to a fixed reference power, and then continuous power allocation is used regardless of the modulation threshold, some subcarriers may transmit below the reliability threshold of the corresponding modulation. If unqualified resources are only deleted after allocation, the overall load requirement may be compromised. After a channel change, a previously constrained configuration may become infeasible in the next effective frame.

[0004] Conventional water-filling processing primarily allocates power based on continuous channel conditions. However, actual modulation schemes come from a finite set, and transmitting equipment is also constrained by total data transmission power, single subcarrier power limits, available resources, and user equipment capabilities. When determining modulation and power independently for each subcarrier, adjacent subcarriers that are highly correlated in the frequency domain may repeatedly cross the discrete modulation threshold due to small estimation disturbances, increasing the control configuration workload. Directly and repeatedly approximating the water-filling level within a continuous real number range also consumes significant baseband computation time when there are many candidate resources or short update cycles.

[0005] Furthermore, when the sum of the lower bounds of the transmit power exceeds the budget, simply downgrading the modulation order based on the power release amount or fixed numbering, without simultaneously considering the equivalent noise cost and channel time-domain fluctuations, may prematurely downgrade resources that are relatively stable and suitable for carrying high-order data. If multipath delay information is only used to select the cyclic prefix and not to describe the frequency domain correlation range, it is also difficult to provide boundaries for the joint configuration of adjacent subcarriers. Existing processing methods also need to clarify the feedback data source, power reference plane, discrete modulation termination conditions, and pause or migration paths when no feasible configuration is available, in order to avoid unbounded iteration and inconsistent configuration states. Summary of the Invention

[0006] The purpose of this invention is to provide a base station signal transmission method based on multi-carrier adaptive modulation. The technical problems this invention aims to solve are: in downlink multi-carrier transmission of base stations with limited power and time-varying channels, there is inconsistency in the received quality after subcarrier modulation selection and allocation; there is a lack of clear update and stop processing when configuration is infeasible or the channel changes; the overhead of independent configuration of adjacent resources is large; the water-filling solution process is lengthy; and the modulation de-leveling does not comprehensively reflect power release, equivalent noise, and channel time-domain fluctuations.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0008] Multi-carrier adaptive base station signal transmission methods include:

[0009] Receive measurement feedback generated by the user equipment based on the downlink pilot, and form a channel state information matrix including reference signal-to-noise ratio, channel power gain, equivalent noise power and multipath delay;

[0010] The candidate data subcarrier set, cyclic prefix length, and candidate modulation scheme are determined based on the channel state information matrix.

[0011] The lowest received signal-to-noise ratio (SNR) corresponding to the candidate modulation scheme is converted into the lower bound of the transmit power. The lower bound of the transmit power, the upper limit of the subcarrier power, and the total data transmit power budget constraint limit the water injection. Based on the allocated SNR, the candidate modulation scheme is verified. For infeasible configurations, only modulation downgrading is performed in the fixed channel state. Priority is given to selecting the downgrading object with a larger power release corresponding to the unit load bit reduction and redistributing the power until the power and modulation constraints are met or no feasible configuration is determined. The transmission configuration table or infeasible mark is output.

[0012] According to the transmission configuration table, the data to be transmitted is modulated, mapped, inverse Fourier transformed, and cyclic prefix added. The multi-carrier signal is then transmitted through the radio frequency front end. If no feasible configuration is available, the corresponding data transmission is paused.

[0013] The newly acquired channel state is periodically compared with the channel state corresponding to the effective transmission configuration table. When the change exceeds the update threshold, the candidate data subcarrier set, cyclic prefix length, modulation scheme and transmit power are re-determined, and the updated configuration that passes the verification is used for the transmission of the next frame.

[0014] As a preferred embodiment of the present invention, the user equipment performs least-squares channel estimation at the downlink pilot position, interpolates the channel frequency response at the pilot position, and forms the channel frequency response at the data subcarrier position; the measurement feedback includes the channel frequency response, equivalent noise power, pilot reference power, measurement timestamp, and multipath delay; the base station normalizes the reference signal-to-noise ratio according to the pilot reference power, and converts the average of the linear reference signal-to-noise ratio over the most recent 10 effective symbol periods into a decibel value.

[0015] As a preferred embodiment of the present invention, the base station marks subcarriers with an average reference signal-to-noise ratio (SNR) higher than 8dB as high quality, subcarriers with an average SNR of not less than 3dB and not more than 8dB as medium quality, and subcarriers with an average SNR lower than 3dB as low quality. Candidate data subcarriers are selected according to channel quality and the data to be transmitted, and 64-QAM, 16-QAM, and QPSK are used as initial candidate modulation schemes. The candidate modulation schemes are determined as the transmission modulation schemes after being verified by the lowest received SNR.

[0016] As a preferred embodiment of the present invention, the cyclic prefix length is determined jointly based on the maximum relative delay of the effective multipath, the receiving timing error margin, and the digital filter memory length, and rounded up to the number of sampling points; subcarriers in the same multicarrier symbol use the same cyclic prefix length; under 1024-point inverse Fourier transform and 15kHz subcarrier spacing, the short cyclic prefix contains 120 sampling points, the long cyclic prefix contains 256 sampling points, and the transmission of the corresponding configuration is suspended when neither cyclic prefix can cover the effective channel memory.

[0017] As a preferred embodiment of the present invention, the base station forms an equivalent noise reduction quantity based on the ratio of equivalent noise power to channel power gain, and multiplies the minimum linear received signal-to-noise ratio corresponding to the candidate modulation scheme with the equivalent noise reduction quantity to obtain the lower bound of the transmit power; the minimum received signal-to-noise ratio is determined by the reliability threshold table corresponding to the modulation scheme, coding configuration and target error rate, and the candidate modulation order is reduced when the lower bound of the transmit power exceeds the upper limit of the subcarrier power.

[0018] As a preferred technical solution of the present invention, the base station divides the total data transmission power budget by the number of candidate data subcarriers at the beginning of the current solution to obtain the reference average power, and sets the subcarrier power upper limit to twice the reference average power; the subcarrier power upper limit is fixed within one solution, and the difference between the water level and the equivalent noise conversion is truncated by upper and lower bounds, and the water level is adjusted by binary search so that the sum of the allocated power does not exceed the total data transmission power budget.

[0019] As a preferred technical solution of the present invention, when the sum of the lower bounds of the transmit power exceeds the total data transmit power budget, the demotion object is selected according to the power release amount corresponding to the reduction in unit loaded bits, and when the sorting values ​​are the same, the demotion object is selected in descending order of the subcarrier number; in the fixed channel state, only the candidate modulation demotion is performed, and the candidate modulation upmotion is not performed; when all the retained subcarriers still cannot meet the minimum modulation method, an infeasibility mark is output, or when the delayed transmission is allowed, the infeasible subcarrier is removed and the solution is resolved.

[0020] As a preferred embodiment of the present invention, the transmission configuration table includes a configuration identifier, an effective frame number, a user equipment identifier, a subcarrier position, a modulation scheme, a coding configuration, a transmit power, and a number of cyclic prefix sampling points. Before the effective frame, the base station sends the transmission configuration table through pre-agreed control resources. After obtaining confirmation from the user equipment, the base station performs 1024-point inverse Fourier transform, cyclic prefix addition, digital filtering, digital-to-analog conversion, and radio frequency transmission in the effective frame.

[0021] As a preferred embodiment of the present invention, the base station calculates the block error rate based on the transport block acknowledgment information returned by the user equipment, divides the number of first transport blocks corresponding to valid negative acknowledgments by the total number of valid first transport blocks; when the block error rate exceeds a preset threshold, the upper limit of candidate modulation of the resources used by the associated transport block is reduced and coding redundancy is increased, and the power is reallocated using the reliability threshold table with increased coding redundancy; the bit error rate is independently calculated by comparing the known test bits bit by bit.

[0022] As a preferred technical solution of the present invention, in multi-user transmission, the base station allocates non-overlapping time-frequency resources according to the channel state information matrix and minimum loading requirements of each user equipment, and the same subcarrier is allocated to one user equipment in the same symbol period; modulation and power calculation are performed within the total data transmission power budget of the base station and the corresponding power budget of the user equipment, and after writing the frequency domain symbols of each user equipment into different resource grid positions, a unified inverse Fourier transform is performed.

[0023] As a preferred embodiment of the present invention, when multiple component carriers participate in transmission, the base station receives the component carrier activation list, the aggregation capability supported by the user equipment, and interference coordination constraints; when an infeasible marker appears on a component carrier, the data that can be migrated is allocated to a backup component carrier that meets the quality constraints; when the interference coordination constraints require reducing interference to neighboring cells, the upper limit of the relevant subcarrier power is lowered and the solution is recalculated; when external interference occurs and power reduction cannot restore quality, resource migration or suspension of transmission is selected.

[0024] As a preferred technical solution of the present invention, in the high-speed mobile scenario, the base station uses continuous channel state records with timestamps to perform linear prediction, sets the prediction target time as the effective time of the transmission configuration table; forms an equivalent noise reduction quantity based on the predicted channel state and inputs it into the amplitude limiting and water injection algorithm; when the prediction span exceeds the allowable range or the historical record is invalid, the latest valid measurement is used and the upper limit of the candidate modulation is limited; the channel monitoring period is set to 2ms and the update threshold is set to 2dB.

[0025] As a preferred technical solution of the present invention, after receiving the measurement feedback generated by the user equipment based on the downlink pilot, the base station extracts the delay power spectrum composed of the non-zero path power value of the effective multipath and the relative arrival delay from the power delay information of the measurement feedback, calculates the normalized frequency domain correlation coefficient of the channel frequency response at the non-zero frequency interval based on the delay power spectrum, and determines the maximum frequency span from the zero frequency interval whose absolute value of the frequency domain correlation coefficient is continuously not lower than the preset correlation threshold as the coherence bandwidth;

[0026] Under the condition that the frequency span does not exceed the coherent bandwidth and the difference between the maximum and minimum values ​​of the channel power gain in dB within the block is lower than the fluctuation threshold, adjacent and continuous candidate data subcarriers are divided into the same subcarrier aggregation block. The same candidate modulation scheme is configured within the same subcarrier aggregation block. The maximum equivalent noise reduction within the subcarrier aggregation block is determined as the block reference equivalent noise reduction, and the lower bound of the transmit power of each candidate data subcarrier is uniformly determined according to the block reference equivalent noise reduction.

[0027] When performing amplitude-limited water injection, the same subcarrier aggregation block is treated as a whole variable in the water injection level search. The total allocated power of the block is calculated based on the water injection level, the lower bound of the transmit power, the upper bound of the subcarrier power, and the block reference equivalent noise reduction. First, the lower bound of the transmit power is configured for each candidate data subcarrier in the block. Then, the remaining power after deducting the sum of the lower bounds of the transmit power in the block from the total allocated power of the block is finely allocated according to the ratio of the equivalent noise reduction of each candidate data subcarrier to the sum of the equivalent noise reduction of the block. The power of candidate data subcarriers that have reached the upper bound of the subcarrier power is stopped from being increased. The unallocated power is redistributed among the remaining candidate data subcarriers that have not reached the upper bound of the subcarrier power according to the ratio of the renormalized equivalent noise reduction.

[0028] As a preferred embodiment of the present invention, during the process of searching and adjusting the water injection level, the base station calculates the lower limit breakpoint formed by adding the lower limit of the transmission power and the equivalent noise reduction, and the upper limit saturation breakpoint formed by adding the upper limit of the subcarrier power and the equivalent noise reduction, based on the equivalent noise reduction, the lower limit of the transmission power, and the upper limit of the subcarrier power for each candidate data subcarrier. Duplicate breakpoints are eliminated, and the breakpoints are sorted from smallest to largest to form an ordered discrete breakpoint vector. In each iteration of the search, based on the adjacent breakpoint interval where the current water injection level is located, the first subcarrier subset reaching the upper limit of the subcarrier power, the second subcarrier subset remaining at the lower limit of the transmission power, and the third subcarrier subset in the power linear adjustment zone are determined.

[0029] When the third subcarrier subset is empty and the second subcarrier subset is not empty, and there is a positive remaining unallocated power after deducting the power occupied by the first and second subcarrier subsets from the total data transmit power budget, the water level search is terminated early. The remaining unallocated power is allocated according to the equivalent noise reduction weight of each candidate data subcarrier in the second subcarrier subset. Power is stopped from being increased for candidate data subcarriers that have reached the subcarrier power limit. Among the remaining candidate data subcarriers in the second subcarrier subset that have not reached the subcarrier power limit, the unallocated power is redistributed according to the proportion of their respective equivalent noise reduction to the sum of the equivalent noise reduction of the remaining candidate data subcarriers. When the second subcarrier subset is empty or the remaining unallocated power exceeds the remaining power space of the second subcarrier subset, the unallocated power is reserved as unused budget.

[0030] As a preferred embodiment of the present invention, when selecting de-leveling targets according to the power release amount corresponding to the reduction in unit loaded bits, the base station establishes a de-leveling priority decision factor. The demotion priority decision factor is calculated according to the following formula:

[0031]

[0032] The following formula is input: (The first...) Current modulation scheme of candidate data subcarriers Corresponding lower limit of transmit power Adjacent lower modulation mode Corresponding lower limit of transmit power The number of load bits reduced by downgrading Equivalent noise reduction Channel frequency domain response statistics and Preset weighting factors and ;

[0033] Output: The Demotion priority decision factor for each candidate data subcarrier The larger the value, the higher the priority of demotion; in the formula, The first in the candidate data subcarrier set The de-priority decision factor for each candidate data subcarrier, in mW / bit; For the first The current candidate modulation scheme for each candidate data subcarrier; For the first Each candidate data subcarrier is adjacent to a candidate modulation scheme of a lower level; For the first Each candidate data subcarrier in the current candidate modulation scheme The corresponding lower bound of the transmit power, in mW; For the first Each candidate data subcarrier is downgraded to the next lower candidate modulation scheme. The corresponding lower bound of the transmit power, in mW; For the first The number of loaded bits reduced by downgrading each candidate data subcarrier, in bits; For the first The equivalent noise reduction of each candidate data subcarrier, in mW; This is the set of indices for all currently degradeable candidate data subcarriers; For set The Middle The equivalent noise reduction of each candidate data subcarrier, in mW; The channel severity penalty weighting factor is a dimensionless constant ranging from 0.1 to 0.5. The channel time-domain fluctuation penalty weighting factor is a dimensionless constant ranging from 0.05 to 0.2. For the first The complex channel frequency response of each candidate data subcarrier within the sliding window is a dimensionless complex number. For the first The variance of the channel power gain of each candidate data subcarrier within the sliding window is a dimensionless quantity. For the first The average channel power gain of each candidate data subcarrier within the sliding window is a dimensionless quantity that is greater than zero. The square of the average channel power gain is a dimensionless quantity greater than zero. Represents a set The index summation is performed; except for the demotion priority decision factor, the lower bound of each transmit power, the number of loaded bits, and each equivalent noise reduction quantity, which have explicit dimensions, the remaining parameters and weighting factor terms are dimensionless; the base station recalculates the set after each candidate modulation demotion. and the priority decision factor for demotion, and according to Candidate data subcarriers are selected in descending order to perform candidate modulation order reduction.

[0034] The present invention also discloses a base station signal transmission system based on multi-carrier adaptive transmission, including a feedback receiving interface, a baseband processor, a memory, a radio frequency front-end, and an antenna. The baseband processor is connected to the feedback receiving interface, the memory, and the radio frequency front-end, and the radio frequency front-end is connected to the antenna.

[0035] The memory stores the reliability threshold table, channel state record, and transmission configuration table. The baseband processor receives measurement feedback from the user equipment based on the downlink pilot through the feedback receiving interface, establishes the channel state information matrix, determines the candidate data subcarrier set, cyclic prefix length, and candidate modulation scheme, converts the lowest received signal-to-noise ratio into a lower bound of transmit power, performs amplitude limiting and water injection within the upper and lower bounds of power and the total data transmit power budget, and reviews the candidate modulation scheme based on the allocated signal-to-noise ratio. For infeasible configurations, only modulation downgrading is performed under fixed channel conditions. Priority is given to selecting downgrading objects with larger power release corresponding to the reduction in unit loaded bits and reallocating power to generate a transmission configuration table or infeasibility mark.

[0036] The baseband processor generates a time-domain sampling sequence with a cyclic prefix according to the transmission configuration table, provides the time-domain sampling sequence to the radio frequency front end, and transmits it through the antenna; the baseband processor compares the channel state changes based on the newly received measurement feedback, and re-solves and updates the configuration for the next frame when the change exceeds the update threshold, and suspends the corresponding data transmission when there is no feasible configuration.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] This invention establishes a channel state information matrix by receiving measurement feedback generated by the user equipment based on the downlink pilot from the base station, so that the candidate data subcarrier set, cyclic prefix length, and candidate modulation scheme are determined under the same channel measurement reference. The lowest received signal-to-noise ratio corresponding to the candidate modulation scheme is converted into the lower bound of the transmit power. Amplification and water injection are performed within the total data transmit power budget and the subcarrier power upper limit. The candidate modulation scheme is verified based on the allocated signal-to-noise ratio. For infeasible configurations, only modulation order reduction and power reallocation are performed under fixed channel conditions, forming a closed loop between modulation selection, power allocation, post-allocation verification, and transmission configuration table output, avoiding the disconnect between modulation selection and post-allocation reception quality.

[0039] Furthermore, subcarrier aggregation blocks are formed according to coherent bandwidth to reduce the overhead of independent configuration of adjacent related resources; discrete water injection breakpoints are used to locate water injection intervals to reduce continuous search processes; a de-priority decision factor is adopted to integrate power release, equivalent noise, and channel time-domain fluctuations, so that limited power is preferentially released from resources with higher channel costs or greater fluctuations. A cyclic prefix orthogonal frequency division multiplexing signal is generated based on the verified configuration, and the configuration of the next frame is updated when the channel state changes, improving reliable loading capability and reducing solution and control overhead. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0041] Figure 1 This is a flowchart of the overall method of the present invention.

[0042] Figure 2 This is a basic transmission flowchart for a single user in this invention.

[0043] Figure 3 This is a flowchart of the multi-user mutual exclusion resource allocation process of the present invention.

[0044] Figure 4 This is a flowchart of the frequency domain coherent block division and discrete water injection breakpoint processing of the present invention. Detailed Implementation

[0045] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the embodiments of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0046] The following is in conjunction with the appendix Figures 1-4 The embodiments of the present invention will be described in detail below.

[0047] This method is used for base stations to transmit CP-OFDM signals to user equipment. The base station and user equipment pre-agree on available subcarrier positions, pilot positions, control resources, allowed cyclic prefix sampling number, modulation scheme, and coding configuration. Subcarrier-by-carrier modulation and custom cyclic prefixes are implemented through the matching receiver; the transmitter cannot change the parameters and require the receiver, which is unaware of the configuration, to continue using the old demodulation method. The reference signal-to-noise ratio (SNR) in this specification is the received SNR corresponding to the specified pilot reference transmit power; the allocated SNR is the received SNR calculated based on the channel gain, equivalent noise power, and the transmit power of the current data. When external interference that can be treated as noise exists, the external interference is included in the equivalent noise power, and the accuracy is maintained within the same measurement record.

[0048] For example, the following numerical calculations use a unit subcarrier reference power of 1mW, a data power budget of 14mW, 20 candidate data resources, a maximum allowed data resource of 14, and a minimum load requirement of 44 bits per data symbol. The reliability calculations use uncoded Gray-mapped modulation and additive Gaussian noise conditions, with a target bit error rate of 0.001 per subcarrier. All power values ​​in the digital baseband are converted to the same transmit reference plane. Pilot and control resources occupy separate reserved budgets and are not included in the 14mW data budget. Fixed gains for filtering, digital-to-analog conversion, and the RF link are incorporated into the reference plane conversion through calibration; unknown link gains cannot be directly assumed to be 1.

[0049] Each solution cycle freezes a snapshot of the channel state. New feedback can be written to the next snapshot, but it does not change the channel gain or noise in the middle of the current binary solution. The transmission configuration table carries the corresponding snapshot identifier, enabling the association of configuration, measurement, and subsequent feedback. The data resource list only includes frequency domain locations that can carry data; pilot, control, and DC locations are not included. The 1024-point transformation indicates that a time-domain symbol is formed by 1024 valid samples, not that all 1024 frequency domain locations must carry user data. The minimum load requirement is provided by the scheduler based on the amount of data to be transmitted, the number of data symbols allowed, and the coding configuration. The scheduler also specifies whether data delays are allowed, whether cross-component carrier migration is allowed, and whether a pause is necessary if the requirement is not met. The baseband processor cannot automatically reduce service requirements and still output a success flag indicating that the original requirements are met. The threshold qualified load is a check metric that sums the number of bits carried by resources that meet the set reliability threshold; the entire configuration must pass the single resource quality, power, and load requirement checks simultaneously before it takes effect.

[0050] Step S1: The user equipment receives the downlink pilot and the pre-known pilot symbols. After completing synchronization and prefix removal, it performs frequency domain transformation and divides the received complex symbols by the known transmitted complex symbols at the pilot position to obtain the least squares frequency response estimate.

[0051] User equipment (UE) interpolates the real and imaginary parts of adjacent valid pilots to generate the frequency response for the data subcarrier positions. For resources outside pilot coverage, measured information is not replaced by unbounded extrapolation values; resources lacking valid estimates are marked as unavailable. Equivalent noise power can be obtained through reserved empty resource measurements or repeated pilot residual estimation, with the measurement method remaining consistent within a single snapshot. UE feeds back frequency response and power delay information to the base station. When feedback containing only the broadband quality index is insufficient to recover arbitrary per-subcarrier complex responses and multipath delays, a pre-agreed extended feedback field is used. The base station does not directly calculate the UE-side channel from unreceived downlink pilot sequences. The channel state information matrix uses candidate resource numbers as row indices, with each row storing the reference signal-to-noise ratio, channel gain modulus squared, equivalent noise power, and effective state. Measurement timestamps, multipath delay records, and pilot reference power are stored as matrix association fields.

[0052] For example, if the pilot reference power is 1mW, the channel power gain of a certain subcarrier is 0.1, and the equivalent noise power is 0.001mW, then the received reference signal power is 0.1mW, the linear reference signal-to-noise ratio is 100, and the decibel value is 20dB. The equivalent noise is calculated by dividing 0.001mW by 0.1, resulting in 0.01mW. A sliding window stores the linear reference signal-to-noise ratio for the most recent 10 effective symbol periods. When the pilot power varies, it is first converted to a unified reference power, and then the arithmetic mean is calculated. After averaging, the logarithm to base 10 is taken and multiplied by 10 to obtain the decibel value used for classification.

[0053] For example, if all 10 records are linearly 100, the average is 100, corresponding to 20dB; if 5 records are 100 and 5 records are 10, the linear average is 55, corresponding to 17.4036dB. Valid states are recorded simultaneously for whether they have expired, whether noise measurements are missing, and whether receiver saturation exists. Invalid records are not included in the data injection. Zero or negative noise power is considered invalid input; when near-zero channel gain results in excessive power cost, it is treated as resource infeasibility to avoid division by zero or generating unbounded power.

[0054] For example, within the same measurement window, if the pilot reference power changes from 1mW to 2mW, the received linear signal-to-noise ratio changes from 100 to 200, while the channel gain and noise remain unchanged, the result after conversion to a 1mW reference power is still 100. The change in pilot power should not be mistakenly identified as a 3.0103dB channel improvement, nor should it trigger a modulation upgrade without physical basis.

[0055] For example, by taking the squared amplitude average of eight independent noise observations after removing the definite pilot component, a noise power estimate can be obtained; the number of observations, empty resource locations, and receiver gain are recorded by the measurement configuration. When using repeated pilot residuals, the estimated definite channel response is eliminated first, and then the residuals are statistically analyzed to prevent the strong pilot itself from being included in the noise. The time origin for multipath delay recording is selected as the first valid path after receiver synchronization. All paths are described relative to the same time origin; the delay origins of different measurement windows cannot be mixed and the maximum value is taken directly.

[0056] In step S2, the base station reads the list of available data resources, the maximum number of resources, data buffer requirements, and minimum loading requirements provided by the scheduler. Pilot, control, and DC positions are excluded before selecting data resources to avoid including reserved resources in the data throughput. When there are 20 candidate data subcarriers, resources with a reference signal-to-noise ratio (SNR) higher than 8dB are marked as high quality, resources with a reference SNR of not less than 3dB and not more than 8dB are marked as medium quality, and resources with a reference SNR lower than 3dB are marked as low quality; resources with a reference SNR of 8dB are classified as medium quality, and resources with a reference SNR of 3dB are classified as medium quality.

[0057] With sufficient data buffer and a resource limit of 14, high-quality resources are prioritized, followed by medium-quality resources, resulting in a maximum of 14 candidate data subcarriers. Low-quality resources are excluded by default; if the service allows continued attempts at low-quality resources, the same power feasibility verification must still be performed, and quality requirements cannot be waived due to high load. High-quality resources start with 64-QAM, medium-quality resources start with 16-QAM, and low-quality resources start with QPSK. 8dB and 3dB are only used for initial candidate sorting and are not thresholds to guarantee the reliability of the corresponding modulation.

[0058] Cyclic prefixes are uniformly selected using the same CP-OFDM symbol. The root mean square delay reported by the user equipment (UE) describes the multipath dispersion, while the maximum effective relative delay is used for coverage checking; the two are not interchangeable. Effective multipath is determined based on the receiver's preset path detection rules. Tail energy below the detection threshold but not negligible needs to be covered by additional margin or a longer prefix. For example, effective multipath delays of 0 μs, 2 μs, and 5 μs, normalized path powers of 0.7, 0.2, and 0.1, average delay of 0.9 μs, second-order delay moment of 3.3 μs², delay variance of 2.49 μs², and root mean square delay of 1.5780 μs. Prefix coverage is still based on a maximum effective delay of 5 μs. A 1024-point transform and a 15 kHz subcarrier spacing correspond to a 15.36 MHz sampling rate, with an effective symbol duration of 66.6667 μs. The short cyclic prefix of 120 points corresponds to 7.8125 μs, and the long cyclic prefix of 256 points corresponds to 16.6667 μs. These two values ​​are the corresponding transmit / receive configuration for this implementation method, and the 120-point configuration is not named the regular cyclic prefix of the Long Term Evolution (LTE) system.

[0059] For example, with a receiving timing error margin of 0.5 μs and a digital filter memory of 2 sampling intervals, corresponding to 0.130208 μs, the maximum effective delay of 5 μs plus the two margins equals 5.630208 μs, which translates to 86.48 sampling intervals. Rounding up to 87 points, a 120-point prefix is ​​sufficient for coverage. When the maximum effective delay becomes 12 μs, calculated using the unrounded sampling interval, the coverage requirement is 12.630208333 μs, equivalent to 194 sampling intervals. 194 points are sufficient for coverage, so a 256-point prefix is ​​selected. If the requirement exceeds 256 points, this implementation does not continue transmitting uncovered configurations but instead selects another allowed frame structure or suspends transmission. Prefix changes affect the data symbol duration. Under short prefix conditions, a complete symbol has 1144 points, while under long prefix conditions it has 1280 points. The same single-symbol load corresponds to a lower load rate under long prefix conditions; therefore, the minimum service rate needs to be recalculated into a per-symbol load requirement. Dividing 48 bits by the durations corresponding to 1144 points and 1280 points respectively yields 644475.524 bits / s and 576000 bits / s. The choice of the long prefix is ​​used to satisfy the delay coverage condition.

[0060] Step S3: The candidate modulation outputs the minimum transmit power, the limiting and water-filling outputs the actual power, and the actual power is returned to the candidate modulation processing to form a verification. The modulation order is a finite discrete quantity and cannot be directly replaced by continuous gradient descent for discrete selection. For a fixed channel snapshot, the index in the candidate data subcarrier set is denoted as... The equivalent noise reduction is denoted as... The minimum received signal-to-noise ratio (SNR) is obtained by dividing the equivalent noise power by the channel power gain, and its unit is mW. The minimum received SNR uses a linear value, obtained by looking up a table based on the candidate modulation and coding configuration. The minimum received SNR is related to... Multiplying them gives the lower bound of the transmit power. At the start of this solution, a reference average power is formed by dividing the total data transmit power budget by the number of candidate data subcarriers, and twice the reference average power is set as the power upper limit. For example, a 14mW budget and 14 candidate data subcarriers correspond to a 1mW reference average power and a 2mW upper limit. The 2mW upper limit is not raised again after removing resources in a single solution. For candidate configurations that satisfy the lower bound not being greater than the upper bound, the following formula is used for amplitude limiting and water injection.

[0061] (1)

[0062] Among them: the Lower bound of transmit power for candidate data subcarriers Subcarrier power limit Equivalent noise reduction Water level Output: The first Allocating transmit power to candidate data subcarriers ;in, For the first The allocated transmit power for each candidate data subcarrier is expressed in mW. For the first The upper limit of subcarrier power for each candidate data subcarrier, in mW; For the first The lower bound of the transmit power of each candidate data subcarrier under the candidate modulation scheme, in mW; The water level is measured in mW. For the first The equivalent noise reduction of each candidate data subcarrier, in mW; This indicates taking the smaller value; This indicates taking the larger value; The first in the candidate data subcarrier set Index of candidate data subcarriers.

[0063] The inputs to formula (1) are the upper and lower bounds of power, the equivalent noise reduction, and the water level, and the output is the allocated power. The lower bound represents the minimum budget required to meet the candidate modulation threshold, the upper bound represents the maximum budget allowed, and the water level in the middle determines the power of the subcarriers that are not fixed by the boundary. If any lower bound exceeds the corresponding upper bound, the corresponding modulation scheme is reduced first; 64QAM is reduced to 16QAM, and then to QPSK. If the minimum modulation is still not feasible, the resource infeasibility status is output. If the service allows for a delay, the relevant data resources are removed; otherwise, the solution is terminated. When the lower bounds do not exceed the upper bounds but the sum of the lower bounds exceeds the total budget, the lower bound power that can be released by each regressable subcarrier is calculated by regressing one level, and then divided by the number of loaded bits reduced by regression; resources with larger ratios are preferred for regression. When the ratios are the same, they are processed in descending order of number. Only one candidate modulation is changed each time, and the sum of the lower bounds is recalculated. After the sum of the lower bounds does not exceed the budget, the water level is solved. The lower end of the binary search is 0mW, and the upper end is all. and The maximum value in the sum. Calculate the midpoint in each round and calculate the total power distribution according to formula (1). Lower the upper end if the sum exceeds the budget, otherwise raise the lower end.

[0064] For example, the total power tolerance is set to 0.000001mW, and the maximum number of binary divisions is set to 40. The process ends when the tolerance is reached; when the maximum number of divisions is reached, the result on the side that does not exceed the budget is used and re-verified. When the total budget is greater than the sum of all upper bounds, only the upper bound is allocated, and the remaining budget is not forcibly used up. Formula (1) is monotonically non-decreasing for the water injection level. Under the condition that both the effective lower bound and the upper bound are fixed, the total power increases continuously from the lower bound sum to the upper bound sum, so when the total budget is between the two, the corresponding allocation can be obtained through binary division. When the total budget is lower than the lower bound sum, binary division without feasible intervals is not performed, but discrete de-leveling is returned. Floating-point errors may cause slight deviations between the calculated value and the threshold. The final configuration saves the power with sufficient precision and recalculates the total and the quality after allocation under the same numerical precision. The allowable binary power error cannot be used to relax the reliability threshold; when there is a rounding result lower than the lower bound, the representation that meets the lower bound is used and the total budget is checked again. The signal-to-noise ratio (SNR) of each candidate data subcarrier after linear allocation is obtained by dividing its allocated transmit power by the equivalent noise reduction factor. Here, the SNR after linear allocation is dimensionless; the units of allocated transmit power and equivalent noise reduction factor are both mW; and the candidate data subcarrier index is... The assigned signal-to-noise ratio is used to compare reliability thresholds and provides input for bit error rate calculation.

[0065] The reliability threshold table can be established from the calibration results of the matching modulation, coding, and receiver. The following uncoded numerical calculations use additive Gaussian noise, coherent reception, Gray mapping, and the condition that there is no uncompensated frequency offset. The bit error rate (BER) estimate for a square constellation is calculated using the following commonly used approximation: First, divide the signal-to-noise ratio (SNR) after three times the linear distribution by the number of constellation points minus 1 and take the square root. Input the resulting value into the right-tail probability function of the standard normal distribution. Then, multiply the right-tail probability function value by four times the difference between one and the reciprocal of the square root of the number of constellation points, and divide by the base-2 logarithm of the number of constellation points. The estimated BER corresponds to the... There are 10 candidate data subcarriers; the number of constellation points can be 4, 16, or 64; the input to the right-tail probability function of the standard normal distribution is a dimensionless real number. For QPSK, the above relationship gives the corresponding bit error probability under the listed conditions; for higher-order square constellations, the above relationship adopts a proximity-decision-error-dominated approximation and is not considered an accurate result at low signal-to-noise ratios.

[0066] For example, when the target bit error rate (BER) is 0.001, 10, 50, and 200 are selected as the linear minimum received signal-to-noise ratios (SNRs) for the three modulation schemes. Based on the approximate BER relationship for the square constellation mentioned above, the estimated BER for QPSK with a linear minimum received SNR of 10 is 0.000782701, for 16QAM with a linear minimum received SNR of 50 it is 0.000587026, and for 64QAM with a linear minimum received SNR of 200 it is 0.000591567, all lower than the target value. These thresholds include discrete margins and do not represent the mathematically smallest decodeable threshold.

[0067] The lowest received signal-to-noise ratio (SNR) in the threshold table represents the threshold at which the corresponding modulation is allowed to be transmitted in this configuration. In actual configurations, continuous thresholds that meet the target error rate can be rounded down to more conservative discrete values; different coding configurations, receivers, or interference models correspond to different entries. For example, a threshold of 10 for QPSK corresponds to 10 dB, a threshold of 50 for 16QAM corresponds to 16.9897 dB, and a threshold of 200 for 64QAM corresponds to 23.0103 dB. Therefore, an initial screening of 8 dB does not directly guarantee that 64QAM meets the 0.001 target of this implementation; the two types of thresholds have different uses.

[0068] If the quality falls below a threshold after allocation, further downgrading or an infeasible state is output. Within a fixed snapshot, candidate modulations only decrease, never increase, with downgrading performed at most according to a finite number of modulation levels for each resource. For example, with 14 resources each having 64QAM, 16QAM, and QPSK constellation candidates, a maximum of 28 modulation downgrading operations are performed from the highest to the lowest order. If removing resources that remain infeasible even at the lowest order is allowed, a maximum of 14 more resource removal operations are performed. The binary search for each candidate state is performed at most 40 times, thus preventing unconditional loops within the same snapshot. This bounded processing generates candidate configurations that satisfy the established constraints.

[0069] Step S4: The transmission configuration table includes a configuration identifier, effective frame number, resource location, user equipment identifier, modulation, coding, power, prefix sampling number, and the channel snapshot identifier used. The user equipment obtains demodulation information through pre-agreed control resources, verifies and confirms it before switching the receive configuration. The control resources use fixed configurations already supported by both parties; the user equipment cannot be required to demodulate unknown new data configurations before obtaining the new configuration.

[0070] If the arrival confirmation is not received, the base station will not send data dependent on the new configuration; if the old configuration still meets the current quality constraints, it can continue to be used; otherwise, it will be paused. Data transmission is encoded and mapped to the constellation according to the configuration. QPSK uses four equal-energy constellation points; 16QAM uses four amplitude levels for both the real and imaginary parts; 64QAM uses eight amplitude levels for both the real and imaginary parts. Gray notation is used to ensure that adjacent levels change by only one bit. The constellation is normalized by average energy and then multiplied by the square root of the corresponding power to ensure consistent power meaning across different modulations. For example, 16QAM uses four levels per axis: -3, -1, +1, and +3, with an average complex constellation energy of 10 amplitude square units, which is normalized by dividing by the square root of 10. 64QAM uses eight levels per axis: -7, -5, -3, -1, +1, +3, +5, and +7, with an average complex constellation energy of 42 amplitude square units, which is normalized by dividing by the square root of 42.

[0071] The complex symbols of the data are written to the frequency domain positions specified in the transmission configuration table. Pilot and control positions are filled with known information separately; positions not enabled, DC, and protection are set to zero. After performing a 1024-point inverse Fourier transform, the last 120 or 256 points of the time-domain sequence are copied to the beginning of the sequence to form a prefixed time-domain sample. Digital filtering is performed after the cyclic prefix is ​​added. The group delay of the filter is compensated by the transmit / receive timing processing, and the filter memory length is included in the prefix coverage. The frequency-to-time domain conversion uses a defined amplitude normalization method, and power calibration is based on the complete transformation and RF link. Software variable values ​​cannot be directly taken as antenna power, nor can the upper limit of twice the average power of a single subcarrier be interpreted as the peak-to-average power ratio of the entire CP-OFDM time-domain waveform not exceeding 2. The prefixed sample is transmitted from the antenna after digital-to-analog conversion, up-conversion, and RF amplification. The receiver uses the same configuration to remove the prefix, performs frequency domain transformation and equalization, and then recovers the data according to the modulation and coding of each resource.

[0072] Step S5: The base station checks new valid channel records at 5ms intervals. It subtracts the new reference signal-to-noise ratio (SNR) in dB from the corresponding resource value in the snapshot of the already effective configuration, taking the magnitude of the change. The preset channel update threshold is 3dB. If the change in the reference SNR of any subcarrier is strictly greater than 3dB, the transmission configuration is re-evaluated and updated; if the change is not greater than 3dB, the original effective configuration is retained. The comparison benchmark is replaced after the new configuration is confirmed to be effective. Ineffective trial configurations cannot cover the benchmark; otherwise, a state not yet used for transmission might be taken as a stable state. 5ms represents the detection period and is not equal to the total delay from channel change to the new configuration taking effect.

[0073] Total latency also includes the time for measurement accumulation, feedback transmission, solution processing, control confirmation, and waiting for the effective frame. For example, if a channel change occurs 0.2ms after a monitoring, there is still a 4.8ms wait before the next 5ms monitoring. If feedback, calculation, and confirmation take a total of 0.6ms, and the effective frame still needs to wait 0.3ms, then the update latency in this timing set is 5.7ms. When new measurements indicate that the old configuration no longer meets the reliability threshold, the old configuration must not continue to be used indefinitely because the new table has not yet been confirmed. The base station suspends associated data, maintains the permitted control and measurement processes, and waits for a new feasible configuration. Invalid feedback and unknown states cannot be used as evidence that quality has been restored. The block error rate is calculated based on the user equipment's confirmation results of the transport blocks.

[0074] When 12 out of the last 100 valid initial transmission blocks have negative acknowledgments, the block error rate is 0.12. When the threshold is 0.10, the upper limit of candidate modulation for the corresponding resource is limited, and the threshold table with added redundancy is used to recalculate the result. Individual measurement of the bit error rate requires the receiver to possess known test bits or an equivalent comparable sequence from the transmitter. The receiver compares bits one by one and returns the number of erroneous bits and the total number of compared bits, which the base station uses to calculate the bit error rate. For ordinary service blocks without known reference content, the number of erroneous bits cannot be determined solely by check failures.

[0075] The baseband processor reads measurement records via the feedback receiving interface and accesses the threshold table, candidate resource table, current configuration, and pending configuration in the memory via the bus. The baseband processor's scheduling process outputs candidate resources, modulation processing outputs the power lower bound, power processing outputs the allocated power, and signal generation processing reads the verified configuration to form time-domain samples. The feedback receiving interface and the RF front-end respectively serve as the feedback input and transmit output. The configuration identifier in the memory connects the two directions, preventing multiple software modules on the same processor from having only names without data relationships.

[0076] Example 1: In this example, steps S1 to S5 are used to support single-user basic transmission. Among the 20 candidate data subcarriers, the linear reference signal-to-noise ratio (SNR) of subcarriers 1 to 8 is 100, corresponding to 20 dB; 9 to 14 is 10, corresponding to 10 dB; and 15 to 20 is 1.584893, corresponding to 2 dB. Each window contains 10 records of the same quality, therefore the average result remains unchanged. At a reference power of 1 mW, the equivalent noise reduction values ​​for the first three groups are 0.01 mW, 0.1 mW, and 0.630957 mW, respectively.

[0077] The first three groups can be formed by channel power gains of 0.1, 0.01, and 0.001584893, respectively, and a common equivalent noise power of 0.001mW. There are 14 resource limits, with sufficient buffer data. The first to 14 reference signal-to-noise ratios are all above 8dB, indicating high initial screening quality, with candidate modulations starting from 64QAM. The 15th to 20th are below 3dB and are not currently used. The cyclic prefix is ​​selected from 120 points based on a maximum effective delay of 5μs and a margin. The minimum linear signal-to-noise ratio for 64QAM is 200. The lower power limits for the first to 8 are 2mW each, and for the 9th to 14th, they are 20mW each. The last 6 exceed the 2mW limit; after reducing to 16QAM, the lower limit is 5mW each, still exceeding the limit; further reducing to QPSK, the lower limit is 1mW each. At this point, the minimum total power is 8 x 2mW plus 6 x 1mW, resulting in 22mW, exceeding the 14mW budget. When subcarriers 1 through 8 are downgraded from 64QAM to 16QAM, each subcarrier releases 1.5mW and reduces 2 bits, with each bit reduction corresponding to a release of 0.75mW. Subcarriers 9 through 14 are already at minimum modulation and are not considered for further downgrading. Subcarriers 8, 7, 6, 5, 4, and 3 are downgraded in descending order of their numbers. The minimum total power after each downgrading is 20.5, 19, 17.5, 16, 14.5, and 13mW, respectively. Subcarriers 1 and 2 retain 64QAM, subcarriers 3 through 8 use 16QAM, and subcarriers 9 through 14 use QPSK.

[0078] When the water injection level is 0.676666667mW, the first and second water injection terms are constrained by both the lower and upper bounds of 2mW, and are each allocated 2mW. The water injection terms from the third to the eighth are 0.666666667mW, which is higher than the lower bound of 0.5mW, and are each allocated 0.666666667mW. The water injection terms from the ninth to the fourteenth are 0.576666667mW, which is lower than the lower bound of 1mW, and are each allocated 1mW. The total is 4mW plus 4mW plus 6mW, which equals 14mW. Based on the above allocation, the signal-to-noise ratios after allocation are 200, 66.666667, and 10.

[0079] Based on the approximate bit error rate relationship of the square constellation mentioned above, the estimated bit error rates are 0.000591567, 0.000097774, and 0.000782701, respectively, all three of which meet the target of 0.001. The single symbol load is 2 x 6 bits + 6 x 4 bits + 6 x 2 bits, resulting in 48 bits, which is higher than the minimum requirement of 44 bits. The effective symbols plus the 120-point prefix total 1144 points, sampled and transmitted at 15.36MHz, corresponding to 74.4791667μs. Calculated only for data symbols and prefixes, the load rate is 644475.524 bits / s. The net service overhead of pilot, control, coding, and retransmission is not deducted from this value.

[0080] The linear reference signal-to-noise ratio (SNR) of the first subcarrier changes from 100 to 25, corresponding to a change from 20 dB to 13.9794 dB, a change of 6.0206 dB, triggering an update. The equivalent noise reduction changes from 0.01 mW to 0.04 mW. The lower bound required for 64QAM is 8 mW, exceeding the 2 mW upper limit, so it is reduced to 16QAM with a lower bound of 2 mW. The same sorting and budget check are re-executed. Finally, the first subcarrier uses 16QAM and is allocated 2 mW, the second uses 64QAM and is allocated 2 mW, the third to eighth subcarriers each use 16QAM and are allocated 0.666666667 mW, and the ninth to fourteenth subcarriers each use QPSK and are allocated 1 mW. After the update, the SNR of the first subcarrier after allocation is 50, and the estimated bit error rate is 0.000587026. The total load is 4 + 6 + 24 + 12, resulting in 46 bits. All resources meet the thresholds, still satisfying the minimum 44-bit load requirement. In this calculation, the reliability threshold and total power budget are checked separately. The average bit error rate estimate obtained by weighting the load bits of each group cannot replace the per-resource threshold; the system requires each reserved resource to pass the threshold, not just the average value.

[0081] Maintaining the same 1mW reference power and 2mW power limit, the channel inputs are changed to 10dB for the first 8, 5dB for the next 6, and 2dB for the remaining 6 to examine the poor channel boundary. The equivalent noise reduction for the first 8 is 0.1mW, for the next 6 it is 0.316227766mW, and for the remaining 6 it is 0.630957mW. Even with 2mW allocated to the first 8, the signal-to-noise ratio after linear allocation is only 20, which does not meet the 16QAM threshold of 50, so QPSK must be used. The power required for the last 6 using QPSK is 3.16227766mW, exceeding the 2mW limit, and therefore not feasible. In the diagnostic branch that allows the removal of infeasible resources, the first 8 resources are retained, and the fixed 2mW limit is maintained for each solution. After the 14mW is evenly distributed across 8 equivalent channels, each channel receives 1.75mW. The signal-to-noise ratio after linear distribution is 17.5, the estimated QPSK bit error rate is 0.0000143654, and the payload is only 16 bits. 16 bits is lower than the minimum requirement of 44 bits in this implementation, therefore the output service requirement is deemed infeasible, and this diagnostic configuration is not considered the final configuration to meet the requirements.

[0082] Example 2: This example, based on the signal generation and modulation power processing of Example 1, adds a two-user mutually exclusive resource allocation. The two users do not simultaneously carry independent data within the same subcarrier and the same symbol period, thus avoiding the introduction of non-orthogonal transmissions superimposed at the same location.

[0083] User equipment A has a linear reference signal-to-noise ratio (SNR) of 100 on resources 1 through 7 and a reference SNR of 1.584893 on resources 8 through 14. User equipment B has a reference SNR of 1.584893 on resources 1 through 7, 100 on resource 8, and 10 on resources 9 through 14. Resources 15 through 20 are of low quality for both users.

[0084] The base station allocates resources 1 through 7 to user equipment A and resources 8 through 14 to user equipment B. After each user's resource allocation status is entered into the same resource table, it forms the same valid inputs of 8 linear 100s and 6 linear 10s as in Example 1. User A's minimum load requirement is 30 bits, and user B's is 14 bits; user A's power budget is 8mW, user B's is 7mW, and the base station's total data budget remains 14mW. The user power budgets are upper limits and cannot be added together to replace the base station's total budget.

[0085] According to the calculations in Example 1, User A loads 6 bits in the first and second resources respectively, and 4 bits in the third to seventh resources respectively, for a total load of 32 bits. User A's power is 4mW plus 5 multiplied by 0.666666667mW, resulting in 7.333333335mW, which does not exceed 8mW. User B loads 4 bits after the eighth resource is downgraded from 64QAM to 16QAM, and 2 bits in the ninth to fourteenth resources respectively, for a total load of 16 bits. User B's power is 0.666666667mW plus 6mW, resulting in 6.666666667mW, which does not exceed 7mW. Calculated using precise fractions, the total power of the two users is 14mW, and the total load is 48 bits.

[0086] The complex symbols of A and B are written to different frequency domain positions, and then a unified inverse Fourier transform is performed. The receiver only extracts the resources allocated in the configuration table. The prefix is ​​taken as the larger value of the effective channel memory requirements of the two users; if the maximum relative delay of the two users does not exceed 5μs, the 120-point prefix continues to be used. If a user's budget is further reduced, the base station cannot hide the excess power in the budget of another user, but should fix the available power of the excess user and re-perform step-down and water-filling, and check the minimum loading requirements for each user. During resource conflict handling, the base station maintains a unique allocation mark for each time-frequency position. When candidate users contend for the same resource, users who have not yet met the minimum requirements are checked first, and then the power cost of related users on this resource is compared; after allocation, the remaining requirements of users are updated, and the already allocated positions are not re-allocated to another user. A's minimum requirement of 30 bits is met with a 32-bit loading, and B's minimum requirement of 14 bits is met with a 16-bit loading. The total requirement of 44 bits is not amplified by inter-user duplication counting.

[0087] Example 3: This example follows the same basic process as Example 1, except that the timestamped linear signal-to-noise ratio (SNR) prediction values ​​are input into the same power solver, and a 2ms monitoring period and a 2dB update threshold are used. The linear reference SNRs of the first subcarrier at 20ms, 22ms, and 24ms are 100, 75, and 50, respectively. The difference between the two most recent equally spaced records is used as the trend, decreasing by 25 between 22ms and 24ms, with a slope of 12.5 decreases per millisecond. The configuration plan takes effect at 26ms, with a prediction lead of 2ms. Starting from the 24ms record of 50, subtracting 2 times 12.5 yields a predicted linear reference SNR of 25 at 26ms.

[0088] When the reference power is 1mW, dividing the unit reference transmit power by the predicted linear reference signal-to-noise ratio (SNR) of 25 yields a predicted equivalent noise level of 0.04mW. With other resources remaining in the state of Example 1, the solution is: 1st 16QAM, 2mW; 2nd 64QAM, 2mW; 3rd to 8th 16QAM, 0.666666667mW each; 9th to 14th QPSK, 1mW each, for a total load of 46 bits. If a linear SNR of 50 is used directly at 24ms, dividing the unit reference transmit power by the linear SNR of 50 yields an equivalent noise level of 0.02mW. Multiplying the minimum linear receive SNR of 50 for 16QAM by the equivalent noise level gives a lower bound of 1mW for the transmit power of the first subcarrier.

[0089] Using the same budget, we obtain 2mW each for the 2nd and 3rd 64QAM, 0.6mW each for the 4th to 8th 16QAM, 1mW each for the 9th to 14th QPSK, and 1mW for the 1st, for a total power of 14mW. When the actual linear reference SNR reaches 25 according to the calculated sequence in 26ms, the SNR after the first allocation in the old measurement configuration is 25, corresponding to a 16QAM bit error rate estimate of 0.00950524, which is higher than 0.001. The SNR after the first allocation in the prediction configuration is 50, corresponding to 0.000587026. The prediction configuration eliminates one threshold violation resource under this input group. The prediction span is determined by the feedback timestamp, calculation time, and effective time. If the allowable span is exceeded, there is insufficient recording, or the prediction results in non-positive quality, the latest valid measurement is used and the upper limit of the candidate modulation is limited, and a pause is taken if necessary.

[0090] Example 4: Based on the signal generation, power allocation and update mechanism of Example 3, this example further introduces frequency domain coherent block power coordination, rapid location of discrete water injection breakpoints and adaptive transmission based on multi-dimensional feature degraded decision factors.

[0091] All subsequent processes use the same channel state snapshot identifier. Measurement feedback, delay power spectrum, candidate resources, power boundaries, and the final transmission configuration table are associated through this snapshot identifier. Wireless channels exhibit frequency-selective fading in the frequency domain, and the fading characteristics of adjacent subcarriers may be correlated. If the modulation order and clipping water filling are calculated completely independently for each subcarrier, the base station needs to transmit a large number of per-subcarrier configurations to the user equipment. When the signal-to-noise ratio is close to the discrete modulation switching threshold, adjacent subcarriers may also produce different modulation results due to minor measurement disturbances. To determine the frequency boundaries that allow joint configuration, after receiving the measurement feedback generated by the user equipment based on the downlink pilot in step S1, the base station's baseband processor reads the relative delays of each effective path from the power delay information. and corresponding path power . For path index, take values ​​from 0 to... Subtract one, This represents the total number of valid paths. Path power is a non-negative value from the same measurement snapshot and the same power reference plane; after removing records below the path detection threshold, the power of the remaining paths is normalized to ensure all... The sum of them equals 1.

[0092] The baseband processor calculates the normalized frequency domain correlation coefficient based on the time-delay power spectrum. Specifically, for a given non-negative frequency interval, the first... Normalized path power of each path Multiplying by the complex phase term and summing over all valid paths, the phase of the complex phase term equals -2π, the frequency interval, and the relative time delay. The product of the three It represents the imaginary unit.

[0093] The absolute value of the obtained complex number is the normalized frequency domain correlation coefficient at the frequency interval; the correlation coefficient at the zero frequency interval is equal to 1. When the correlation threshold is set to 0.5, starting from the zero frequency interval, the correlation coefficient is checked sequentially with the subcarrier interval as the sampling step size. The maximum frequency span where the absolute value of the correlation coefficient is continuously not lower than 0.5 is determined as the coherent bandwidth. When adjacent sampling points cross 0.5, the boundary can be determined after linear interpolation; the discontinuous interval formed by the correlation curve rising again is not included in the coherent bandwidth starting from zero. Assume the subcarrier interval is 15kHz. If the coherent bandwidth obtained from the current channel snapshot is 180kHz, then the number of complete subcarrier intervals covered by the coherent bandwidth is 12. The baseband processor scans candidate data subcarriers from low frequency to high frequency. Only when the frequency positions of the candidate data subcarriers are continuous, the total span of the current aggregation block does not exceed 180kHz, and the difference between the maximum and minimum values ​​of the channel power gain in dB within the block is less than 0.4dB, will the corresponding candidate data subcarriers be included in the same subcarrier aggregation block. If any condition is not met, the current aggregation block is terminated and a new block is established from the next candidate data subcarrier, thereby avoiding the aggregation block from crossing the coherent bandwidth simply because the threshold is met by two adjacent points.

[0094] In Example 1, among the 20 candidate data subcarriers, the linear reference signal-to-noise ratio (SNR) of subcarriers 1 to 8 is 100, and the channel power gain is 0.1, satisfying the continuous flatness condition and forming the first subcarrier aggregation block; the linear reference SNR of subcarriers 9 to 14 is 10, and the channel power gain is 0.01, forming the second subcarrier aggregation block. The frequency span of each aggregation block does not exceed 12 complete subcarrier intervals.

[0095] Subcarriers 15 through 20 were not included in the candidate data subcarrier set and did not participate in this aggregation. Candidate data subcarriers within the aggregation block are configured with the same candidate modulation scheme. To prevent edge subcarriers within the block from falling below the reliability threshold due to channel estimation errors, the baseband processor takes the largest equivalent noise reduction within the subcarrier aggregation block as the block reference equivalent noise reduction. The lowest linear received signal-to-noise ratio corresponding to the selected modulation scheme within the block is multiplied by the block reference equivalent noise reduction to obtain the lower bound of the transmit power used by each candidate data subcarrier within the block. The block reference equivalent noise reduction for the first subcarrier aggregation block is 0.01mW, and for the second subcarrier aggregation block it is 0.1mW. If the first block uses 64QAM, the lower bound of the transmit power for each subcarrier is 2mW; even if the second block is downgraded to QPSK, the lower bound of the transmit power for each subcarrier is still 1mW, resulting in a total lower bound of 22mW, exceeding the 14mW budget. Since the modulation within the block must be consistent, the first block is reduced to 16QAM, the lower limit of each subcarrier becomes 0.5mW, and the sum of the lower limits becomes 10mW, entering the feasible water-filling range.

[0096] When performing amplitude limiting water injection in step S3, each subcarrier aggregation block participates in the water injection level search as a whole variable. For a given water injection level, the limited power corresponding to each subcarrier within the block is first calculated using the common transmit power lower bound of the block, the block's reference equivalent noise reduction, and the upper limit of each subcarrier's power. Then, the total allocated power of the block is obtained by summing these values. Under the conditions of a total data transmit power budget of 14mW and a subcarrier power upper limit of 2mW, the total power of the first block is calculated to be 8mW, and the total power of the second block is calculated to be 6mW. Within the block, a common transmit power lower bound is first configured for each subcarrier. Then, the remaining part after deducting the sum of the common transmit power lower bounds from the total allocated power of the block is finely adjusted according to the proportion of the actual equivalent noise reduction of each subcarrier to the sum of the equivalent noise reductions within the block. If a subcarrier reaches the 2mW upper limit, the subcarrier power is fixed at 2mW, and the weights of the remaining subcarriers with remaining power space and not reaching the subcarrier power upper limit are re-normalized according to the proportion of their respective equivalent noise reductions to the sum of the equivalent noise reductions of these unsaturated subcarriers. The equivalent noise reduction values ​​within the first and second blocks are equal, and the fine-tuning weights degenerate into a proportional allocation. Each subcarrier in the first block receives 1mW, and each subcarrier in the second block receives 1mW. All power is not lower than the lower bound of the common transmit power and does not exceed the upper bound of the subcarrier power, totaling 14mW. The first block uses 16QAM to carry 32 bits, and the second block uses QPSK to carry 12 bits, totaling 44 bits, meeting the minimum loading requirements.

[0097] The limited-amplitude water injection is a piecewise linear continuous mapping with respect to the water injection level. The slope changes when the subcarrier moves away from the lower bound of the transmit power or reaches the upper bound of the subcarrier power. For the first... One candidate data subcarrier, the baseband processor will set the lower bound of the transmit power. Equivalent noise reduction The sum is used as the lower limit to break the breakpoint, and the upper limit of the subcarrier power is set. Equivalent noise reduction The sum is used as the upper limit saturation breakpoint.

[0098] Summary Two types of breakpoints in each candidate data subcarrier are identified, duplicate points are removed, and the breakpoints are sorted in ascending order of numerical value, resulting in a breakpoint count not exceeding [number missing]. The discrete breakpoint vector. For 14 candidate data subcarriers, with no more than 28 breakpoints, the target adjacent breakpoint interval can be determined by binary search of the breakpoint index, requiring a maximum of 5 interval judgments. Within the interval formed by any two adjacent breakpoints, the baseband processor divides the candidate data subcarriers into three non-overlapping subsets:

[0099] The first subcarrier subset contains subcarriers that have reached the upper limit of subcarrier power, with the power fixed at the corresponding upper limit; the second subcarrier subset contains subcarriers that are still at the lower limit of transmit power, with the power fixed at the corresponding lower limit; the third subcarrier subset contains subcarriers in the power linear adjustment region, with the power equal to the water level minus the corresponding equivalent noise reduction.

[0100] The total allocated power within the specified interval is equal to the sum of the upper bound power of the first subcarrier subset, the sum of the lower bound power of the second subcarrier subset, and the sum of the water level of the third subcarrier subset. Therefore, when the third subcarrier subset is not empty, the total data transmission power budget can be reduced by the sum of the two fixed power values, added to the sum of the equivalent noise reduction of the third subcarrier subset, and then divided by the number of subcarriers in the third subcarrier subset to obtain the target water level in one step.

[0101] If the third subcarrier subset is empty, the total allocated power of the current interval does not change with the water level. At this point, the baseband processor terminates the water level search early and calculates the remaining unallocated power after deducting the sum of the upper limit power of the first subcarrier subset and the sum of the lower limit power of the second subcarrier subset from the total data transmit power budget. If the remaining power is positive and the second subcarrier subset is not empty, safe injection is performed according to the equivalent noise reduction weights of each subcarrier in the second subcarrier subset; subcarriers that have reached their subcarrier power upper limit are removed from subsequent allocations, and the remaining weights are normalized. If the second subcarrier subset is empty, or the remaining power is greater than the total remaining power space of the second subcarrier subset, the portion that cannot be further allocated is retained as unused budget, without exceeding any subcarrier power upper limit. A negative remaining power indicates that the current discrete modulation and power boundary are infeasible, and modulation de-leveling is returned without performing reverse allocation.

[0102] Regarding the demotion mechanism, Implementation 1 uses the power release amount corresponding to the reduction in loaded bits per unit as the ranking index. To further distinguish candidate resources with the same power release amount, this implementation constructs a demotion priority decision factor. (Also known as the multidimensional benefit decision factor), it integrates power release efficiency, equivalent noise relative severity, and time-domain fluctuations in channel power gain. The demotion priority decision factor is calculated according to the following formula:

[0103] (2)

[0104] In formula (2): the first Current modulation scheme of candidate data subcarriers Corresponding lower limit of transmit power Adjacent lower modulation mode Corresponding lower limit of transmit power The number of load bits reduced by downgrading Equivalent noise reduction Channel frequency domain response statistics and Preset weighting factors and ;

[0105] Output: The Demotion priority decision factor for each candidate data subcarrier The larger the value, the higher the priority of demotion; in the formula, For the first The de-priority decision factor for each candidate data subcarrier, in mW / bit; For the first The current candidate modulation scheme for each candidate data subcarrier; For the first Each candidate data subcarrier is adjacent to a candidate modulation scheme of a lower level; For the first Each candidate data subcarrier in the current candidate modulation scheme The corresponding lower bound of the transmit power, in mW; For the first Each candidate data subcarrier is downgraded to the next lower candidate modulation scheme. The corresponding lower bound of the transmit power, in mW; For the first The number of loaded bits reduced by downgrading each candidate data subcarrier, in bits; For the first The equivalent noise reduction of each candidate data subcarrier, in mW; This is the set of indices for all currently degradeable candidate data subcarriers; For set The Middle The equivalent noise reduction of each candidate data subcarrier, in mW; The channel severity penalty weighting factor is a dimensionless constant ranging from 0.1 to 0.5. The channel time-domain fluctuation penalty weighting factor is a dimensionless constant ranging from 0.05 to 0.2. For the first The complex channel frequency response of each candidate data subcarrier within the sliding window is a dimensionless complex number. For the first The variance of the channel power gain of each candidate data subcarrier within the sliding window is a dimensionless quantity. For the first The average channel power gain of each candidate data subcarrier within the sliding window is a dimensionless quantity that is greater than zero. The square of the average channel power gain is a dimensionless quantity greater than zero. Represents a set The index summation in the table; except for the demotion priority decision factor, the lower bound of each transmit power, the number of loaded bits and each equivalent noise conversion quantity which have explicit dimensions, the other parameters and weighting factor terms are dimensionless quantities.

[0106] Each time a degrade is executed, the baseband processor updates the set. The lower bound and decision factor. The equivalent noise term in the parentheses of formula (2) gives higher de-ranking priority to subcarriers with larger noise reduction and lower reception quality per unit power; the time-domain fluctuation term uses positive penalty to give priority to de-ranking of subcarriers with larger channel power gain fluctuation, avoiding the continued occupation of a large reliability margin with high-order modulation. The base station follows Sort by size from largest to smallest; If the subcarrier numbers are the same, the descending order rule of Example 1 shall be used.

[0107] To verify the ranking effect of the demotion priority decision factor independently, the common modulation constraint of the aggregated block is temporarily disabled, and the 1st to 8th demotionable subcarriers from Example 1 are used. The equivalent noise reduction of each subcarrier is 0.01mW, and the initial set... The sum of the equivalent noise reduction is 0.08mW; when 64QAM is reduced to 16QAM, the power release corresponding to the reduction in the number of loaded bits is 0.75mW / bit.

[0108] Let the normalized variance of the channel power gain of the 1st to 4th subcarriers be 0.25, and the normalized variance of the 5th to 8th subcarriers be 0.01. Take 0.3, The value is set to 0.1. In the initial decision calculation, the decision factor for the 1st to 4th high-fluctuation subcarriers is 0.75 multiplied by one, plus 0.0375, plus 0.025, resulting in 0.796875 mW / bit; the decision factor for the 5th to 8th stable subcarriers is 0.75 multiplied by one, plus 0.0375, plus 0.001, resulting in 0.778875 mW / bit. Therefore, the 1st to 4th high-fluctuation subcarriers enter the de-sequence sequence first. Each de-sequence is executed and the set is dynamically updated. At this time, the relative decision factor difference between high-fluctuation subcarriers and stable subcarriers remains greater than zero, and they are arranged in descending order of subcarrier number within the same fluctuation level. Therefore, when six de-electrons are required, the multi-dimensional decision processes the 4th, 3rd, 2nd, 1st, 8th, and 7th subcarriers sequentially, allowing the 5th and 6th stable subcarriers to retain higher-order modulation; in Example 1, only the 8th to 3rd subcarriers are processed sequentially in descending order of number, which allows the 1st and 2nd high-fluctuation subcarriers to retain higher-order modulation. The comparison illustrates that the decision direction is consistent with the goal of prioritizing de-electrons for high-fluctuation resources.

[0109] Comparative Example 1: This comparative example maintains the 20 candidate resources, 14 data resources, reference signal-to-noise ratio, total budget, subcarrier upper limit, prefix, reliability threshold, and transmission conditions of Example 1 unchanged. Only the relationship between candidate modulation generating a minimum power lower bound and returning power allocation is disabled. Conventional water injection without a modulation lower bound is used, and modulation is selected independently based on the allocated quality. The equivalent noise reduction for 8 resources is 0.01mW, and for 6 resources it is 0.1mW. All resources do not reach the 2mW upper limit. The water injection level is 14mW plus 8 times 0.01mW plus 6 times 0.1mW, divided by 14 to obtain 1.048571429mW. 1.038571429mW is allocated to each of the 1st to 8th resources, and 0.948571429mW is allocated to each of the 9th to 14th resources, for a total power of 14mW. The signal-to-noise ratios after allocation are 103.857143 and 9.485714, respectively. The first 8 are insufficient to support 64QAM with a linear threshold of 200, so they are reduced to 16QAM independently; the last 6 are reduced to QPSK but are still below the threshold of 10.

[0110] If the last 6 resources are still transmitted using the minimum modulation, the total nominal load is 8 x 4 + 6 x 2, resulting in 44 bits. Of these, 12 bits are located in resources that do not meet the threshold, leaving a threshold-compliant load of 32 bits. The estimated bit error rate (BER) of the last 6 bits is 0.00103539, higher than the target of 0.001. Dividing 44 bits and 32 bits by 74.4791667 μs yields a nominal load rate of 590769.231 bits / s and a threshold-compliant load rate of 429650.350 bits / s, respectively. The difference between the two arises from the power constraint imposed by the modulation lower bound, not from increasing resources, increasing total power, or changing the prefix. The threshold-compliant load is the sum of the bits carried by resources that meet the threshold, and does not equal the actual effective throughput after packet loss. If the conventional scheme also incorporates minimum power backhaul, recalculation, and failure handling, the relationship intentionally removed in this comparison has been restored.

[0111] Comparative Example 2: This comparative example maintains the initial input, power configuration, and subsequent trajectory of the first subcarrier changing linearly from 100 to 25, as in Example 1. Only the update action of resolving after the change exceeds the threshold is disabled, and the old transmission configuration table is continued. In the old configuration, the first subcarrier uses 64QAM and is allocated 2mW. After the change, the equivalent noise reduction is 0.04mW, and the signal-to-noise ratio (SNR) after allocation is 50. The above approximate bit error rate relationship for the square constellation gives a bit error rate estimate of 0.03582327 for 64QAM, which is higher than the target of 0.001, forming one threshold violation resource. The nominal load capacity in the old configuration remains 48 bits. The first resource carrying 6 bits does not meet the threshold, and the threshold-compliant load capacity is 42 bits. After resolving, the first subcarrier uses 16QAM, still allocated 2mW, and the SNR after allocation remains 50, but the load capacity becomes 4 bits and meets the threshold; other resources remain unchanged, therefore the threshold-compliant load capacity is 46 bits. A 42-bit threshold load rate corresponds to 563,916.084 bit / s, and a 46-bit threshold load rate corresponds to 617,622.378 bit / s. The update reduces the nominal load size while placing more load bits in resources that meet the threshold.

[0112] When multiple component carriers participate in transmission, carrier aggregation management reads the active list from the base station scheduler and the allowed aggregation combinations, maximum receive bandwidth, and number of simultaneous receptions from the user equipment capability configuration. Each component carrier maintains its own reference power, noise, data budget, and resource location, and uses the same power reference plane when comparing across component carriers.

[0113] When component carrier A has 14 available data resources and component carrier B has 4 spare data resources, if a certain data queue is marked as infeasible on carrier A, the carrier aggregation management reads the latest measurement and remaining budget of carrier B, confirms that the user equipment supports simultaneous reception by carrier A and carrier B, and adds the data that can be migrated to the candidate queue of carrier B.

[0114] The migration operation includes removing the data to be transmitted from the original resource table, establishing a new resource correspondence in the target resource table, recalculating the modulation and power of the target component carrier, and sending a receive configuration with the new configuration identifier to the user equipment. Only after the target configuration is confirmed and takes effect will the relevant data be transmitted from the target component carrier. The target component carrier has two candidate resources with equivalent noise reduction values ​​of 0.05mW and 0.1mW respectively, a reserved data budget of 2mW, and a power upper limit of 2mW for both. When the two resources use QPSK, the threshold 10 corresponds to lower limits of 0.5mW and 1mW, with a total lower limit of 1.5mW. When the fill level is 1.05mW, the allocated power is 1mW and 1mW, satisfying the 2mW budget. After allocation, the signal-to-noise ratios of the two target resources are 20 and 10, both reaching the QPSK threshold and capable of carrying 4 bits of uncoded data. If the data to be migrated requires 8 bits within this time limit, but the target resources can only provide 4 bits, then this migration is insufficient to meet the requirement, and partial migration cannot be directly registered as full service recovery. The component carrier budget is derived from the allocation of the base station's overall budget. The power released from the original resources and the power added to the target resources are registered separately before and after the migration; the target resource budget cannot be reused before the original component carriers have been released.

[0115] Interference detection reads co-channel interference measured by user equipment, base station measurement information, and neighboring cell coordination constraints. External co-channel interference is used to update the equivalent noise; coordination information from neighboring cells requiring the base station to limit its transmission is used to reduce the power limit. Since the control directions of the two inputs are different, reducing the base station's transmission power cannot be considered a universal method to reduce external interference at the receiver. For example, a resource originally has a power limit of 2mW, and coordination constraints reduce this limit to 0.8mW, resulting in an equivalent noise reduction of 0.1mW. The lower bound required for QPSK threshold 10 is 1mW, exceeding 0.8mW, therefore minimum modulation is also infeasible. The system outputs an infeasibility flag and attempts confirmed backup resources, but cannot forcibly allocate 1mW exceeding the coordination limit. For the same resource, only an increase in external interference changes the equivalent noise reduction from 0.1mW to 0.2mW, while the base station's transmission power remains at 1mW, resulting in a signal-to-noise ratio (SNR) dropping from 10 to 5 after allocation. Further reducing the transmission power to 0.8mW will cause the SNR to drop to 4 after allocation, failing to restore reception quality. At this point, allow for power increases, resource migrations, or branch pauses.

[0116] To more intuitively illustrate the differences between Example 1 and Comparative Examples 1 and 2 in terms of the impact of modulation lower bound constraints, amplitude limiting water injection, and channel update closed loop on threshold qualified loading capacity and transmission reliability, the comparison results of the three are summarized in Table 1.

[0117] Table 1:

[0118] plan Key issues nominal loading Qualified loading Does it meet the 44-bit requirement? Example 1 Initial -2:64QAM; 3-8:16QAM; 9-14:QPSK 48bit 48bit yes Comparative Example 1 The last 6 QPSKs have an SNR of 9.49 (<10), indicating that the BER exceeds the limit. 44bit 32bit no Comparative Example 2 is not updated The first 64QAM, SNR=50<200, BER exceeds the standard. 48bit 42bit no After the update The first one was reduced to 16QAM. 46bit 46bit yes

[0119] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0120] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A base station signal transmission method based on multi-carrier adaptive transmission, characterized in that, Includes the following steps: Receive measurement feedback generated by the user equipment based on the downlink pilot, and form a channel state information matrix including reference signal-to-noise ratio, channel power gain, equivalent noise power and multipath delay; The candidate data subcarrier set, cyclic prefix length, and candidate modulation scheme are determined based on the channel state information matrix. The lowest received signal-to-noise ratio (SNR) corresponding to the candidate modulation scheme is converted into the lower bound of the transmit power. The lower bound of the transmit power, the upper limit of the subcarrier power, and the total data transmit power budget constraint limit the water injection. Based on the allocated SNR, the candidate modulation scheme is verified. For infeasible configurations, only modulation downgrading is performed in the fixed channel state. Priority is given to selecting the downgrading object with a larger power release corresponding to the unit load bit reduction and redistributing the power until the power and modulation constraints are met or no feasible configuration is determined. The transmission configuration table or infeasible mark is output. According to the transmission configuration table, the data to be transmitted is modulated, mapped, inverse Fourier transformed, and cyclic prefix added. The multi-carrier signal is then transmitted through the radio frequency front end. If no feasible configuration is available, the corresponding data transmission is paused. The newly acquired channel state is periodically compared with the channel state corresponding to the effective transmission configuration table. When the change exceeds the update threshold, the candidate data subcarrier set, cyclic prefix length, modulation scheme and transmit power are re-determined, and the updated configuration that passes the verification is used for the transmission of the next frame.

2. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 1, characterized in that, User equipment performs least-squares channel estimation at the downlink pilot position, interpolates the channel frequency response at the pilot position to form the channel frequency response at the data subcarrier position; the measurement feedback includes the channel frequency response, equivalent noise power, pilot reference power, measurement timestamp and multipath delay; the base station normalizes the reference signal-to-noise ratio according to the pilot reference power, and converts the average of the linear reference signal-to-noise ratio in the most recent 10 effective symbol periods into a decibel value.

3. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 2, characterized in that, The base station marks subcarriers with an average reference signal-to-noise ratio (RSR) higher than 8dB as high quality, subcarriers with an RRS of not less than 3dB and not more than 8dB as medium quality, and subcarriers with an RRS lower than 3dB as low quality. Candidate data subcarriers are selected according to channel quality and the data to be transmitted. 64-orthogonal amplitude modulation, 16-orthogonal amplitude modulation, and orthogonal phase shift keying are used as initial candidate modulation schemes, respectively. The candidate modulation schemes are determined as the transmission modulation schemes after being verified by the lowest received RRS.

4. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 1, characterized in that, The cyclic prefix length is determined by the maximum relative delay of the effective multipath, the receiving timing error margin, and the digital filter memory length, and rounded up to the number of sampling points. Subcarriers in the same multicarrier symbol use the same cyclic prefix length. Under 1024-point inverse Fourier transform and 15kHz subcarrier spacing, the short cyclic prefix contains 120 sampling points, and the long cyclic prefix contains 256 sampling points. When neither cyclic prefix can cover the effective channel memory, the corresponding configuration of transmission is suspended.

5. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 1, characterized in that, The base station forms an equivalent noise reduction value based on the ratio of equivalent noise power to channel power gain. It multiplies the minimum linear received signal-to-noise ratio corresponding to the candidate modulation scheme with the equivalent noise reduction value to obtain the lower bound of the transmit power. The minimum received signal-to-noise ratio is determined by the reliability threshold table corresponding to the modulation scheme, coding configuration and target error rate. When the lower bound of the transmit power exceeds the upper limit of the subcarrier power, the candidate modulation order is reduced.

6. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 5, characterized in that, The base station divides the total data transmission power budget by the number of candidate data subcarriers at the start of the current solution to obtain the reference average power, and sets the subcarrier power upper limit to twice the reference average power. Within one solution, the subcarrier power upper limit is fixed, and the difference between the water level and the equivalent noise reduction is truncated by upper and lower bounds. The water level is adjusted by binary search so that the sum of the allocated power does not exceed the total data transmission power budget.

7. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 6, characterized in that, When the sum of the lower bounds of the transmit power exceeds the total data transmit power budget, the demotion object is selected according to the power release amount corresponding to the reduction in unit loaded bits, and when the sorting values ​​are the same, the demotion object is selected in descending order of subcarrier number; in fixed channel state, only candidate modulation demotion is performed, and candidate modulation upmotion is not performed; If all retained subcarriers still cannot satisfy the minimum modulation scheme, an infeasibility flag is output, or if delayed transmission is allowed, the infeasible subcarriers are removed and the solution is resolved.

8. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 1, characterized in that, The transmission configuration table includes configuration identifier, effective frame number, user equipment identifier, subcarrier position, modulation method, coding configuration, transmit power, and number of cyclic prefix sampling points. Before the effective frame, the base station sends the transmission configuration table through pre-agreed control resources. After obtaining confirmation from the user equipment, the base station performs 1024-point inverse Fourier transform, cyclic prefix addition, digital filtering, digital-to-analog conversion, and radio frequency transmission in the effective frame.

9. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 1, characterized in that, The base station calculates the block error rate based on the transport block acknowledgment information returned by the user equipment, and divides the number of first transport blocks corresponding to valid negative acknowledgments by the total number of valid first transport blocks; When the block error rate exceeds a preset threshold, the upper limit of candidate modulation for the resources used by the associated transport block is reduced and coding redundancy is increased. Power is then reallocated using the reliability threshold table with increased coding redundancy. Bit error rate is independently calculated by comparing each known test bit.

10. The base station signal transmission method based on multi-carrier adaptive transmission according to claim 1, characterized in that, In multi-user transmission, the base station allocates non-overlapping time-frequency resources based on the channel state information matrix and minimum load requirements of each user equipment. The same subcarrier is allocated to one user equipment within the same symbol period. Modulation and power calculation are performed within the total data transmit power budget of the base station and the corresponding power budget of the user equipment. After writing the frequency domain symbols of each user equipment into different resource grid positions, a unified inverse Fourier transform is performed.