Bridge method and system for inference and assurance of reliable communication for massive connectivity

Abstract

The application discloses a bridging method and system for inference and guarantee of reliable communication for large-scale connection, comprising: jointly inferring an active terminal set and a channel thereof from pilot observation, and generating posterior samples and confidence information; converting the posterior samples and the confidence information into provable signal power confidence lower bound and residual term upper bound, forming an equivalent confidence lower bound of signal-to-interference-and-noise ratio; combining an interference power consistent upper bound generated by spatial regularization condition, and outputting a reliability certificate. The application upgrades the system from "higher performance" to "provable, controllable and auditable performance".

Description

Technical Field

[0001] This invention relates to the field of wireless communication and artificial intelligence, and in particular to a bridging method and system for inference and guarantee of reliable communication for large-scale connections. Background Technology

[0002] In massive machine-type communication (mMTC) scenarios, base stations need to simultaneously make closed-loop decisions on "whether the terminal is active, what the channel parameters are, and whether the target reliability is met" under conditions of extremely short pilot signals and strong interference.

[0003] On the one hand, existing access-side methods typically output statistical indicators such as detection accuracy and mean squared error. Even if the average performance is strong, it often remains at the level of being better on average, and it is difficult to provide usable and auditable reliability guarantees for a specific time slot or a specific link. When the statistical patterns of the training data are inconsistent with those of the real deployment environment, the algorithm may experience performance degradation, and the system lacks quantification and protection measures for the degree of degradation.

[0004] On the other hand, network-side reliability analysis often uses the Poisson point process (PPP) to characterize the randomness of node locations. Such models are easy to obtain results such as average success rate, but they naturally describe the average performance of typical links and are difficult to give a direct guarantee that all links are valid at the same time. More importantly, when spatially controlled conditions are lacking, the total interference is often difficult to be uniformly bounded in almost everywhere, which leads to a lack of provable basis for the reliability lower bound that is valid for all links.

[0005] Therefore, what is lacking in engineering is a cross-layer bridging mechanism: one that can leverage the advantages of generative inference under complex channel distributions while injecting inference errors and uncertainties into the reliability lower bound framework in a computable and provable manner, enabling the system to not only provide estimation results but also output reliability certificates that can be used for admission, scheduling, and retransmission. This invention proposes an "inference and guarantee bridging" scheme to address this gap. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a bridging method and system for inference and assurance of reliable communication for large-scale connections.

[0007] The objective of this invention is achieved through the following technical solution: a bridging method for inferring and guaranteeing reliable communication, comprising:

[0008] The active terminal set and its channel are jointly inferred from pilot observations, and posterior samples and confidence information are generated.

[0009] The posterior samples and confidence information are transformed into a provable lower bound for signal power confidence and an upper bound for residual terms, forming an equivalent lower bound for signal-to-interference-plus-noise ratio confidence.

[0010] Based on the upper bound of the interference power generated by the spatial regularity conditions, a reliability certificate is output.

[0011] Furthermore, the active terminal set and its channel are jointly inferred from pilot observations, and posterior samples and confidence information are generated, including:

[0012] Fractional generative posterior inference models based on pilot observations With pilot set Input, output activity information and channel point estimation for each terminal (i.e., posterior sample), and calculate the error confidence radius. (i.e., confidence information).

[0013] Furthermore, the transformation of the lower confidence bound of the signal power is as follows:

[0014] Let receive the merged vector From the estimated channel The actual effective signal amplitude is constructed as follows: ,in, For modulo operations on complex numbers, superscript This indicates the conjugate transpose. This is the uplink channel vector of the k-th terminal;

[0015] In confidence events Using the triangle inequality and the Cauchy-Schwarz inequality, we have:

[0016] ;

[0017] in It is the vector norm (the default is the Euclidean norm). Let the error confidence radius be . For confidence level parameters;

[0018] Thus, the lower bound of the signal power in the confidence sense is obtained:

[0019] ,in It is a non-negative truncation operator.

[0020] Furthermore, the upper bound of the residual term is:

[0021] ;

[0022] Where α1, α2 ≥ 0 are coefficients. Let be the error confidence radius, where Here, K represents the terminal index, and K represents the total number of candidate terminals. This represents the confidence level parameter.

[0023] Furthermore, let This represents the actual external interference power of link k in the current time slot. Given the thermal noise power, provide a lower bound for the signal power under the confidence event. and with Characterizing the upper bound of the residual term, the lower bound of the equivalent signal-to-interference-plus-noise ratio confidence is:

[0024] .

[0025] Furthermore, the upper bound of the interference power generated by the spatial regularity condition is consistent. Thus, a computable and auditable lower bound is defined under space regularization:

[0026] ;

[0027] Thus, a reliability certificate is defined:

[0028] ;

[0029] Where θ is the reliability threshold. This represents the confidence level parameter.

[0030] Furthermore, let Φ denote the definition in The stationary point process of all potential launch nodes on the [the network]; [denotes] For Center, radius The opening ball; if there exist finite constants σ≥0, ρ≥0, ν≥0 such that the following event occurs with probability 1:

[0031] ;

[0032] Then Φ is said to satisfy strong(σ,ρ,ν)-ball regulation; where (σ,ρ,ν) is called the regulation strength parameter. Indicates all With all The condition set is taken as the intersection, that is, the above inequality must hold for any center of the sphere and any radius (which holds almost everywhere in a probabilistic sense).

[0033] Let the path loss function be defined. It is a non-negative, bounded, monotonically non-increasing function, and satisfies and ;

[0034] At any receiving location Define the geometric shot-noise generated by the potential emitting node (the sum of distance decays generated by spatial point processes):

[0035] ;

[0036] in, Point representation process One of the node positions; The cutoff radius indicates that only the distance from the receiving position is considered. No more than The node contribution; Indicates all and node position To sum up, that is, to sum the results of the opening ball. The potential launch nodes within are stacked one by one;

[0037] When Φ satisfies the strong(σ,ρ,ν)-ball regulation, there exists a definite upper bound for all R>0:

[0038]

[0039] Let R→∞ to obtain the computable upper bound constant for shot-noise in the entire space:

[0040] ;

[0041] Thus for any have ;

[0042] Assume that the transmit power of each active transmitting node in the current resource block (RB) satisfies 0 ≤ P x ≤P max Then the aggregated interference power at any receiving point satisfies a consistent upper bound:

[0043] .

[0044] The present invention also provides a bridging system for inference and guarantee of reliable communication, comprising:

[0045] Generative inference layer: Jointly infers the set of active terminals and their channels from pilot observations, and generates posterior samples and confidence information;

[0046] Bridging layer: Transforms posterior samples and confidence information into a provable lower bound of signal power confidence and an upper bound of residual terms, forming an equivalent lower bound of signal-to-interference-plus-noise ratio confidence;

[0047] Guarantee layer: Based on the upper limit of the interference power generated by the spatial regularity conditions, output a reliability certificate.

[0048] The present invention also provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, is used to implement the above-described bridging method for inferring and guaranteeing reliable communication.

[0049] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described bridging method for inferring and guaranteeing reliable communication.

[0050] Compared with the prior art, the beneficial effects of the embodiments of the present invention are:

[0051] 1) Upgrading AI-driven posterior inference outputs to provable reliability certificates: Score-generative posterior inference models using pilot observations With pilot set Input, Output ,in As an indicator of the probability of activity, For channel point estimation, The error confidence radius; the bridging layer and the guarantee layer further calculate the computable and auditable lower bound under spatial regularization. And generate a certificate This enables the system to not only provide detection and estimation, but also to provide auditable reliability criteria;

[0052] 2) Standardization of cross-layer bridging interfaces: using error confidence radius Signal power confidence lower bound Upper bound of residual terms Upper bound consistent with interference power A small amount of scalar material is sufficient to connect to the guarantee layer, facilitating project deployment and reuse;

[0053] 3) Not dependent on averaged location model: It adopts strong sphere regularization conditions instead of PPP average analysis, which is more suitable for controlled deployments such as park private networks, factory networks, and warehouse networks;

[0054] 4) More robust to distribution changes: Changes in environmental statistical patterns will be reflected as greater posterior uncertainty and tighten certificates. The system can then automatically tighten access or increase protection resources to avoid imperceptible degradation.

[0055] 5) Multi-antenna natural compatibility: for The received and combined vector of the kth terminal has an effective signal amplitude determined by... The present invention constructs a universal lower bound for the inner product under the constraint of the error radius, thereby reducing uncertainty. The standardized form is used in the calculation of the lower bound of the signal power and the certificate. This lower bound of the inner product applies to any number of antennas M and any linear merging vector. All of these are valid and can be easily extended to more antenna configurations and different merging strategies. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating a bridging method for inferring and guaranteeing reliable communication, provided in an embodiment of the present invention. Detailed Implementation

[0058] The present invention will now be described in detail with reference to the accompanying drawings. Unless otherwise specified, the features of the following embodiments and implementations can be combined with each other.

[0059] The present invention provides a bridging method for inferring and guaranteeing reliable communication, see [link to relevant documentation]. Figure 1 It includes the following three steps (corresponding to the generative inference layer, bridging layer, and guarantee layer, respectively):

[0060] (1) Jointly infer the active terminal set and its channel from pilot observations, and generate posterior samples and confidence information;

[0061] (2) Transform the posterior samples and confidence information into a provable lower bound of signal power confidence and an upper bound of residual terms to form an equivalent lower bound of signal-to-interference-plus-noise ratio confidence;

[0062] (3) Combine the upper limit of the interference power generated by the spatial regularity conditions and output the reliability certificate.

[0063] Specifically, step (1) involves the base station obtaining the observation matrix during the pilot signal period:

[0064] (1)

[0065] Where K is the total number of candidate terminals. Indicates whether the k-th terminal is active. Let M be the uplink channel vector from the terminal to the M receiving antennas. Given a pilot sequence of length L, , This is the noise matrix (which may include thermal noise and modeling errors), with superscript... This indicates transpose.

[0066] The goal of generative inference layers is to... Joint recovery and Unlike simply providing point estimates, this invention emphasizes outputting posterior uncertainty for subsequent "guarantee" calculations.

[0067] For example, this embodiment provides a type of score-generating posterior inference model that can be directly implemented.

[0068] (Variable Organization and Conditional Posterior) Complex variables are solved into real vectors based on their real and imaginary parts. Define... ,in To extract the real part operator, To extract the imaginary part operator. The aggregated unknowns are:

[0069] (2)

[0070] Generative inference layer learns conditional posterior distribution fractional function and using fractional networks To approximate, where This is the time index during the sampling process.

[0071] (Training and Inference Implementation) The training phase employs denoising score-matching or its equivalent; the inference phase uses reverse diffusion sampling or predictor-corrector iteration to generate posterior samples. And extract accordingly , The uncertainty measure is S, where S is the posterior sample size.

[0072] (Extracting active probability and channel point estimation from posterior samples) Let the first... A posterior sample can be written as ,in For active samples, For the corresponding channel samples (by (Obtained by restoration).

[0073] The activity probability is then calculated using the posterior mean (Monte Carlo estimation):

[0074] (3)

[0075] If the system implementation requires a binary activity indicator, thresholding can be used:

[0076] (4)

[0077] in The threshold is configurable.

[0078] Channel point estimation uses the posterior mean. To align with active events, it is preferable to use a posterior mean. Conditional mean estimate for weights:

[0079] (5)

[0080] Where ε>0 is a minimal constant in the implementation, used to avoid numerical problems where the denominator is zero (when all samples are inactive). (It can be set as a zero vector or keep the previous time slot estimate, depending on the implementation).

[0081] (Standardized Output Interface) To ensure the input closure between the bridging layer and the guarantee layer, the generative inference layer provides a unified output for each terminal k:

[0082] (6)

[0083] in As an indicator of the probability of activity, For channel point estimation, Let be the error confidence radius.

[0084] In one implementation, the inference phase generates a posterior sample set. And calculated by equation (8) In another implementation, the model does not explicitly output the sample set, but instead directly regresses... (Or first output the covariance / confidence interval and then map it to obtain the result) Regardless of the implementation method used, the external interface is... .

[0085] (Definition and Realizable Calculation of Confidence Radius) The confidence radius is defined as satisfying:

[0086] (7)

[0087] in For probability operators, This is a configurable confidence level parameter. If a posterior sample set is output, a directly achievable calculation formula can be given using empirical quantiles:

[0088] (8)

[0089] The above definition uses the Euclidean norm as an example; in engineering implementation, the weighted norm can also be used to reflect antenna correlation or a priori structure, but without changing the interface form of the bridging layer.

[0090] Step (2) is as follows:

[0091] Transformation of the lower confidence bound of signal power:

[0092] Taking linear merging as an example (such as maximum ratio merging), let the receiving merge vector be... Constructed from the estimated channel, for example The actual effective signal amplitude is .

[0093] In confidence events Using the triangle inequality and the Cauchy-Schwarz inequality, we have:

[0094] (9)

[0095] superscript Conjugate transpose.

[0096] Thus, the lower bound of the signal power in the confidence sense is obtained:

[0097] (10)

[0098] in This is a non-negative truncation operator. This formula does not depend on the specific channel distribution details, but only on the uncertainty radius. Therefore, it is robust and feasible.

[0099] Transformation of the lower confidence bound of signal power:

[0100] Estimation errors can introduce residual self-interference or equivalent noise enhancement. To ensure provability and auditability, this invention employs a non-negative conservative upper bound:

[0101] (11)

[0102] Where α1, α2≥0 are coefficients related to receiver normalization, power configuration, combining method and implementation details, which can be fixed in the system implementation or obtained through offline calibration.

[0103] Equivalent signal-to-interference-plus-noise ratio confidence lower bound:

[0104] True instantaneous signal-to-interference-plus-noise ratio: Let This represents the actual external interference power of link k in the current time slot. Given thermal noise power, the true instantaneous signal-to-interference-plus-noise ratio (SINNR) is defined as:

[0105] (12)

[0106] This quantity depends on the actual channel and transient interference, and cannot be obtained directly.

[0107] Inference-induced confidence lower bound: The bridging layer provides a lower bound on the signal power under confidence events. and with Characterize the upper bound of the residual terms, defined as follows:

[0108] (13)

[0109] This formula expresses the following: Given an observation Y, with a confidence level of at least 1-δ, the true effective signal power is not less than... Therefore, at the same confidence level, the equivalent signal-to-interference-plus-noise ratio is no less than This quantity still contains unknown transient disturbances. It is used for theoretical connections and interface definitions.

[0110] Step (3) is as follows:

[0111] Computable and auditable lower bound under spatial regularization: In the guarantee layer, the spatial regularization condition is used to obtain... Consistent upper bound Thus, a computable and auditable lower bound is defined under space regularization:

[0112] (14)

[0113] This formula serves as the auditable lower bound for final system control and certificate output.

[0114] Let Φ represent the definition in The stationary point process (set of potential transmitter positions) of all potential transmitter nodes on the [planet name]. [Denotes this process]. For Center, radius The opening ball. If there exist finite constants σ≥0, ρ≥0, ν≥0 such that the following event occurs with probability 1:

[0115] (15)

[0116] Then Φ is said to satisfy strong(σ,ρ,ν)-ball regulation. Where (σ,ρ,ν) is called the regulation strength parameter. Indicates all With all The condition set must intersect, meaning the above inequality must hold simultaneously for any sphere center and any radius (and almost everywhere in a probabilistic sense). The key to this definition is that it holds almost everywhere for all positions y and all radii r, thus providing a mathematical basis for the consistent upper bound on interference and the lower bound on reliability that "holds for all links".

[0117] (Monotonicity of Active Subsets) The set of active transmitting nodes that simultaneously transmit in the current resource unit (e.g., within a certain time slot / frequency band) is denoted as The source of the substance is not limited; it only needs to meet the following conditions. .

[0118] Then for any , have Therefore, the upper bound of disturbance derived from the strong ball regularity of potential nodes holds for any active subset (resulting in a more conservative but robust guarantee).

[0119] (Path Loss Function) Let the path loss function be... It is a non-negative, bounded, and monotonically non-increasing function. Typical possible forms include:

[0120] Bounded power law: ,in ;

[0121] Cut-off power law: ,in (exist (Bounded).

[0122] In satisfying and Under the condition that, the subsequent upper bound is a finite constant.

[0123] (shot-noise upper bound and) At any receiving location Define the geometric shot-noise (i.e., aggregated path loss) generated by potential emitter nodes:

[0124] (16)

[0125] in, Point representation process The position of a node (the two-dimensional spatial coordinates of a potential launch node) is summed for all nodes located at the opening sphere. Perform on nodes within; The cutoff radius (observation radius) indicates that only the distance from the receiving position is considered in the statistics. No more than The node contribution; Indicates all and node position To sum up, that is, to sum the results of the opening ball. The potential launch nodes within are stacked one by one. When ordered... When the above equation converges to the shot-noise of the entire space, it corresponds to the geometric aggregate contribution of all potential emission nodes.

[0126] When Φ satisfies the strong(σ,ρ,ν)-ball regulation, there exists a definite upper bound for all R>0:

[0127] (17)

[0128] Let R→∞ to obtain the computable upper bound constant for shot-noise in the entire space:

[0129] (18)

[0130] Thus for any have

[0131] (Upper bound of interference power) Assume that the transmit power of each active transmitting node on the same frequency RB satisfies 0 ≤ P x ≤P max Then the aggregated interference power at any receiving point satisfies a consistent upper bound:

[0132] (19)

[0133] Depend on As can be seen from the upper bound of power, the interference power generated by any set of activities is all... Upper bound coverage.

[0134] (A computable and auditable equivalent lower bound) is given in the bridging layer. and Based on this, the auditable lower bound under spatial regularization is obtained using equation (14). .

[0135] (Certificate Definition) For a given threshold θ>0 and confidence level Define a reliability certificate:

[0136] (20)

[0137] when When established, the equivalent signal-to-interference-plus-noise ratio of link k is conservatively guaranteed to be no less than the threshold θ at confidence level 1-δ, thus it can be used for admission, scheduling, coding and modulation selection and retransmission control.

[0138] Specifically, certificates are used to control:

[0139] like If successful, data transmission is permitted and the coding rate or scheduling priority can be configured.

[0140] If this is not the case, measures such as reducing concurrent access, increasing pilot resources, adjusting power, or triggering retransmissions will be taken to improve the inference confidence or tighten concurrent access on the same frequency, thereby restoring the guaranteeability.

[0141] Engineering applications of this invention (typical scenarios):

[0142] Access Admission Control: Only allow access that meets the following criteria. When the terminal enters the data period, it is necessary to avoid the unreliable link from lowering the overall reliability;

[0143] Reliability-driven scheduling: As a ranking metric, prioritizing services can demonstrate a more reliable link;

[0144] Adaptive resource configuration: When the reliability certificate of link k If the certificate is invalid, the system prioritizes the resource action most effective for the certificate criterion (such as extending the pilot signal, reducing concurrent access, adjusting power, or triggering retransmission) to reduce the radius of uncertainty. Or reduce the upper limit of interference This makes the certificate criterion valid;

[0145] Critical business safeguards: For businesses that must meet the standards, adopt smaller delta outputs and more conservative certificates in exchange for stronger reliability assurance.

[0146] The meanings of the above symbols are shown in Table 1:

[0147] Table 1: Meaning of Symbols

[0148]

[0149] The present invention also provides a bridging system for inference and guarantee of reliable communication, comprising:

[0150] Generative inference layer: Jointly infers the set of active terminals and their channels from pilot observations, and generates posterior samples and confidence information;

[0151] Bridging layer: Transforms posterior samples and confidence information into a provable lower bound of signal power confidence and an upper bound of residual terms, forming an equivalent lower bound of signal-to-interference-plus-noise ratio confidence;

[0152] Guarantee layer: Based on the upper limit of the interference power generated by the spatial regularity conditions, output a reliability certificate.

[0153] It should be noted that the system embodiment shown in this embodiment matches the content of the above method embodiment, and the content of the above method embodiment can be referred to, and will not be repeated here.

[0154] The present invention also provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, is used to implement the above-described bridging method for inferring and guaranteeing reliable communication.

[0155] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described bridging method for inferring and guaranteeing reliable communication.

[0156] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0157] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0158] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0159] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0160] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A bridging method for inferring and guaranteeing reliable communication, characterized in that, include: The active terminal set and its channel are jointly inferred from pilot observations, and posterior samples and confidence information are generated. The posterior samples and confidence information are transformed into a provable lower bound for signal power confidence and an upper bound for residual terms, forming an equivalent lower bound for signal-to-interference-plus-noise ratio confidence. Based on the upper bound of the interference power generated by the spatial regularity conditions, a reliability certificate is output. The transformation of the lower confidence bound of signal power is as follows: Let receive the merged vector From the estimated channel The actual effective signal amplitude is constructed as follows: ,in, For vector modulo, superscript This indicates the conjugate transpose. This is the uplink channel vector of the k-th terminal; In confidence events Using the triangle inequality and the Cauchy-Schwarz inequality, we have: ； in For vector norm, Let the error confidence radius be . For confidence level parameters; Thus, the lower bound of the signal power in the confidence sense is obtained: ； The upper bound of the residual term is: ； Where α1, α2 ≥ 0 are coefficients. Let be the error confidence radius, where Here, K represents the terminal index, and K represents the total number of candidate terminals. For confidence level parameters; make This represents the actual external interference power of link k in the current time slot. Given the thermal noise power, provide a lower bound for the signal power under the confidence event. and with Characterizing the upper bound of the residual term yields the lower bound of the equivalent signal-to-interference-plus-noise ratio confidence level. for: 。 2. The method according to claim 1, characterized in that, The active terminal set and its channels are jointly inferred from pilot observations, and posterior samples and confidence information are generated, including: Fractional generative posterior inference models based on pilot observations With pilot set Input, output activity information and channel point estimation for each terminal And calculate the error confidence radius. ,in Here, K represents the terminal index, and K represents the total number of candidate terminals. This represents the confidence level parameter.

3. The method according to claim 1, characterized in that, The upper bound of interference power generated by combining spatial regularity conditions is consistent. This defines an auditable lower bound under spatial regularization. : ； Thus defining a reliability certificate : ； Where θ is the reliability threshold. This represents the confidence level parameter.

4. The method according to claim 3, characterized in that, Let Φ represent the definition in The stationary point process of all potential launch nodes on the [the network]; [denotes] For Center, radius The opening ball; if there exist finite constants σ≥0, ρ≥0, ν≥0 such that the following event occurs with probability 1: ； Then Φ is said to satisfy strong sphere regularity; Where (σ,ρ,ν) is called the regularity strength parameter. Indicates all With all The condition sets intersect; Let the path loss function be defined. It is a non-negative, bounded, monotonically non-increasing function, and satisfies and ; At any receiving location Define the geometric aggregation path loss generated by the potential launch node: ； in, Point representation process One of the node positions; The cutoff radius indicates that only the distance from the receiving position is considered. No more than The node contribution; Indicates all and node position To sum up, that is, to sum the results of the opening ball. The potential launch nodes within are stacked one by one; When Φ is strongly spherically regular, there exists a definite upper bound for all R>0: Let R→∞ to obtain a computable upper bound constant for the sum of aggregated path losses in the entire space. : ； Thus for any have ; Assume that the transmit power of each active transmitting node on the same frequency RB satisfies 0 ≤ P x ≤P max Then the aggregated interference power at any receiving point satisfies a consistent upper bound. 。 5. A bridging system for reliable communication inference and guarantee, characterized in that, include: Generative inference layer: Jointly infers the set of active terminals and their channels from pilot observations, and generates posterior samples and confidence information; Bridging layer: Transforms posterior samples and confidence information into a provable lower bound of signal power confidence and an upper bound of residual terms, forming an equivalent lower bound of signal-to-interference-plus-noise ratio confidence; The transformation of the lower confidence bound of signal power is as follows: Let receive the merged vector From the estimated channel The actual effective signal amplitude is constructed as follows: ,in, For vector modulo, superscript This indicates the conjugate transpose. This is the uplink channel vector of the k-th terminal; In confidence events Using the triangle inequality and the Cauchy-Schwarz inequality, we have: ； in For vector norm, Let the error confidence radius be . For confidence level parameters; Thus, the lower bound of the signal power in the confidence sense is obtained: ； The upper bound of the residual term is: ； Where α1, α2 ≥ 0 are coefficients. Let be the error confidence radius, where Here, K represents the terminal index, and K represents the total number of candidate terminals. For confidence level parameters; make This represents the actual external interference power of link k in the current time slot. Given the thermal noise power, provide a lower bound for the signal power under the confidence event. and with Characterizing the upper bound of the residual term yields the lower bound of the equivalent signal-to-interference-plus-noise ratio confidence level. for: ； Guarantee layer: Based on the upper limit of the interference power generated by the spatial regularity conditions, output a reliability certificate.

6. A computer-readable storage medium having a program stored thereon, characterized in that, When executed by a processor, the program is used to implement a bridging method for inferring and guaranteeing reliable communication as described in any one of claims 1-4.

7. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements a bridging method for inferring and guaranteeing reliable communication as described in any one of claims 1-4.

Images