Statistical prior assisted multi-user scheduling method

Abstract

The invention provides a statistical prior assisted multi-user scheduling method, which comprises the following steps of: firstly, establishing a statistical prior channel knowledge map from a user geographic position to channel state information based on regional division under the coverage of a base station and long-term observed channel data, and evaluating the reliability of statistical prior; and then selecting scheduling users by adopting a two-stage user screening method. In the first stage, all sub-regions covered by a base station are grouped by using statistical prior information, and a preliminary active user set is obtained by using the groups. And in the second stage, the correlation between the users is taken as a criterion, and orthogonal users are selected by using an angle domain power spectrum. In the first stage, a limited number of active users can be selected under the scene of massive users, and the pilot frequency overhead of channel estimation is reduced. In the second stage, an effective orthogonal user set can be selected by using instantaneous channel information and statistical prior, and the whole algorithm has relatively low calculation complexity.

Description

Technical Field

[0001] The present invention belongs to the field of communications, and particularly relates to a multi-user scheduling method assisted by statistical prior information. Background Art

[0002] In the context of the era of all things intelligent connection, the demand for massive data interaction drives the continuous improvement of the transmission efficiency of wireless networks. By deploying a high-density antenna array on the base station side, an exponential increase in channel capacity can be achieved without increasing the existing spectrum bandwidth and transmission power.

[0003] User scheduling refers to selecting appropriate users for data transmission in a multi-user communication system to maximize system performance. Generally, user scheduling algorithms need to make full use of the channel state information of the wireless environment where users are located to achieve interference coordination and resource optimization allocation during multi-user pairing. Traditional user scheduling methods rely on the accurate acquisition of real-time channel state information, and evaluate the channel quality and mutual interference degree of users through a periodic channel estimation and feedback mechanism. However, in the scenario of a large-scale antenna array, with the increase in the number of antennas and the scale of users, the complexity of estimating, feedbacking, and processing complete instantaneous channel state information increases exponentially, resulting in a significant increase in system latency and resource consumption. In addition, existing scheduling strategies fail to fully exploit the long-term statistical prior information related to user locations, resulting in the potential value not being fully utilized. Summary of the Invention

[0004] Object of the Invention: Aiming at the problem that traditional scheduling mechanisms highly rely on real-time channel state information, the object of the present invention is to propose a user scheduling method assisted by statistical prior information, which improves the resource efficiency of multi-user scheduling by deeply exploring the potential value of statistical prior information, and at the same time reduces the system's dependence on real-time channel information, thereby optimizing the performance of the overall communication system.

[0005] Technical Solution: To achieve the above object of the invention, a multi-user scheduling method based on statistical prior information provided by the present invention includes the following steps:

[0006] Divide the coverage area of the base station into multiple sub-regions, collect the sampled channel information obtained from long-term observations within the sub-regions, and construct a channel knowledge map, including the statistical angular power spectrum, statistical correlation, and statistical channel gain;

[0007] Calculate the variance of the statistical prior information within each sub-region, and evaluate the reliability of the sub-region statistical prior information according to the variance threshold;

[0008] Group the sub-regions according to the statistical channel prior information, meet the orthogonality requirements between the sub-regions within the same group, and use the grouping to initially screen the active user set;

[0009] For active users in the unreliable region, their instantaneous channel information is measured; for active users in the reliable region, the statistical prior of the sub-region is used as their statistical prior, and the user angular power spectrum, inter-user correlation, and user channel gain are updated.

[0010] With the criterion of minimizing channel correlation, the angular power spectrum combining the instantaneous channel information and the statistical prior information is used to select the set of users to be finally scheduled from the set of active users.

[0011] Furthermore, the statistical angular power spectrum of sub-region l is the mean of the angular power spectra of the observed sampled channels; the statistical correlation between sub-regions l1 and l2 is the degree of similarity of the statistical angular power spectra between sub-regions; the statistical channel gain of sub-region l is the mean of the channel gains of the observed sampled channels.

[0012] Furthermore, the reliability σ of the statistical prior of sub-region l l is obtained by calculating the variance of the correlation values of all sampled channels with the channel at the region center. The correlation of the angular power spectrum between the s-th sampled channel in sub-region l and the channel at the region center where ω l,c and ω l,s represent the angular power spectrum of the channel at the region center and the angular power spectrum of the s-th sampled channel, respectively.

[0013] Furthermore, according to the statistical channel prior information, the L sub-regions are grouped into N G groups. The steps include:

[0014] Step1) Initialize i = 1, let the grouped regions the ungrouped regions

[0015] Step2) Initialize the i-th group Select the region with the largest statistical channel gain and put it into the group the temporary variable of the grouped regions

[0016] Step3) Search for the regions in that meet the conditions where represents the statistical correlation, and α is used to limit the orthogonality between users, and put it into the group Repeat this step until there are no sub-regions in that meet the conditions;

[0017] Step4) Search for The area that meets the conditions And put it into a group Repeat this step until There is no sub-area that meets the conditions;

[0018] Step5) Let i = i + 1, and repeat Step2) until i > N G Or

[0019] Furthermore, the method for initially screening the active user set using sub-area grouping is: map the user location to the area index where the user is located Statistically calculate the user coverage of each group, select the group with the highest coverage, and output the users covered by this group as the active user set

[0020] Furthermore, based on the active user set, using the location coordinates of the users, map the users to the index of the sub-area where they are located; when the variance of the sub-area where the user is located is less than the set threshold, the user is considered to be in a reliable area; when the variance of the sub-area where the user is located is greater than or equal to the set threshold, the user is considered to be in an unreliable area.

[0021] Furthermore, update the user angular power spectrum, user - to - user correlation, and user channel gain according to the statistical prior reliability of the area where user k is located;

[0022] The angular power spectrum of user k is expressed as

[0023]

[0024] The channel gain of user k is expressed as

[0025]

[0026] The correlation between user k and user j is expressed as

[0027]

[0028] Where Is the reliable area, Is the unreliable area, Are the indices of the sub - areas where user k and user j are located respectively, Respectively represent the statistical angular power spectrum, statistical correlation, and statistical channel gain corresponding to the area index, h k 、 Respectively represent the channel state information in the spatial domain and angular domain obtained through uplink channel estimation, and ⊙ is the Hadamard product.

[0029] Furthermore, the base station starts from the select from users for scheduling. The steps include:

[0030] Step1) Initialize the set scheduling user set

[0031] Step2) For select a user with low correlation and high gain with the selected users and where g k and g s represent the channel gains of user k and user s respectively, and ρ k,s represents the correlation between user k and user s;

[0032] Step3) If the number of selected users is less than then update the user set to make the orthogonality between users satisfy where α is used to limit the orthogonality between users; if is non-empty and the number of elements in is less than then let i = i + 1 and repeat Step2) until the algorithm ends.

[0033] The present invention also provides a computer system, including a memory, a processor, and a computer program / instructions stored on the memory and executable on the processor. When the computer program / instructions are executed by the processor, the steps of the above-mentioned multi-user scheduling method assisted by statistical prior are implemented.

[0034] The present invention also provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by the processor, the steps of the above-mentioned multi-user scheduling method assisted by statistical prior are implemented.

[0035] Beneficial effects: By dividing the base station coverage area and observing the historical channel state information of each sub-area for a long time, the present invention constructs a mapping relationship from geographical location to statistical prior information - a channel knowledge map. Further, by using the variance of the statistical prior information and setting a variance threshold, all sub-areas are divided into areas with reliable statistical prior information and areas with unreliable statistical prior information. This scheduling method adopts a two-stage user selection strategy to achieve efficient scheduling: The first stage: Group the sub-areas based on the statistical prior information to ensure good orthogonality among the sub-areas within the same group. On this basis, a batch of active users are screened out to form an active user set. In this stage, by using the statistical prior information, the number of users that need to be further processed is effectively reduced, and the pilot overhead of the system is significantly reduced. The second stage: With the minimization of channel correlation as the criterion, combining the instantaneous channel information of the users and the statistical prior information, the final scheduled users are further selected from the active user set. This stage makes full use of the synergistic advantages of the instantaneous channel information and the statistical prior information to ensure the efficiency and accuracy of user selection. Compared with the existing user scheduling technologies, the present invention realizes user scheduling in two stages: In the first stage, a large number of users are quickly screened out by using the statistical prior information, greatly reducing the burden of the system pilot overhead; in the second stage, the instantaneous channel information and the statistical prior information of the users are synergistically used in the active user set for refined selection; this method overcomes the high dependence on real-time channel state information and significantly improves the resource efficiency of multi-user scheduling. Description of the Drawings

[0036] Figure 1 It is a schematic diagram of the sub-area division of the multi-user communication system in the embodiment of the present invention;

[0037] Figure 2 It is the overall flowchart of the method in the embodiment of the present invention;

[0038] Figure 3 It is the detailed method flowchart of the embodiment of the present invention. Detailed Embodiment

[0039] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be described in detail below with reference to the accompanying drawings in the embodiments of the present invention.

[0040] A statistical prior-assisted multi-user scheduling method disclosed in an embodiment of the present invention Figure 1It is a system schematic diagram. The wireless communication transmission system is a MU-MISO system, where the base station is equipped with M antennas, and there are K single-antenna terminals in the cell coverage area. The base station selects K user terminals for data transmission through a user scheduling scheme. During the operation of the system, first, the coverage area of the base station is divided into several sub-areas. Subsequently, the system collects the historical channel state information in each sub-area through the uplink channel. After processing these historical channel state information, a channel knowledge map is formed. This channel knowledge map is stored in the server in the form of a database for subsequent use by the scheduling algorithm.

[0041] Figure 2 The flowchart of this embodiment is shown. For the statistical prior-assisted multi-user scheduling method, first, the coverage area of the base station is divided, and the historical channel state information of each sub-area is observed for a long time, and then a mapping relationship from geographical location to statistical prior information - the channel knowledge map is constructed. Based on the variance of the statistical prior information, a variance threshold is set to divide all sub-areas into two categories: areas with reliable statistical prior and areas with unreliable statistical prior. Subsequently, the sub-areas are grouped using the statistical prior information to ensure good orthogonality within the same group. On this basis, a batch of active users is further selected to form an active user set. In the second stage, with the criterion of minimizing channel correlation, the instantaneous channel information of users in the unreliable area and the statistical channel information of users in the reliable area are jointly processed to further select the final scheduled users from the active user set. The following combines Figure 3 , and describes the detailed steps of the statistical prior-assisted multi-user scheduling method as follows:

[0042] Step 1: Divide the coverage area of the base station into multiple sub-areas. Collect the sampled channel information obtained by long-term observation in each sub-area through the uplink channel, and construct a channel knowledge map, including the statistical angular power spectrum, statistical correlation, and statistical channel gain.

[0043] Specifically, the statistical angular power spectrum of the sub-area can be characterized by the mean of the angular power spectrum of the observed sampled channel; the statistical correlation can be characterized by the similarity degree of signals between different sub-areas, including the similarity degree of the statistical angular power spectrum between sub-areas; the statistical channel gain of the sub-area can be characterized by the mean of the channel gain of the observed sampled channel.

[0044] Exemplarily, considering a single-cell large-scale MIMO system with the number of antennas equipped at the base station being M. First, at preset intervals d x and d y , the coverage range of the base station is divided into L square sub-areas of d x ×d y The regional center is {c1, c2,..., c L}. It should be noted that in addition to the above simple division, in some embodiments, the sub-region division interval can also be dynamically adjusted according to the type of propagation environment and the historical user distribution density. For example, the division interval in the urban area scenario is smaller than that in the open area scenario.

[0045] Suppose there are S observed historical sampled channels {h l,1 ,..., h l,S} in the channel of sub-region l. Then the statistical angular power spectrum of sub-region l can be expressed as:

[0046]

[0047] where is the angular power spectrum of the s-th sampled channel, which describes the energy distribution of the channel in the angular domain, and the expression is where is the channel state information in the angular domain, is the DFT matrix, is the s-th spatial domain sampled channel state information of sub-region l, and ⊙ is the Hadamard product. The statistical correlation describes the similarity degree of signals between different sub-regions. The statistical correlation between sub-regions l1 and l2 can be expressed as:

[0048]

[0049] The statistical channel gain, which reflects the change in power after the signal is transmitted through the wireless channel, can be expressed as:

[0050]

[0051] Using the above statistical angular power spectrum, statistical correlation, and statistical channel gain, a channel knowledge map for user scheduling can be constructed.

[0052] Step 2: Calculate the variances {σ1, σ2,..., σ L} of the statistical prior information in each sub-region according to the long-term observed channel data. The reliability of the statistical prior of sub-region l is obtained by calculating the variance of the correlation values between all sampled channels and the channel at the regional center, which is expressed as:

[0053]

[0054] where represents the mean of the correlation between the S sampled channels in the sub-region and the channel at the regional center, represents the correlation between the s-th sampled channel in sub-region l and the channel at the regional center, and the expression is:

[0055]

[0056] where ω l,c and ω l,s respectively represent the angular power spectrum of the regional center channel and the angular power spectrum of the sth sampled channel. Set the variance threshold σ. If σ l ≥σ, it is determined that the statistical prior information of sub-region l is unreliable; if σ l <σ, it is determined that the statistical prior information of sub-region l is reliable. The region satisfying is called the reliable region, and the region satisfying is called the unreliable region.

[0057] Step 3: According to the statistical channel prior information, group the sub-regions so that the sub-regions within the same group have good orthogonality (i.e., meet the orthogonality requirements). Further, use the grouped initial screening user set to select the active user set from the total user set

[0058] Specifically, based on the statistical channel prior information, group the sub-regions into N G groups, and the core adopts a double-layer iterative search strategy: First, select the optimal sub-region that meets the spatial correlation constraint from the ungrouped region set , and then perform secondary optimization screening in the grouped region set . The specific steps are as follows:

[0059] Step1) Initialize i = 1, let the grouped region the ungrouped region

[0060] Step2) Initialize the ith group Select the region with the largest gain and put it into the group Update the temporary variable of the grouped region

[0061] Step3) Search for the regions that meet the conditions in and put them into the group Repeat this step until there are no sub-regions that meet the conditions in .

[0062] Step4) Search for the regions that meet the conditions in and put them into the group Repeat this step until there are no sub-regions that meet the conditions in .

[0063] Step 5) Let \(i = i + 1\), and repeat Step 2) until \(i\gt N\). G or where \(\alpha\) is a positive number between 0 and 1, used to limit the orthogonality between users. Both \(\alpha\) and \(N\) G two hyperparameters need to be set reasonably. The smaller \(\alpha\) is, the stricter the orthogonality constraint between users is. If it is too small, the number of sub-regions in each group will be small. Generally, it can be set to about 0.8. Similarly, G if \(N\) is too small, some sub-regions will not be included in the grouping. When setting, it is necessary to ensure that the vast majority of sub-regions are included in the grouping. Considering a multi-user transmission system with \(K\) mobile users, according to the regional index where the users are located statistically calculate the user coverage of each group select the group with the highest coverage output the users covered by this group as the active user set The number of users in the active user set is denoted by .

[0064] Step 4, for the users in , use the real-time user location information to distinguish whether the user is in a reliable area or an unreliable area. For users in the unreliable area, measure their instantaneous channel information; for users in the reliable area, use the statistical prior of the sub-region as their statistical prior.

[0065] Use the location coordinates of user \(k\) to determine the area where the \(k\)-th user is located, and map the user to the index of the sub-region where it is located.

[0066]

[0067] On this basis, determine whether the statistical prior of this area is reliable. When , the variance of the statistical prior is small, and it is considered that user \(k\) is in the reliable area; when , the variance of the statistical prior is large, and it is considered that user \(k\) is in the unreliable area.

[0068] Furthermore, according to whether the statistical prior of the area where user \(k\) is located is reliable or not, update the user angular power spectrum, the correlation between users, and the user channel gain. The correlation between users can be characterized by the similarity degree of the angular power spectra between users, and the specific representation is as follows:

[0069] The angular power spectrum of user \(k\) can be expressed as

[0070]

[0071] The channel gain of user \(k\) can be expressed as

[0072]

[0073] The correlation between user k and user j can be expressed as

[0074]

[0075] Step 5: Taking the minimization of channel correlation as the criterion, extract the angular power spectra of the instantaneous channel and the statistical prior, and select users with small power spectrum correlation as the scheduled user set

[0076] Based on the set of active users selected in Step 3, the base station selects users from users for scheduling. First, select one user to be included in the scheduling set according to the channel gain preferentially; subsequently, perform iterative user expansion based on spatial correlation and channel gain; the user selection scheme is described as follows:

[0077] Step1) Initialize the set

[0078] Step2) For Select a user with small correlation and high gain with the selected users And

[0079] Step3) If the number of selected users is less than Then update the user set To make the orthogonality between users satisfy where α is a positive number between 0 and 1, used to limit the orthogonality between users. If is not empty, and the number of elements in is less than Then let i = i + 1, repeat Step2), until the algorithm ends.

[0080] An embodiment of the present invention also discloses a computer system, including a memory, a processor, and a computer program / instructions stored on the memory and executable on the processor. When the computer program / instructions are executed by the processor, the steps of the statistical prior-assisted multi-user scheduling method are implemented.

[0081] An embodiment of the present invention also discloses a computer program product, including computer program / instructions. When the computer program / instructions are executed by the processor, the steps of the statistical prior-assisted multi-user scheduling method are implemented.

[0082] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or a controller of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, so that when the program codes are executed by the processor or the controller, the steps of the method of the present invention are implemented. The program codes can be executed entirely on the machine, partially on the machine, partially on the machine as an independent software package and partially on a remote machine, or entirely on a remote machine or server. Where the present invention is not described in detail, it is common knowledge to those skilled in the art.

Claims

1. A multi-user scheduling method assisted by statistical prior, characterized in that Including the following steps: Divide the base station coverage area into multiple sub - regions, collect the sampled channel information obtained from long - term observations within the sub - regions, and construct a channel knowledge map, including the statistical angular power spectrum, statistical correlation, and statistical channel gain; Calculate the variance of the statistical prior information within each sub - region, and evaluate the reliability of the sub - region statistical prior information according to the variance threshold; Group the sub - regions based on the statistical channel prior information, meet the orthogonality requirements between the sub - regions within the same group, and initially screen the active user set using the grouping; For the active users in the unreliable regions, measure their instantaneous channel information; for the active users in the reliable regions, use the statistical prior of the sub - region as their statistical prior, and update the user angular power spectrum, user - to - user correlation, and user channel gain; Taking the minimization of channel correlation as the criterion, jointly use the angular power spectrum of the instantaneous channel information and the statistical prior information to select the final scheduled user set from the active user set.

2. The multi-user scheduling method assisted by statistical prior according to claim 1, wherein: Statistical angular power spectrum of sub-region l is the mean of the angular power spectra of the observed sampled channels; the statistical correlation between sub-regions l1 and l2 is the degree of similarity of the statistical angular power spectra between sub-regions; the statistical channel gain of sub-region l is the mean of the channel gains of the observed sampled channels.

3. A multi-user scheduling method assisted by statistical prior according to claim 1, characterized in that: Reliability σ of the sub-region l statistical prior l Obtained by calculating the variance of the correlation values of all sampled channels with the region center channel, the correlation of the angular power spectrum between the sth sampled channel in sub-region l and the channel of the region center where ω l,c and ω l,s represent the angular power spectrum of the region center channel and the angular power spectrum of the sth sampled channel respectively.

4. A multi-user scheduling method assisted by statistical prior according to claim 1, characterized in that: Group L sub-regions into N groups according to the statistical channel prior information. The steps include: G ​ Step1) Initialize i = 1, and let the grouped area the ungrouped area Step2) Initialize the i-th group Select the statistical channel gain The largest area And put it into the group Temporary variable for grouped areas Step3) Search for the regions that meet the conditions where represents statistical correlation, α is used to limit the orthogonality between users, and it is put into the grouping Repeat this step until there are no sub-regions that meet the conditions; Step4) Search for the area that meets the conditions and put it into a group Repeat this step until there is no sub - area that meets the conditions; Step5) Let i = i + 1, and repeat Step2) until i > N G or 5. A multi-user scheduling method assisted by statistical prior according to claim 1, characterized in that: The method of initially screening the active user set by grouping sub-regions is as follows: map the user location to the index of the region where the user is located Statistically calculate the user coverage of each group, select the group with the highest coverage, and output the users covered by this group as the active user set 6. A multi-user scheduling method assisted by statistical prior according to claim 1, characterized in that: Based on the active user set, use the location coordinates of the users to map the users to the indices of their respective sub - regions; when the variance of the sub - region where the user is located is less than the set threshold, the user is considered to be in the reliable region; when the variance of the sub - region where the user is located is greater than or equal to the set threshold, the user is considered to be in the unreliable region.

7. A multi-user scheduling method assisted by statistical prior according to claim 1, characterized in that: Update the user angular power spectrum, user - to - user correlation, and user channel gain according to whether the statistical prior of the region where user k is located is reliable or not; The angular power spectrum of user k is expressed as The channel gain of user k is expressed as The correlation between user k and user j is expressed as Among them is the reliable region, is the unreliable region, are the indices of the sub-regions where user k and user j are located respectively, represent the statistical angular power spectrum, statistical correlation and statistical channel gain corresponding to the region index respectively, h k and represent the channel state information in the spatial domain and angular domain obtained through uplink channel estimation respectively, and ⊙ is the Hadamard product.

8. A multi-user scheduling method assisted by statistical prior according to claim 1, characterized in that: The base station selects users from the active user set and schedules these users. The steps include: Step1) Initialize the set Dispatch the user set Step2) For Select users with low correlation and high gain with the selected users And where g k and g s represent the channel gains of user k and user s respectively, and ρ k,s represents the correlation between user k and user s; Step3) If the number of selected users is insufficient Then update the user set So that the orthogonality among users satisfies Where α is used to limit the orthogonality among users; If Is not empty, and The number of elements in is insufficient Then let i = i + 1, repeat Step2) until the algorithm ends.

9. A computer system, comprising a memory, a processor, and a computer program / instructions stored on the memory and executable on the processor, characterized in that, When the computer program / instructions are executed by the processor, the steps of a statistical prior - assisted multi - user scheduling method according to any one of claims 1 - 8 are implemented.

10. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, the steps of a statistical prior - assisted multi - user scheduling method according to any one of claims 1 - 8 are implemented.

Images