A distributed MIMO energy efficiency optimization method based on joint subcarrier and power allocation under network slicing

Abstract

The present invention discloses a distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under network slicing, comprising: establishing a downlink distributed MIMO system model; establishing an objective function and constraints for the energy efficiency problem of joint power and subcarrier allocation under eMBB and URLLC slicing; converting the original problem into a convex problem based on a large M method, integer relaxation, fractional programming, auxiliary variable approximation and convex difference programming method; the present invention can effectively improve the spectrum efficiency and energy efficiency of a distributed MIMO system under a network slicing architecture, meet the quality of service (QoS) requirements of different users, and has broad application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed MIMO system energy efficiency optimization, and in particular to a distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under network slicing. Background Art

[0002] 5G and the emerging 6G wireless networks are envisioned to support a variety of services, such as enhanced mobile broadband (eMBB) with high data rates, ultra-reliable low-latency communications (URLLC) that enable ultra-high reliability and very low latency, and massive machine-type communications (mMTC) that provide connectivity for a large number of devices.

[0003] In order to enable these services to coexist under a shared physical architecture, especially eMBB and URLLC services, network slicing is proposed to divide the entire network into multiple virtual and logically independent sub-networks. The virtual sub-networks are named slices, and each slice supports specific business requirements. Radio resources such as subcarriers and transmit power are dynamically allocated to slices according to business needs. In addition, to promote network slicing, distributed MIMO is considered as a physical architecture for network slicing as a 5G key technology. Unlike traditional centralized MIMO systems, in distributed MIMO systems, remote antenna units are distributed in cells and connected to the baseband unit pool via optical fiber. Distributed remote antenna units facilitate multi-point collaboration, and multiple antennas facilitate channel hardening, which greatly improves the reliability of URLLC and the data rate of eMBB.

[0004] However, when eMBB and URLLC slices coexist in a distributed MIMO system, how to determine the optimal resource allocation is a challenging problem. To solve this problem, a resource puncturing method is proposed, whose main mechanism is to puncture the spectrum resources that have been allocated to the eMBB slices, and then reallocate these spectrum resources to the URLLC slices. This allows for fast scheduling of URLLC services and low latency. However, the eMBB slices lose spectrum resources and suffer performance degradation. Moreover, because each carrier is only allocated to one user to avoid multi-user interference, such a resource strategy does not fully utilize spectrum resources. To this end, a subcarrier and power hybrid allocation strategy is proposed to improve spectrum efficiency, in which each subcarrier can be used by more than one user, and eMBB does not need to sacrifice its own performance for URLLC. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under network slicing. The method jointly optimizes subcarrier allocation and power allocation, and studies the problem of maximizing the energy efficiency of the downlink distributed MIMO system while meeting QoS requirements.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under network slicing, the method comprising:

[0008] Step S1: for a distributed MIMO system, construct a distributed MIMO mathematical model under eMBB and URLLC slice services, wherein the distributed MIMO mathematical model includes a data transmission model and an energy efficiency expression;

[0009] Step S2: for the distributed MIMO mathematical model constructed in step S1, construct an optimization problem with maximizing energy efficiency as the objective function, and with QoS requirements of different slices, maximum power consumption limit and integer constraints as constraints, wherein the maximum power consumption limit is that the sum of the power of all users on all subcarriers must be less than or equal to the maximum value of the system transmission energy consumption, and the sum of the power of all users on all subcarriers is in the form of the product of the subcarrier allocation variable and the power;

[0010] Step S3, sequentially using the big M method, integer relaxation, fractional programming, auxiliary variable approximation and convex difference programming to transform the optimization problem constructed in step S2 into a convex problem, which includes: first, decomposing the product form of the subcarrier allocation variable and the power according to the big M method, and then reconstructing the integer constraint by using the integer relaxation method, and then using the fractional programming to convert the fractional objective function into a subtraction form, and obtaining the reconstructed objective function and constraints, and finally by introducing auxiliary variables and convex difference programming methods, the reconstructed system energy efficiency problem is transformed into a convex problem; wherein the introduced auxiliary variables are used to constrain the user's non-convex signal-to-interference-noise ratio;

[0011] Step S4: Iteratively solve the convex problem obtained in step S3 to obtain optimal subcarrier allocation and power allocation.

[0012] Furthermore, in step S1, the distributed MIMO system specifically includes:

[0013] 1 baseband unit pool, H remote antenna units with L antennas and K single-antenna users;

[0014] The entire frequency band is divided into M orthogonal subcarriers, each subcarrier is allocated to multiple single-antenna users mentioned above, wherein the slices in the distributed MIMO system are eMBB slices and URLLC slices, and the baseband unit pool has known channel state information.

[0015] Furthermore, the data transmission model specifically includes:

[0016] The data rate of the kth user in the eMBB slice is:

[0017] The data rate of the kth user in the URLLC slice is:

[0018]

[0019] The weighted throughput of the system is:

[0020] In the above formula, k∈{1,...,K}, m∈{1,...,M}, It is expressed as the signal-to-interference-to-noise ratio of user k on subcarrier m, which is specifically expressed as: is the conjugate transpose of the channel gain matrix between user k and RRU on subcarrier m, is the normalized precoding direction of user k on subcarrier m, is the power allocated to user k on subcarrier m, σ 2 is the noise variance, is an integer variable used for subcarrier allocation. If user k receives a signal from subcarrier m, otherwise a=log2(e),∈ k is the decoding error probability, n is the codeword block length, Q -1 (·) is the inverse function of the Gaussian function, T(γ k ) is the short packet fading, μ k is the weight coefficient of user k, μ k The larger the value of , the higher the priority of the user. log represents logarithmic operation, |·| represents absolute value operation, and ‖·‖ represents norm operation.

[0021] Furthermore, the energy efficiency expression is:

[0022]

[0023] In this formula, is the total power consumption of the system, Q B is the power consumption of the BBU pool to ensure that η EE The denominator of is positive.

[0024] Furthermore, in step S2, the maximization of energy efficiency is the objective function, which is specifically:

[0025]

[0026] In this formula, p represents all A collection of It indicates that the allocation of subcarriers and the power allocation of the corresponding subcarriers of the distributed MIMO system are optimized respectively so that the subsequent function value is maximized.

[0027] Furthermore, in step S2, the constraint conditions are specifically:

[0028] QoS requirements for eMBB slices and URLLC slices:

[0029]

[0030]

[0031] In this formula, k∈{1,...,K}, Minimum throughput requirements for users served by eMBB slices and URLLC slices, They represent the user sets associated with the eMBB network slice and the URLLC network slice respectively;

[0032] Power consumption constraints of distributed MIMO systems:

[0033]

[0034]

[0035] In the formula, k∈{1,...,K}, h∈{1,...,H}, m∈{1,...,M}, is the normalized precoding direction of user k on subcarrier m of RRU h, is the maximum transmission energy consumption of each RRU;

[0036] Create an integer constraint:

[0037]

[0038] Further, in the step S2, the product form of the subcarrier allocation variable and the power is decomposed according to the large M method, and then the integer relaxation method is used to reconstruct the integer constraint, and then the fractional objective function is equivalently converted into a subtraction form by using fractional programming to obtain the reconstructed objective function and constraints, which specifically include:

[0039] Step S3011: Introduce auxiliary variables Then the objective function of the original problem and the constraints Replace with And without changing the feasible solution, the variables Relax and introduce new constraints as follows;

[0040]

[0041] This constraint ensures that if a subcarrier is not assigned to a user, The power transmitted to user k on subcarrier m is also zero, that is, the auxiliary variable Otherwise auxiliary variables Maximum reachable

[0042] Step S3012: Reconstruct the integer constraint in step S204 using the integer relaxation method. Then introduce the following new constraints:

[0043]

[0044]

[0045] in All variables The above two constraints will be integer variables Slack is a continuous variable;

[0046] Step S3013: Use fractional programming to convert the fractional objective function into a subtraction form. The reconstructed objective function and constraints are as follows:

[0047]

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055] In the formula, represents the optimal energy efficiency of the i-th iteration, yes A collection of is the data rate of the kth user in the eMBB slice, The signal-to-noise ratio In Replace with The result after that, The data rate expression of the kth user in the URLLC slice is In Replace with As a result, the first constraint is the replaced eMBB slice rate constraint, the second constraint is the replaced URLLC slice rate constraint, the third constraint is the replaced power constraint, and the fourth constraint ensures that if the subcarrier is not allocated to the user, that is, The power transmitted to user k on subcarrier m is also zero, that is, the auxiliary variable Otherwise auxiliary variables Maximum reachable The fifth and sixth constraints set the integer variable Relaxed to a continuous variable, the seventh constraint ensures that the power allocated to user k on subcarrier m is non-negative, st indicates that it satisfies the following conditions, and i indicates the number of iterations.

[0056] Furthermore, the method of introducing auxiliary variables and convex difference programming is used to transform the reconstructed system energy efficiency problem into a convex problem, which specifically includes:

[0057] Step S3021: Introduce auxiliary variables To constrain the non-convex signal-to-interference-noise ratio of user k in eMBB and URLLC That is, the following constraints, using auxiliary variables For all The new constraints introduced are as follows:

[0058]

[0059] In the formula, This constraint represents the auxiliary variable introduced The value range of

[0060] Step S3022: Since the above constraints are non-convex constraints, an auxiliary variable is further introduced. Further constraints on the numerator and denominator of SINR are expressed as follows:

[0061]

[0062]

[0063]

[0064] In the formula, is a temporary variable, is the channel gain matrix between user k and RRU on subcarrier m, where the first constraint represents that the power of user k receiving information needs to be greater than The second constraint represents that the interference and noise power of user k needs to be less than Due to the existence of these two constraints, it is avoided that when optimizing energy efficiency, the user's received power and the interference noise power become smaller simultaneously, and the signal to noise ratio is reduced. Still a very undesirable situation, the third constraint is actually a restatement of the right part of the constraint in step S3021, that is, The numerator and denominator of are replaced by new auxiliary variables respectively;

[0065] Step S3023, the fifth constraint of the reconstructed problem obtained in step S3013, the third constraint in S3022, and the throughput of the URLLC slice are still non-convex constraints, but they conform to the form of convex difference programming, so they are transformed using the first-order Taylor approximation method, and slack variables are introduced to transform the reconstructed system energy efficiency problem obtained in step S301 into a convex problem;

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077] where τ is the number of slack variables τ k is a set of, λ is the penalty factor, φ, α, β are auxiliary variables A collection of is the result of the first-order Taylor approximation of the throughput of the URLLC slice, G (i) (s) represents the result of first-order Taylor approximation of G(s), Express The results of the first-order Taylor approximation are as follows:

[0078]

[0079]

[0080]

[0081] in,

[0082] Furthermore, the step S4 specifically includes:

[0083] Step S401, initializing various parameters of the loop, including: setting the maximum number of iterations I max , s (0) ,φ (0) ,α (0) ,β (0) and the penalty factor λ;

[0084] Step S402: Solve the convex problem obtained in step S3023 to obtain the current optimal subcarrier allocation coefficient s * , Power Distribution and the corresponding auxiliary variable φ * ,α * ,β * ;

[0085] Step S403, update the number of iterations i=i+1, and use the following expression to s (i) ,φ (i) ,α (i) ,β (i) , Update, expressed as:

[0086] s (i) =s * ,φ (i) =φ * ,α (i) =α * ,β (i) =β

[0087]

[0088] Step S404: determine whether the loop has converged. If the algorithm has converged or reached the maximum number of iterations, max Go to step S405, otherwise go to step S402;

[0089] Step S405: Output the optimal subcarrier allocation and power allocation s for the current cycle * ,

[0090] The beneficial effects of the present invention are:

[0091] The present invention considers the energy efficiency maximization problem of distributed MIMO systems under eMBB and URLLC slicing by jointly allocating subcarriers and power, while ensuring the constraints of QoS requirements. The present invention can improve spectrum efficiency by allocating subcarriers to multiple users, meet the service quality requirements of different users on a shared physical architecture, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 A schematic diagram of a flow chart of a distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under a network slice provided in Example 1;

[0093] Figure 2 A schematic diagram of the flow of executing the algorithm in step S4 provided in Example 1;

[0094] Figure 3 It is a comparison chart of energy efficiency obtained by the method of Example 1 and the case of considering only eMBB, the case of considering only URLLC, the case of allocating subcarriers in OFDM mode, the case of optimizing the total rate and the case of equal power allocation. DETAILED DESCRIPTION

[0095] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0096] Example 1

[0097] See also Figure 1-Figure 3 This embodiment provides a distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under network slicing. The specific process of the method is as follows Figure 1 As shown, the method specifically comprises the following steps:

[0098] Step S1: for a distributed MIMO system, construct a distributed MIMO mathematical model under eMBB and URLLC slice services, wherein the distributed MIMO mathematical model includes a data transmission model and an energy efficiency expression;

[0099] Specifically, in this embodiment, step S1 includes:

[0100] Step S101: construct a distributed MIMO system, the system comprising: 1 baseband unit pool, H remote antenna units equipped with L antennas, and K single-antenna users;

[0101] The entire frequency band is divided into M orthogonal subcarriers, each subcarrier is allocated to multiple single-antenna users mentioned above, wherein the slices in the distributed MIMO system are eMBB slices and URLLC slices, and the baseband unit pool has known channel state information;

[0102] Step S102: for the distributed MIMO system in step S101, establish its data transmission model, the data transmission model specifically includes:

[0103] The data rate of the kth user in the eMBB slice is:

[0104] The data rate of the kth user in the URLLC slice is:

[0105]

[0106] The weighted throughput of the system is:

[0107] In the above formula, k∈{1,...,K}, m∈{1,...,M}, It is expressed as the signal-to-interference-to-noise ratio of user k on subcarrier m, which is specifically expressed as: is the conjugate transpose of the channel gain matrix between user k and RRU on subcarrier m, is the normalized precoding direction of user k on subcarrier m, is the power allocated to user k on subcarrier m, σ 2 is the noise variance, is an integer variable used for subcarrier allocation. If user k receives a signal from subcarrier m, otherwise a=log2(e),∈ k is the decoding error probability, n is the codeword block length, Q -1 (·) is the inverse function of the Gaussian function, T(γ k ) is the short packet fading, μ k is the weight coefficient of user k, μ k The larger the value of , the higher the priority of the user. log represents logarithmic operation, |·| represents absolute value operation, and ‖·‖ represents norm operation.

[0108] Step S103: construct a performance measurement index for the distributed MIMO system constructed in step S101. The performance measurement index is energy efficiency, which is specifically expressed as:

[0109]

[0110] In this formula, is the total power consumption of the system, Q B is the power consumption of the BBU pool to ensure that η EE The denominator of is positive.

[0111] Step S2: for the distributed MIMO mathematical model constructed in step S1, construct an optimization problem with maximizing energy efficiency as the objective function, and with QoS requirements of different slices, maximum power consumption limit and integer constraint as constraints; wherein the maximum power consumption limit is that the sum of the power of all users on all subcarriers must be less than or equal to the maximum value of the system transmission energy consumption, and the sum of the power of all users on all subcarriers is in the form of the product of the subcarrier allocation variable and the power;

[0112] Specifically, in this embodiment, step S2 specifically includes:

[0113] Step S201: construct an objective function, which is specifically expressed as:

[0114]

[0115] In this formula, p represents all A collection of It indicates respectively optimizing the allocation of subcarriers of the distributed MIMO system and the power allocation of the corresponding subcarriers so that the subsequent function value is maximized;

[0116] Step S202: Construct QoS requirements for eMBB slices and URLLC slices:

[0117]

[0118]

[0119] In this formula, k∈{1,...,K}, Minimum throughput requirements for users served by eMBB slices and URLLC slices, They represent the user sets associated with the eMBB network slice and the URLLC network slice respectively;

[0120] Step S203: construct power consumption constraints for a distributed MIMO system:

[0121]

[0122]

[0123] In the formula, k∈{1,...,K}, h∈{1,...,H}, m∈{1,...,M}, is the normalized precoding direction of user k on subcarrier m of RRU h, is the maximum transmission energy consumption of each RRU, where represents the power obtained by user k on subcarrier m from all RRUs, represents the power sent by RRU h to user k on subcarrier m, represents the sum of the power sent by RRU h to all users on all subcarriers;

[0124] Step S204: Establish integer constraints:

[0125]

[0126] Step S3, transforming the optimization problem constructed in step S2 based on big M method, integer relaxation, fractional programming, auxiliary variable approximation and convex difference programming;

[0127] Specifically, in this embodiment, step S3 specifically includes:

[0128] Step S301, decomposing the product form of the subcarrier allocation variable and the power according to the big M method, reconstructing the integer constraint using the integer relaxation method, and then using the fractional programming to convert the fractional objective function into a subtraction form;

[0129] More specifically, in this embodiment, step S301 includes:

[0130] Step S3011: Introduce auxiliary variables Then the objective function of the original problem and the constraints Replace with And without changing the feasible solution, the variables Relax and introduce new constraints as follows;

[0131]

[0132] This constraint ensures that if a subcarrier is not assigned to a user, The power transmitted to user k on subcarrier m is also zero, that is, the auxiliary variable Otherwise auxiliary variables Maximum reachable

[0133] Step S3012: Reconstruct the integer constraint in step S204 using the integer relaxation method. Then introduce the following new constraints:

[0134]

[0135]

[0136] in All variables The above two constraints will be integer variables Slack is a continuous variable;

[0137] Step S3013: Use fractional programming to convert the fractional objective function into a subtraction form. The reconstructed objective function and constraints are as follows:

[0138]

[0139]

[0140]

[0141]

[0142]

[0143]

[0144]

[0145]

[0146] In the formula, represents the optimal energy efficiency of the i-th iteration, yes A collection of is the data rate of the kth user in the eMBB slice, The signal-to-noise ratio In Replace with The result after that, The data rate expression of the kth user in the URLLC slice is In Replace with As a result, the first constraint is the replaced eMBB slice rate constraint, the second constraint is the replaced URLLC slice rate constraint, the third constraint is the replaced power constraint, and the fourth constraint ensures that if the subcarrier is not allocated to the user, that is, The power transmitted to user k on subcarrier m is also zero, that is, the auxiliary variable Otherwise auxiliary variables Maximum reachable The fifth and sixth constraints set the integer variable Relaxed to a continuous variable, the seventh constraint ensures that the power allocated to user k on subcarrier m is non-negative, st represents the condition to be satisfied, and i represents the number of iterations;

[0147] Step S302: transforming the reconstructed system energy efficiency problem obtained in step S301 into a convex problem by introducing auxiliary variables and convex difference programming;

[0148] More specifically, in this embodiment, step S302 includes:

[0149] Step S3021: Introduce auxiliary variables To constrain the non-convex signal-to-interference-noise ratio of user k in eMBB and URLLC That is, the following constraints, using auxiliary variables For all Replace (this replacement is equivalent because the eMBB and URLLC capacity formulas are based on is a monotonic function of the variable, so When the optimization problem is solved optimally, The new constraints introduced are as follows:

[0150]

[0151] In the formula, This constraint represents the auxiliary variable introduced The value range of

[0152] Step S3022: Since the above constraints are non-convex constraints, an auxiliary variable is further introduced. Further constraints on the numerator and denominator of SINR are expressed as follows:

[0153]

[0154]

[0155]

[0156] In the formula, is a temporary variable, is the channel gain matrix between user k and RRU on subcarrier m, where the first constraint represents that the power of user k receiving information needs to be greater than The second constraint represents that the interference and noise power of user k needs to be less than Due to the existence of these two constraints, it is avoided that when optimizing energy efficiency, the user's received power and the interference noise power become smaller simultaneously, and the signal to noise ratio is reduced. Still a very undesirable situation, the third constraint is actually a restatement of the right part of the constraint in step S3021, that is, The numerator and denominator of are replaced by new auxiliary variables respectively;

[0157] Step S3023, the fifth constraint of the reconstructed problem obtained in step S3013, the third constraint in S3022, and the throughput of the URLLC slice are still non-convex constraints, but they conform to the form of convex difference programming, so they are transformed using the first-order Taylor approximation method, and slack variables are introduced to transform the reconstructed system energy efficiency problem obtained in step S301 into a convex problem;

[0158]

[0159]

[0160]

[0161]

[0162]

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169] where τ is the number of slack variables τ k is a set of, λ is the penalty factor, φ, α, β are auxiliary variables A collection of is the result of the first-order Taylor approximation of the throughput of the URLLC slice, G (i) (s) represents the result of first-order Taylor approximation of G(s), Express The results of the first-order Taylor approximation are as follows:

[0170]

[0171]

[0172]

[0173] in

[0174] Step S4, solving the convex problem obtained in step S3 by iteration;

[0175] Specifically, in this embodiment, step S4 specifically includes:

[0176] Step S401, initializing various parameters of the loop, including: setting the maximum number of iterations I max , s (0) ,φ (0) ,α (0) ,β (0) and the penalty factor λ;

[0177] Step S402: Solve the convex problem obtained in step S3023 to obtain the current optimal subcarrier allocation coefficient s * , Power Distribution and the corresponding auxiliary variable φ * ,α * ,β * ;

[0178] Step S403, update the number of iterations i=i+1, and use the following expression to s (i) ,φ (i) ,α (i) ,β (i) , Update, expressed as:

[0179] s (i) =s * ,φ (i) =φ * ,α (i) =α * ,β (i) =β

[0180]

[0181] Step S404: determine whether the loop has converged. If the algorithm has converged or reached the maximum number of iterations, max Go to step S405, otherwise go to step S402;

[0182] Step S405: Output the optimal subcarrier allocation and power allocation s for the current cycle * ,

[0183] In summary, the present invention provides a distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under network slicing, including: establishing a downlink distributed MIMO system model; establishing the objective function and constraints of the energy efficiency problem of joint power and subcarrier allocation under eMBB and URLLC slicing; converting the original problem into a convex problem based on the big M method, integer relaxation, fractional programming, auxiliary variable approximation and convex difference programming method; the present invention can effectively improve the spectrum efficiency and energy efficiency of the distributed MIMO system under the network slicing architecture, meet the service quality (QoS) requirements of different users, and has broad application prospects.

[0184] The matters not described in detail in the present invention are all known technologies to those skilled in the art.

[0185] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A distributed MIMO energy efficiency optimization method for joint subcarrier and power allocation under network slicing, characterized in that: The method comprises: Step S1: for a distributed MIMO system, construct a distributed MIMO mathematical model under eMBB and URLLC slice services, wherein the distributed MIMO mathematical model includes a data transmission model and an energy efficiency expression; Step S2: for the distributed MIMO mathematical model constructed in step S1, construct an optimization problem with maximizing energy efficiency as the objective function, and with QoS requirements of different slices, maximum power consumption limit and integer constraints as constraints, wherein the maximum power consumption limit is that the sum of the power of all users on all subcarriers must be less than or equal to the maximum value of the system transmission energy consumption, and the sum of the power of all users on all subcarriers is in the form of the product of the subcarrier allocation variable and the power; Step S3, sequentially using the big M method, integer relaxation, fractional programming, auxiliary variable approximation and convex difference programming to transform the optimization problem constructed in step S2 into a convex problem, which includes: first, decomposing the product form of the subcarrier allocation variable and the power according to the big M method, and then reconstructing the integer constraint by using the integer relaxation method, and then using the fractional programming to convert the fractional objective function into a subtraction form, and obtaining the reconstructed objective function and constraints, and finally by introducing auxiliary variables and convex difference programming methods, the reconstructed system energy efficiency problem is transformed into a convex problem; wherein the introduced auxiliary variables are used to constrain the user's non-convex signal-to-interference-noise ratio; Step S4, using iteration to solve the convex problem obtained in step S3 to obtain optimal subcarrier allocation and power allocation; In the step S1, the distributed MIMO system specifically includes: 1 baseband unit pool, H remote antenna units with L antennas and K single-antenna users; The entire frequency band is divided into M orthogonal subcarriers, each subcarrier is allocated to multiple single-antenna users mentioned above, wherein the slices in the distributed MIMO system are eMBB slices and URLLC slices, and the baseband unit pool has known channel state information; The data transmission model specifically includes: The data rate of the kth user in the eMBB slice is: The data rate of the kth user in the URLLC slice is: The weighted throughput of the system is: In the above formula, k∈{1,...,K}, m∈{1,...,M}, It is expressed as the signal-to-interference-to-noise ratio of user k on subcarrier m, which is specifically expressed as: is the conjugate transpose of the channel gain matrix between user k and RRU on subcarrier m, is the normalized precoding direction of user k on subcarrier m, is the power allocated to user k on subcarrier m, σ 2 is the noise variance, is an integer variable used for subcarrier allocation. If user k receives a signal from subcarrier m, otherwise a=log2(e),∈ k is the decoding error probability, n is the codeword block length, Q -1 (·) is the inverse function of the Gaussian function, T(γ k ) is the short packet fading, μ k is the weight coefficient of user k, μ k The larger the value of , the higher the priority of the user. log represents logarithmic operation, |·| represents absolute value operation, and ‖·‖ represents norm operation. The energy efficiency expression is: In this formula, is the total system power consumption, Q B is the power consumption of the BBU pool to ensure that η EE The denominator of is positive; In step S2, the maximization of energy efficiency is the objective function, which is specifically: In this formula, p represents all A collection of It indicates respectively optimizing the allocation of subcarriers of the distributed MIMO system and the power allocation of the corresponding subcarriers so that the subsequent function value is maximized; In step S2, the constraint conditions are specifically: QoS requirements for eMBB slices and URLLC slices: In this formula, k∈{1,...,K}, Minimum throughput requirements for users served by eMBB slices and URLLC slices, They represent the user sets associated with the eMBB network slice and the URLLC network slice respectively; Power consumption constraints of distributed MIMO systems: In the formula, k∈{1,...,K}, h∈{1,...,H}, m∈{1,...,M}, is the normalized precoding direction of user k on subcarrier m of RRU h, is the maximum transmission energy consumption of each RRU; Create an integer constraint: In the step S2, the product form of the subcarrier allocation variable and the power is decomposed according to the large M method, and then the integer relaxation method is used to reconstruct the integer constraint, and then the fractional objective function is equivalently converted into a subtraction form by using fractional programming to obtain the reconstructed objective function and constraints, which specifically include: Step S3011: Introduce auxiliary variables Then the objective function of the original problem and the constraints Replace with And without changing the feasible solution, the variables Relax and introduce new constraints as follows; This constraint ensures that if a subcarrier is not assigned to a user, The power transmitted to user k on subcarrier m is also zero, that is, the auxiliary variable Otherwise auxiliary variables Maximum reach Step S3012: Reconstruct the integer constraint in step S204 using the integer relaxation method. Then introduce the following new constraints: in All variables The above constraints will be integer variables Slack is a continuous variable; Step S3013: Use fractional programming to convert the fractional objective function into a subtraction form. The reconstructed objective function and constraints are as follows: In the formula, represents the optimal energy efficiency of the i-th iteration, yes A collection of is the data rate of the kth user in the eMBB slice, The signal-to-noise ratio In Replace with The result after that, The data rate expression of the kth user in the URLLC slice is In Replace with As a result, the first constraint is the replaced eMBB slice rate constraint, the second constraint is the replaced URLLC slice rate constraint, the third constraint is the replaced power constraint, and the fourth constraint ensures that if the subcarrier is not allocated to the user, that is, The power transmitted to user k on subcarrier m is also zero, that is, the auxiliary variable Otherwise auxiliary variables Maximum reach The fifth and sixth constraints set the integer variable Relaxed to a continuous variable, the seventh constraint ensures that the power allocated to user k on subcarrier m is non-negative, st represents the condition to be satisfied, and i represents the number of iterations; The method of introducing auxiliary variables and convex difference programming transforms the reconstructed system energy efficiency problem into a convex problem, which specifically includes: Step S3021: Introduce auxiliary variables To constrain the non-convex signal-to-interference-noise ratio of user k in eMBB and URLLC That is, the following constraints, using auxiliary variables For all The new constraints introduced are as follows: In the formula, This constraint represents the auxiliary variable introduced The value range of Step S3022: Since the above constraints are non-convex constraints, an auxiliary variable is further introduced. Further constraints on the numerator and denominator of SINR are expressed as follows: In the formula, is a temporary variable, is the channel gain matrix between user k and RRU on subcarrier m, where the first constraint represents that the power of user k receiving information needs to be greater than The second constraint represents that the interference and noise power of user k needs to be less than Due to the existence of these two constraints, it is avoided that when optimizing energy efficiency, the user's received power and the interference noise power become smaller simultaneously, and the signal to noise ratio is reduced. Still a very undesirable situation, the third constraint is actually a restatement of the right part of the constraint in step S3021, that is, The numerator and denominator of are replaced by new auxiliary variables respectively; Step S3023, the fifth constraint of the reconstructed problem obtained in step S3013, the third constraint in S3022, and the throughput of the URLLC slice are still non-convex constraints, but they conform to the form of convex difference programming, so they are transformed using the first-order Taylor approximation method, and slack variables are introduced to transform the reconstructed system energy efficiency problem obtained in step S301 into a convex problem; where τ is the number of slack variables τ k is a set of, λ is the penalty factor, φ, α, β are auxiliary variables A collection of is the result of the first-order Taylor approximation of the throughput of the URLLC slice, G (i) (s) represents the result of first-order Taylor approximation of G(s), Express The results of the first-order Taylor approximation are as follows: in, The step S4 specifically includes: Step S401, initializing various parameters of the loop, including: setting the maximum number of iterations I max , s (0) ,φ (0) ,α (0) ,β (0) and the penalty factor λ; Step S402: Solve the convex problem obtained in step S3023 to obtain the current optimal subcarrier allocation coefficient s * , Power Distribution and the corresponding auxiliary variable φ * ,α * ,β * ; Step S403, update the number of iterations i=i+1, and use the following expression to Update, expressed as: Step S404: determine whether the loop has converged. If the algorithm has converged or reached the maximum number of iterations, max Go to step S405, otherwise go to step S402; Step S405: Output the optimal subcarrier allocation and power allocation of the current cycle

Images