A three-dimensional space-time-frequency resource allocation method for MU-MIMO system

Abstract

The present invention discloses a space-time-frequency three-dimensional resource allocation method for a MU-MIMO system, comprising the following steps: Step 1: Establishing a comprehensive multi-dimensional resource optimization problem; Step 2: Analyzing the optimization problem and developing a heuristic resource allocation method to improve spectrum efficiency while balancing complexity and system performance; Step 3: To address the waste of time and frequency resources in the resource allocation method in Step 2, the heuristic resource allocation method is improved into a traffic-aware heuristic resource allocation method based on the user's personalized traffic requirements; Step 4: To improve the spatial multiplexing capability of the resource allocation method, a beam selection criterion is proposed by balancing the matching performance of the beams with the repetition rate selected by the user; Step 5: Based on the beam selection criterion, a traffic-aware heuristic resource allocation method is used to transmit data for the MU-MIMO system. The present invention achieves higher spectrum efficiency and higher system sum rates, and is applicable to hardware system complexity.

Description

Technical Field

[0001] The present invention belongs to the technical field of mobile communications, and in particular relates to a space-time-frequency three-dimensional resource allocation method for a MU-MIMO system. Background Art

[0002] In the latest 5G and future 6G mobile communication systems, with the rapid growth in the number of end users, high-density equipment environments still need to achieve reliable and high-speed communications. Resources can be viewed as elements of a three-dimensional structure, corresponding to the frequency, time, and space domains. To obtain the optimal solution to the resource problem, it is necessary to compare the allocation of each possible beam on each time-frequency resource to improve resource utilization in each dimension. In fact, as mentioned earlier, the general sum rate maximization problem has been demonstrated to be a non-deterministic problem. The high complexity caused by the large number of degrees of freedom and the large number of users is also not conducive to hardware implementation. How to achieve high transmission rates for a large number of users with limited resources is a problem worthy of research in current 5G NR system communications. Three-dimensional joint efficient resource allocation can greatly improve the system's channel capacity and spectrum efficiency.

[0003] Currently, several classic algorithms are used in existing networks, such as round-robin scheduling, proportional fair scheduling, and first-come, first-served scheduling. These methods employ simple criteria to adapt to limited hardware computing resources and lack interactive feedback with end users before scheduling, often leading to user congestion or multiple retransmissions. Existing theoretical research, however, mostly targets maximum throughput for resource scheduling. This assumes that comprehensive channel information about end users is known before scheduling, and the optimal scheduling method is obtained through extensive iterative searches. Alternatively, a low-complexity heuristic scheduling method is developed based on prior information. Given the computational power of actual communication systems, low-complexity methods are more applicable. However, these approaches often overlook constraints such as users' personalized traffic demands and quality of service requirements during communication.

[0004] In resource scheduling on existing networks, for example, in the paper "Research on Codebook Design Standardization for 5G New Air Interface," the authors proposed an R15 stage 5G new air interface codebook design that basically continues the codebook design ideas of R13 and R14. However, the base station and end users do not generate interactive feedback before scheduling. When multiple users choose the same beam to transmit data, it will cause user congestion or multiple retransmissions. Theoretical research work often uses iterative search to solve the optimal solution, and the computational complexity is within the computational complexity that the hardware can bear. It ignores constraints such as users' personalized traffic requirements and quality of service requirements in communications, and does not design a wireless resource allocation problem model based on multiple constraints. Summary of the Invention

[0005] To overcome the shortcomings of the aforementioned prior art, the present invention aims to provide a space-time-frequency three-dimensional resource allocation method for MU-MIMO systems. This method combines user personalized needs with spatial interference constraints to establish a space-time-frequency multi-dimensional resource allocation optimization problem. A heuristic resource allocation method based on a greedy algorithm is developed to solve the problem in a step-by-step manner. Furthermore, a traffic-aware improved resource allocation method and a new beam selection criterion are proposed, integrating user needs and channel characteristics. This method achieves higher spectral efficiency and higher system sum rates, while also being adaptable to hardware system complexity.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A space-time-frequency three-dimensional resource allocation method for a MU-MIMO system comprises the following steps:

[0008] Step 1: For the MU-MIMO system, a comprehensive multi-dimensional resource optimization problem is established with the goal of maximizing system and rate requirements under the joint constraints of user rate requirements and airspace QoS (quality of service) requirements.

[0009] Step 2: Analyze the comprehensive multi-dimensional resource optimization problem, and use a heuristic resource allocation method to allocate resources while balancing complexity and system performance to improve spectrum efficiency;

[0010] Step 3: To solve the time-frequency resource waste in the resource allocation method in step 2, the heuristic resource allocation method is improved to a traffic-aware heuristic resource allocation method based on the user's personalized traffic requirements;

[0011] Step 4: To improve the spatial reuse capability of the resource allocation method in step 3, a beam selection criterion is proposed by balancing the beam matching performance with the repetition rate selected by the user.

[0012] Step 5: Based on the beam selection criteria described in step 4, a traffic-aware heuristic resource allocation method is used to transmit data for the MU-MIMO system.

[0013] The MU-MIMO system is a MU-MIMO downlink communication system, including a base station equipped with a uniform planar array, which provides services to U single-antenna user terminals at the same time, wherein the base station performs N B beamforming, generating N B Different beam shapes are grouped into beams, and the beam grouping is used to reduce spatial interference; the beam grouping is: N BThe beams are divided into K beam groups, that is, in the same time-frequency unit, the user is only interfered by other beams in the same beam group. The beams between different beam groups rely on occupying different time-frequency resources to eliminate strong spatial interference. Suppose the beam group where beam b is located is g k , the signal-to-interference-noise ratio when the u-th user selects the b-th beam is:

[0014]

[0015] w b represents the codebook of beam b, g k represents the beam group to which beam b belongs. The u-th user schedules the same data on multiple beams to improve reliability. Therefore, the final signal-to-interference-and-noise ratio of the u-th user can be expressed as:

[0016]

[0017] Among them, F u,k ∈{0,1} is a binary variable, when F u,k When it is 1, it means that the u-th user selects beam group g. k Provide services for it, when F u,k When it is 0, it means that the u-th user does not select beam group g. k Provide services for it; u,b ∈{0,1} is a binary variable, when I u,b When it is 1, it means that the u-th user chooses beam b to provide service. u,b When it is 0, it means that the u-th user does not choose beam b to provide service; k represents the k-th beam group.

[0018] In order to formally describe the joint beam selection and resource allocation problem, two decision binary variables are defined: J u,f,t is the binary variable for the decision of the u-th user on the f-th frequency resource in the t-th OFDM symbol. u,f,t When J is 1, the uth user chooses to use the fth frequency resource in the tth OFDM symbol. u,f,t When E is 0, the u-th user does not select the f-th frequency resource in the t-th OFDM symbol; t,k is the beam group g k For the decision binary variable of the tth OFDM symbol, when E t,k When it is 1, the beam group g k Occupies the tth OFDM symbol, when E t,k When it is 0, the beam group g k The tth OFDM symbol is not occupied. Then the actual transmission rate of the uth user is r u Expressed as:

[0019]

[0020] J u,f,t is the binary variable for the decision of the u-th user on the f-th frequency resource in the t-th OFDM symbol. u,f,t When J is 1, the uth user chooses to use the fth frequency resource in the tth OFDM symbol. u,f,t When E is 0, the u-th user does not select the f-th frequency resource in the t-th OFDM symbol; t,k is the beam group g k For the decision binary variable of the tth OFDM symbol, when E t,k When it is 1, the beam group g k Occupies the tth OFDM symbol, when E t,k When it is 0, the beam group g k The tth OFDM symbol is not occupied. Represents the bandwidth of a single schedulable frequency domain unit, SINR u represents the final signal-to-interference-and-noise ratio after the u-th user schedules multiple beams;

[0021] The final result is the actual transmission rate of the u-th user, which is used to characterize the variables in the following optimization problem: u .

[0022] In the step 1, the multi-dimensional resource optimization problem is to set R to represent the multi-user MIMO and rate, d u represents the minimum traffic demand of the u-th user, and T represents the number of OFDM symbols. Even if the user traffic demand remains unchanged, T will change with the change of resource allocation strategy. T is large enough to meet the minimum traffic requirements of all users. In order to maximize the sum rate of all end users, the resource allocation problem in step 1 is expressed as:

[0023]

[0024] The c1 constraint indicates that each symbol in the time domain can only be occupied by one beam group; the c2 constraint indicates that the total signal-to-interference-and-noise ratio (SINR) of each user should be greater than the SINR threshold SINR threshold , to ensure the user's airspace QoS requirements; the c3 constraint indicates that each time-frequency unit can only be occupied by one user or no user; the c2 constraint and c4 constraint are essentially different. The former is to ensure the reliability of transmitted data and select multiple beams to improve the airspace diversity effect, while the latter is a restriction on the number of time-frequency resources to meet the user's minimum traffic requirements.

[0025] The specific process of the heuristic resource allocation method in step 2 is as follows:

[0026] Step 1.1: The prior information is the beam numbers corresponding to all beam groups and the minimum traffic requirements of all users. Users select the best beam in turn based on the greedy algorithm:

[0027] b * =argmaxSINR u,b

[0028] Assumptions In order to satisfy the c2 constraint, the beam group is selected Several beams with the largest SINR transmit signals to the user at the same time, and determine the variable F u,k and I u,b , and then determine the user's transmission rate on a single RB u :

[0029]

[0030] Step 1.2: Allocate RBs to users sequentially, assuming U b represents the set of users who select beam b to transmit data, and the idle users on beam b are RBs are allocated to U b The user u who is not assigned a time-frequency unit, is a function that rounds up to an integer, and also assigns the same number and position of beams to other beams selected by user u in the beam group. RBs. The same operation is performed for each beam.

[0031] Step 1.3: Count the number of RBs M required for beam b in turn b , assuming that the number of RBs contained in a single OFDM symbol is F, determine the beam group g k The number of OFDM symbols required:

[0032]

[0033] Finally, the number of OFDM symbols required to meet all user constraints is determined:

[0034]

[0035] T OFDM symbols are sequentially assigned to K beam groups, and then combined with F u,k and I u,b , sequentially determine which users use which beam shapes on which OFDM symbols for downlink data transmission, and determine the variables

[0036] The traffic-aware heuristic resource allocation method in step 3 is to allocate excess time-frequency resources to users based on their traffic demands. kThe number of occupied OFDM symbols T k After that, beam group g k The number of RBs that can be scheduled is T k F, assuming the user set U b Beam group g is selected k Medium beam b, assigned to U b The number of RBs for user u is:

[0037]

[0038] Among them D b In order to satisfy U b The minimum traffic requirement of all users in beam b is the minimum number of RBs to be allocated:

[0039]

[0040] g k The idle RBs on the first and third beams are extended to the corresponding users according to the demand ratio.

[0041] In step 4, the user selects the optimal beam based on the greedy algorithm in step 2. Although each user selects the optimal beam and obtains the best spatial characteristics, this does not necessarily achieve the optimal spatial utilization of the current cell. This is because within the same beam group, multiple beams may have very similar SINRs on the user's NLOS path, but the user may not choose an idle beam to transmit data, thereby reducing the system's spatial multiplexing capability.

[0042] The beam selection criteria are used to balance the beam matching performance with the repetition rate selected by the user to improve spatial multiplexing capabilities. The beam selection criteria are:

[0043]

[0044] in

[0045]

[0046] Its physical meaning is that when user u matches beam b, SINR u,b The ratio of the number of users that have been matched on the current beam b is different from the ratio of the number of users that have been matched on the current beam b. u,b Compared with the maximization criterion, As a criterion for selecting beams, it is more objective because Also taking into account the repetition rate of beam b selected by the user, It gives the opportunity to select beams with similar SINR performance but idle. At the same time, the fewer users choose the same beam b, the less time-frequency resources are needed to orthogonalize the users who choose b, which can reduce b. *Beam group Requires T k ,Therefore, time domain resources are further saved.

[0047] Beneficial effects of the present invention:

[0048] This paper combines user personalized needs with constraints such as limited spatial interference to establish a space-time-frequency multi-dimensional resource allocation optimization problem. Based on the concept of a greedy algorithm, it derives a heuristic resource allocation method that solves the problem in a step-by-step manner. Furthermore, it integrates user needs and channel characteristics to propose an improved traffic-aware resource allocation method and a new beam selection criterion. While ensuring communication effectiveness and reliability, it enables the system to serve more active users, achieving higher spectrum efficiency and higher system sum rates, and is adaptable to hardware system complexity.

[0049] This invention aims to maximize system summation and rate, taking into account user personalized traffic requirements. It proposes a space-time-frequency three-dimensional resource allocation method for MU-MIMO systems. This method improves the system's signal-to-interference-and-noise ratio, spectrum efficiency, and spatial domain multiplexing capabilities during communication, further improving system summation and rate. Any equivalent variations of the algorithm described in this invention are intended to be included within the scope of this patent application. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of a space-time-frequency three-dimensional resource allocation method for a MU-MIMO system provided by an embodiment of the present invention.

[0051] Figure 2 This is a schematic diagram of the time-frequency unit architecture based on the 5G-NR protocol provided by an embodiment of the present invention.

[0052] Figure 3 It is a schematic diagram comparing the heuristic resource allocation method provided by an embodiment of the present invention and the traffic-aware heuristic resource allocation method.

[0053] Figure 4 This is a simulation result diagram of the influence of the SINR threshold provided by the embodiment of the present invention on the number of OFDM symbols occupied by different resource allocation methods.

[0054] Figure 5 This is a simulation result diagram showing the impact of the SINR threshold on different resource allocation methods and rates provided by an embodiment of the present invention.

[0055] Figure 6 This is a simulation result diagram of the impact of the number of users on different resource allocation methods and rates provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The present invention will be described in further detail below with reference to the accompanying drawings.

[0057] like Figure 1 As shown, the space-time-frequency three-dimensional resource allocation method for a MU-MIMO system provided in an embodiment of the present invention includes the following steps:

[0058] Step 1: For the MU-MIMO system, a comprehensive multi-dimensional resource optimization problem is established with the goal of maximizing system and rate requirements under the joint constraints of user rate requirements and airspace QoS (quality of service) requirements.

[0059] Step 2: Analyze the comprehensive multi-dimensional resource optimization problem and develop a heuristic resource allocation method based on a greedy algorithm to improve spectrum efficiency while balancing complexity and system performance.

[0060] Step 3: To solve the time-frequency resource waste in the resource allocation method in step 2, the heuristic resource allocation method is improved to a traffic-aware heuristic resource allocation method based on the user's personalized traffic requirements;

[0061] Step 4: To improve the spatial reuse capability of the resource allocation method in step 3, a beam selection criterion is proposed by balancing the beam matching performance with the repetition rate selected by the user.

[0062] Step 5: Based on the beam selection criteria described in step 4, a traffic-aware heuristic resource allocation method is used to transmit data for the MU-MIMO system.

[0063] The time-frequency unit architecture of the entire communication system is based on the 5G-NR protocol, and its configuration is as follows: Figure 2 As shown. A frame is 10ms long and contains 10 subframes; each subframe contains 2 time slots; each time slot consists of 14 OFDM symbols; we regard the minimum granularity that can be allocated in the time domain as an OFDM symbol, and define a single subcarrier on an OFDM symbol as a resource element (RE), with a subcarrier spacing of 30kHz. In the 5G protocol, 12 consecutive subcarriers in the frequency domain are defined as a resource block (RB), and its bandwidth is 360kHz. RB is the smallest frequency unit that can be allocated to any user in 5G NR. The system bandwidth is 20MHz. Excluding the guard interval, etc., the number of RBs contained in a single OFDM symbol is 51.

[0064] MU-MIMO downlink communication system, the system includes a base station equipped with a uniform planar array, the base station simultaneously provides services for U single-antenna user terminals, where the base station can perform N B beamforming, generating N B Different beam shapes.

[0065] In order to reduce spatial interference, the present invention introduces beam groups. The beam groups are: BBeams are divided into K beam groups. That is, in the same time-frequency unit, users are only subject to interference from other beams within the same beam group. Beams between different beam groups eliminate strong spatial interference by occupying different time-frequency resources.

[0066] Therefore, the signal-to-interference-and-noise ratio when the u-th user selects the b-th beam is:

[0067]

[0068] w b represents the codebook of beam b. g k represents the beam group to which beam b belongs. To improve reliability, the u-th user can schedule the same data on multiple beams. Therefore, the final signal-to-interference-and-noise ratio of the u-th user can be expressed as:

[0069]

[0070] Among them, SINR u,b represents the signal-to-interference-and-noise ratio when the u-th user selects the b-th beam, F u,k ∈{0,1} is a binary variable indicating whether the u-th user selects beam group g k Provide services for it. u,b ∈{0,1} is also a binary variable, indicating whether the u-th user chooses beam b to provide service.

[0071] In order to formally describe the joint beam selection and resource allocation problem, two decision binary variables are defined: J u,f,t is a binary variable indicating whether the u-th user chooses to use the f-th frequency resource in the t-th OFDM symbol; E t,k Represents the bundle group g k Whether it occupies the tth OFDM symbol alone. Then the actual transmission rate of the uth user is r u It can be expressed as:

[0072]

[0073] In step 1, a multi-dimensional resource optimization problem is established. Assume that R represents the system sum rate, d u represents the minimum traffic requirement of the uth user, and T represents the number of OFDM symbols. Even if the user traffic requirement remains unchanged, T will change with the resource allocation strategy. In the present invention, T is large enough to meet the minimum traffic requirements of all users. In order to maximize the sum rate of all end users, the resource allocation problem can be expressed as:

[0074]

[0075] The c1 constraint indicates that each symbol in the time domain can only be occupied by one beam group; the c2 constraint indicates that the total signal-to-interference-and-noise ratio (SINR) of each user should be greater than the SINR threshold SINR threshold , to ensure users' spatial QoS requirements; the c3 constraint indicates that each time-frequency unit can only be occupied by one user or no user; the c2 and c4 constraints are fundamentally different. The former ensures data transmission reliability by selecting multiple beams to enhance spatial diversity, while the latter imposes a constraint on the number of time-frequency resources to meet users' minimum traffic requirements. Clearly, the various variables to be optimized in this optimization problem, including subcarrier allocation, time-domain resource allocation, and beam resources, constitute a non-convex mixed integer linear programming (NM-ILP) problem.

[0076] For the optimization problem described in step 1, a heuristic resource allocation method is developed based on a greedy algorithm to improve spectrum efficiency while balancing complexity and system performance. Based on the characteristics of narrowband subcarriers in the OFDM structure, it is assumed that the channel within multiple consecutive symbols is flat fading and that the channel within multiple consecutive symbols is time-invariant. The user's transmission rate is affected only by the number of allocated RBs, and is not affected by the allocation time sequence or subcarrier sequence. The specific process of the heuristic resource allocation method is as follows:

[0077] Step 1.1: The prior information is the beam numbers corresponding to all beam groups and the minimum traffic requirements of all users. Users select the best beam in turn based on the greedy algorithm:

[0078] b * =argmaxSINR u,b

[0079] Assumptions In order to satisfy the c2 constraint, the beam group is selected Several beams with the largest SINR transmit signals to the user at the same time, and determine the variable F u,k and I u,b Then determine the user's transmission rate on a single RB. u :

[0080]

[0081] Step 1.2: Allocate RBs to users sequentially. Assume U b represents the set of users who select beam b to transmit data, and the idle users on beam b are RBs are allocated to U b The user u who is not assigned a time-frequency unit, is a function that rounds up to an integer, and also assigns the same number and position of beams to other beams selected by user u in the beam group. RBs. The same operation is performed for each beam.

[0082] Step 1.3: Count the number of RBs M required for beam b in turn b , assuming that the number of RBs contained in a single OFDM symbol is F, determine the beam group g k The number of OFDM symbols required:

[0083]

[0084] Finally, the number of OFDM symbols required to meet all user constraints is determined:

[0085]

[0086] T OFDM symbols are sequentially assigned to K beam groups, and then combined with F u,k and I u,b , sequentially determine which users use which beam shapes on which OFDM symbols for downlink data transmission, and determine the variables

[0087] This low-complexity algorithm can obtain a feasible solution to the resource allocation optimization problem. However, in step 1.2, the operation of rounding up may make T k There are unassigned RBs in the last OFDM symbol. k In the process of taking the maximum operation, although it ensures that the user set U on each beam b This meets the traffic demand of users in the area, but also results in idle OFDM symbols in some beams.

[0088] Furthermore, although each user selects the optimal beam and achieves optimal spatial characteristics, this does not necessarily achieve optimal spatial utilization for the current cell. This is because within the same beam group, multiple beams may have very similar SINRs on the user's NLOS path, but the user may not choose an idle beam for data transmission, thus reducing the system's spatial multiplexing capability.

[0089] Based on a heuristic resource allocation method based on a greedy algorithm, we address the above two shortcomings. Considering the waste of time and frequency resources, we design a traffic-aware heuristic resource allocation method. To further improve spatial reuse capabilities, we propose a new and more accurate beam selection criterion by balancing beam matching performance with the repetition rate selected by the user.

[0090] The traffic-aware heuristic resource allocation method in step 3 is to allocate excess time-frequency resources to users based on their traffic demands. k The number of occupied OFDM symbols T k After that, beam group gk The number of RBs that can be scheduled is T k F. Assume that the user set U b Beam group g is selected k Medium beam b, assigned to U b The number of RBs for user u is:

[0091]

[0092] Among them D b In order to satisfy U b The minimum traffic requirement of all users in beam b is the minimum number of RBs to be allocated:

[0093]

[0094] The schematic diagram of the heuristic resource allocation method and the traffic-aware heuristic resource allocation method is as follows: Figure 3 As shown, it can be seen that g k The idle RBs on the first and third beams are extended to the corresponding users according to the demand ratio.

[0095] Furthermore, the new beam selection criterion is:

[0096]

[0097] in

[0098]

[0099] Its physical meaning is that when user u matches beam b, SINR u,b The ratio of the number of users that have been matched on the current beam b. u,b Compared with the maximization criterion, As a criterion for selecting beams, it is more objective. The repetition rate of beam b selected by the user is also taken into account. This gives idle beams with similar SINR performance an opportunity to be selected. At the same time, the fewer users choose the same beam b, the less time-frequency resources are needed to orthogonalize the users who choose b, which may reduce b. * Beam group Requires T k Therefore, time domain resources are further saved.

[0100] To demonstrate the performance improvement, the following four resource allocation schemes are used: Scheme 1: Ignoring the OFDM structure of the NR system, a heuristic resource allocation method based on a greedy algorithm is used; Scheme 2: Considering the OFDM architecture, a heuristic resource allocation method based on a greedy algorithm is used; Scheme 3: A traffic-aware heuristic resource allocation method is used; Scheme 4: Applying a new beam selection criterion and a traffic-aware heuristic resource allocation method are used.

[0101] Assuming the number of users is 20, as the SINR threshold increases, Figure 4 In this paper, the present invention demonstrates the change in time resources occupied by four different schemes. Regardless of how the SINR threshold changes, the OFDM characteristics of the NR system can reduce the total number of symbols required. This is because, in schemes that do not consider the OFDM structure, users selecting the same beam can only be distinguished by time domain resources. However, the minimum schedulable unit of the OFDM structure is more refined. As the SINR threshold increases, the effect of the new beam selection criterion can be analyzed in three parts. First, within the SINR threshold range of 2dB to 6dB, the SINR threshold is low, and users can meet the threshold requirement by selecting beams with very low SINRs. The SINR corresponding to these beams is significantly lower than that of the optimal beam. Although the new beam selection criterion reduces the beam selection repetition rate, the increased spatial multiplexing capability is insufficient to offset the SINR gap. Therefore, under the same traffic demand, the new beam selection criterion actually requires more OFDM symbols. As the SINR increases from 6dB to 10dB, the number of selectable beams that meet the SINR threshold decreases. This means that the SINRs corresponding to the user's selectable beams are relatively close. The performance of the new beam selection criterion is gradually being demonstrated. Finally, when the SINR threshold just exceeds 10dB, the number of required OFDM symbols suddenly increases. This is because some users require more beams to meet the SINR threshold. This also leads to a higher beam selection repetition rate, requiring more RBs to distinguish users selecting the same beam. However, when the SINR threshold changes from 10dB to 13dB, the new beam selection accuracy still reduces the number of occupied symbols.

[0102] As the SINR threshold increases, Figure 5 In this paper, the present invention demonstrates the sum rate variation on a single OFDM symbol using four different schemes. The OFDM characteristics of the NR system, the adaptive RA algorithm, and the new beam selection criterion all improve the total achievable transmission rate on a single OFDM symbol. The adaptive RA algorithm achieves the most significant performance improvement, followed by the new beam selection criterion. Furthermore, the new beam selection criterion is more suitable for scenarios with a large number of users.

[0103] Assume that the number of users increases from 10 to 50. Figure 6In this paper, the present invention demonstrates the sum rate evolution for a single OFDM symbol using four different schemes. Clearly, OFDM characteristics, the adaptive RA algorithm, and the novel beam selection criteria all improve the system's achievable rate per symbol. The adaptive RA algorithm delivers the greatest performance gain. When the number of users exceeds 15, the achievable rate per symbol is increased to over 50%.

Claims

1. A space-time-frequency three-dimensional resource allocation method for a MU-MIMO system, characterized in that: The following steps are included: Step 1: For the MU-MIMO system, under the joint constraints of user rate requirements and airspace QoS requirements, a comprehensive multi-dimensional resource optimization problem is established with the goal of maximizing system and rate. Step 2: Analyze the comprehensive multi-dimensional resource optimization problem, and use a heuristic resource allocation method to allocate resources while balancing complexity and system performance to improve spectrum efficiency; Step 3: To solve the time-frequency resource waste in the resource allocation method in step 2, the heuristic resource allocation method is improved to a traffic-aware heuristic resource allocation method based on the user's personalized traffic requirements; The traffic-aware heuristic resource allocation method in step 3 is to allocate excess time-frequency resources to users based on their traffic demands. k The number of occupied OFDM symbols T k After that, beam group g k The number of RBs that can be scheduled is T k F, assuming the user set U b Beam group g is selected k Medium beam b, assigned to U b The number of RBs for user u is: d u represents the minimum traffic demand of the u-th user, where D b In order to satisfy U b The minimum traffic requirement of all users in beam b is the minimum number of RBs to be allocated: g k The idle RBs on the first and third beams are extended to the corresponding users according to the demand ratio; The specific process of the heuristic resource allocation method in step 2 is as follows: Step 1.1: The prior information is the beam numbers corresponding to all beam groups and the minimum traffic requirements of all users. Users select the best beam in turn based on the greedy algorithm: b * =argmaxSINR u,b Assumptions In order to meet the c2 constraint, the c2 constraint indicates that the total signal to interference and noise ratio (SINR) of each user should be greater than the SINR threshold SINR. threshold , select beam group Several beams with the largest SINR transmit signals to the user at the same time, and determine the binary variable F u,k and binary variable I u,b , and then determine the user's transmission rate l on a single RB u :SINR u represents the final signal-to-interference-and-noise ratio after the u-th user schedules multiple beams; Represents the bandwidth of a single schedulable frequency domain unit, SINR u represents the final signal-to-interference-and-noise ratio after the u-th user schedules multiple beams; Step 1.2: Allocate RBs to users sequentially, assuming U b represents the set of users who select beam b to transmit data, and the idle users on beam b are RBs are allocated to U b The user u who is not assigned a time-frequency unit, is a function that rounds up to an integer, and also assigns the same number and position of beams to other beams selected by user u in the beam group. RBs; each beam performs the same operation; Step 1.3: Count the number of RBs M required for beam b in turn b , assuming that the number of RBs contained in a single OFDM symbol is F, determine the beam group g k The number of OFDM symbols required: Finally, the number of OFDM symbols required to meet all user constraints is determined: T OFDM symbols are sequentially assigned to K beam groups, and then combined with F u,k and I u,b , sequentially determine which users use which beam shapes on which OFDM symbols for downlink data transmission, and determine the variables J u,f,t is the decision binary variable of the u-th user on the f-th frequency resource in the t-th OFDM symbol.

2. The space-time-frequency three-dimensional resource allocation method for MU-MIMO system according to claim 1, characterized in that: The MU-MIMO system is a MU-MIMO downlink communication system, including a base station equipped with a uniform planar array, which provides services to U single-antenna user terminals at the same time, wherein the base station performs N B beamforming, generating N B Different beam shapes are grouped into beams, and the beam grouping is used to reduce spatial interference; the beam grouping is: N B The beams are divided into K beam groups, that is, in the same time-frequency unit, the user is only interfered by other beams in the same beam group. The beams between different beam groups rely on occupying different time-frequency resources to eliminate spatial interference. Suppose the beam group where beam b is located is g k , the signal-to-interference-noise ratio when the u-th user selects the b-th beam is: w b represents the codebook of beam b, g k represents the beam group to which beam b belongs. The u-th user schedules the same data on multiple beams to improve reliability. The final signal-to-interference-and-noise ratio of the u-th user is expressed as: Among them, F u,k ∈{0,1} is a binary variable, when F u,k When it is 1, it means that the u-th user selects beam group g. k Provide services for it, when F u,k When it is 0, it means that the u-th user does not select beam group g. k Provide services for it; u,b ∈{0,1} is a binary variable, when I u,b When it is 1, it means that the u-th user chooses beam b to provide service. u,b When it is 0, it means that the u-th user does not choose beam b to provide service; k represents the k-th beam group; Define two decision binary variables: J u,f,t is the binary variable for the decision of the u-th user on the f-th frequency resource in the t-th OFDM symbol. u,f,t When J is 1, the uth user chooses to use the fth frequency resource in the tth OFDM symbol. u,f,t When E is 0, the u-th user does not select the f-th frequency resource in the t-th OFDM symbol; t,k is the beam group g k For the decision binary variable of the tth OFDM symbol, when E t,k When it is 1, the beam group g k Occupies the tth OFDM symbol, when E t,k When it is 0, the beam group g k The tth OFDM symbol is not occupied; the actual transmission rate of the uth user is r u Expressed as: Represents the bandwidth of a single schedulable frequency domain unit, SINR u represents the final signal-to-interference-and-noise ratio after the u-th user schedules multiple beams; The final result is the actual transmission rate of the u-th user, which is used to represent the variables in the optimization problem: r u .

3. The space-time-frequency three-dimensional resource allocation method for MU-MIMO system according to claim 2, characterized in that: In the step 1, the multi-dimensional resource optimization problem is to set R to represent the multi-user MIMO and rate, d u represents the minimum traffic demand of the u-th user, and T represents the number of OFDM symbols. Even if the user traffic demand remains unchanged, T will change with the change of resource allocation strategy. T is large enough to meet the minimum traffic requirements of all users. In order to maximize the sum rate of all end users, the resource allocation problem in step 1 is expressed as: The c1 constraint indicates that each symbol in the time domain can only be occupied by one beam group; the c2 constraint indicates that the total signal to interference and noise ratio (SINR) of each user should be greater than the SINR threshold SINR threshold , to ensure the user's airspace QoS requirements; the c3 constraint means that each time-frequency unit can only be occupied by one user or no user.

4. The space-time-frequency three-dimensional resource allocation method for MU-MIMO system according to claim 1, characterized in that: Step 4: To improve the spatial multiplexing capability of the resource allocation method in step 3, a beam selection criterion is proposed by balancing the matching performance of the beams with the repetition rate selected by the user.

5. The space-time-frequency three-dimensional resource allocation method for MU-MIMO system according to claim 4, characterized in that: In step 4, the beam selection criteria are used to balance the beam matching performance and the repetition rate selected by the user to improve the spatial multiplexing capability. The beam selection criteria are: in Its physical meaning is that when user u matches beam b, SINR u,b The ratio of the number of users that have been matched on the current beam b is different from the ratio of the number of users that have been matched on the current beam b. u,b Compared with the maximization criterion, As a criterion for selecting beams, it is more objective because Also taking into account the repetition rate of beam b selected by the user, It gives the opportunity to select beams with similar SINR performance but idle. At the same time, the fewer users choose the same beam b, the less time-frequency resources are needed to orthogonalize the users who choose b, which can reduce b. * Beam group Requires T k ,Therefore, time domain resources are further saved.

6. The space-time-frequency three-dimensional resource allocation method for MU-MIMO system according to claim 5, characterized in that: Step 5: Based on the beam selection criteria described in step 4, a traffic-aware heuristic resource allocation method is used to transmit data for the MU-MIMO system.

Images