Method and device for wireless network interference coordination and resource scheduling based on hierarchical clustering algorithm

Abstract

The application provides a wireless network interference coordination and resource scheduling method based on a hierarchical clustering algorithm, and relates to the technical field of wireless communication, which comprises the following steps: obtaining a set of user equipment with unmet demands; according to the hierarchical clustering algorithm, classifying the equipment with mutual interference less than a threshold value into the same class and the equipment with mutual interference greater than the threshold value into different classes to obtain an optimal non-overlapping clustering result; adjusting and optimizing the optimal non-overlapping clustering result to cause overlapping between different classes and obtain an optimal clustering result; and according to the optimal clustering result, allocating a wireless resource unit to each class, stopping the allocation when the demands of all the equipment are met or the resources are exhausted, and obtaining a final resource allocation result. The application adopting the above scheme can effectively reduce the consumption of wireless resources while meeting the QoS of all the equipment during wireless resource scheduling.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication, and particularly relates to a wireless network interference coordination and resource scheduling method and device based on a hierarchical clustering algorithm. BACKGROUND

[0002] With the rapid growth of network capacity demand, wireless networks nowadays tend to increase the deployment density of network devices to meet the demand for network capacity. In today's 5G wireless networks, whether in the scenarios of enhanced mobile broadband (eMBB) or ultra-reliable low-latency communications (URLLC), more intensive base station deployment can significantly reduce the transmission distance between the base station and the mobile terminal, thereby reducing the transmission loss and improving the network capacity and reliability. However, the intensive deployment of small base stations will cause serious co-channel interference (CCI), bringing new challenges to wireless communication networks.

[0003] In existing cellular networks, the wireless resource allocation function is completed by the base station, and each cell basically manages and allocates wireless resources independently. As one of the core functions of the medium access control (MAC) layer, the scheduler in the base station performs wireless resource scheduling based on the feedback of the retransmission state, buffer state, channel quality state and other information of each device in the cell to the hybrid automatic repeat request (HARQ) process, as well as the buffer state at the base station side, the historical throughput of each device and the quality of service (QoS) requirements of the service, etc. as the basis, in accordance with the results given by the wireless resource scheduling algorithm. It can be seen that the wireless resource scheduling algorithm is the core of wireless resource scheduling.

[0004] The common resource scheduling algorithms in the existing network mainly include the following three types:

[0005] Maximum throughput type: This type of algorithm mainly focuses on the maximum throughput that the wireless network can achieve. The base station using this type of algorithm tends to allocate resources to a number of devices with good channel quality as much as possible, and almost no resources to other devices. Therefore, although this type of algorithm can achieve high network throughput, the network fairness is very poor;

[0006] Polling: This algorithm mainly focuses on the fairness of the network. The base station using this algorithm will allocate resources to each device in turn, and each device is completely equal. Using this algorithm can guarantee better fairness, but the network performance is low, and when the number of users in the cell is large, the delay caused by polling is often serious, which cannot meet the QoS requirements of some services;

[0007] Proportional fairness: This algorithm takes into account both fairness and performance, and is the most commonly used algorithm. The core idea of this algorithm is to calculate a priority index according to the historical throughput and channel quality of each device in each scheduling period, and then schedule each device according to the index. Generally speaking, the smaller the historical throughput of the device and the better the channel quality, the higher the priority. In addition, in addition to historical throughput and channel quality, the priority index of the proportional fairness algorithm can also be calculated in combination with other information to match the needs of the wireless network.

[0008] The intra-cell resource allocation algorithm used in the current network does not take into account the inter-cell CCI, so in a dense networking environment, the performance will be severely affected. Therefore, the current network often uses additional inter-cell interference coordination mechanisms such as ICIC, eICIC or CoMP to compensate for this deficiency.

[0009] However, ICIC and eICIC technologies are extremely dependent on signaling exchange, resulting in poor timeliness of interference information and severe limitations on data volume. Moreover, in a densely deployed network, the number of neighboring cells is extremely large, and signaling exchange will result in extremely serious overhead; and CoMP requires a large amount of computing resources for signal processing, and it needs to perform frequent channel measurements, which will occupy a large number of pilot resources and affect the normal operation of the network. Therefore, the current wireless network cannot well handle the problem of inter-cell interference.

[0010] In the academic community, most existing research on multi-cell resource allocation algorithms is based on the assumption that the network channel information is completely and timely known, but this assumption does not hold in real wireless networks. Moreover, based on this assumption, the academic community's research on resource allocation algorithms also usually ignores the impact of small-scale fading, so in scenarios with high QoS requirements such as URLLC, existing algorithms often cannot well guarantee QoS. SUMMARY

[0011] The present application aims to at least partially solve one of the technical problems in the related art.

[0012] To this end, the first purpose of the present application is to propose a wireless network interference coordination and resource scheduling method based on a hierarchical clustering algorithm, which solves the technical problem that existing methods cannot well handle inter-cell interference, and achieves effective reduction of wireless resource consumption while meeting the QoS of all devices during wireless resource scheduling.

[0013] A second object of the present application is to provide a wireless network interference coordination and resource scheduling device based on a hierarchical clustering algorithm.

[0014] To achieve the above object, the first aspect of the present application provides a wireless network interference coordination and resource scheduling method based on a hierarchical clustering algorithm, comprising: obtaining a set of user equipment whose demands are not satisfied; according to the hierarchical clustering algorithm, grouping devices in the set of user equipment into the same class if the interference between the devices is less than a threshold value, and grouping devices into different classes if the interference between the devices is greater than the threshold value, to obtain an optimal non-overlapping clustering result; adjusting and optimizing the optimal non-overlapping clustering result to cause overlapping between different classes, to obtain an optimal clustering result; and allocating a wireless resource unit to each class according to the optimal clustering result, and stopping the allocation if the demands of all devices are satisfied or the resources are exhausted, to obtain a final resource allocation result.

[0015] The wireless network interference coordination and resource scheduling method based on a hierarchical clustering algorithm of the present application obtains the interference relationship between network devices by using existing interference modeling technology without needing to master accurate and complete channel measurement information, and avoids and coordinates the interference between network devices and schedules resources by means of a hierarchical clustering algorithm, so as to reduce the consumption of wireless resources as much as possible while satisfying the QoS of all devices.

[0016] Optionally, in an embodiment of the present application, the hierarchical clustering algorithm is a condensation-type hierarchical clustering algorithm, and the classification of the set of user equipment according to the hierarchical clustering algorithm comprises:

[0017] The set of user equipment is clustered according to the difference between classes by the condensation-type hierarchical clustering algorithm, a binary agglomerative tree is obtained, and a set of clustering results is obtained by setting different heights for the binary agglomerative tree, wherein the difference between classes is determined by a linkage function, and each device in each class included in each clustering result uses the same resource, and different classes use different resources;

[0018] The set of clustering results is screened according to the expected resource occupation amount and the achievable total rate of each clustering result, to obtain an optimal non-overlapping clustering result, wherein the expected resource occupation amount is the sum of the maximum resource amount expected to be required by all classes in the clustering result, and the achievable total rate is the sum of the rates achievable by all classes in the clustering result, and the sum of the rates achievable by each class is the sum of the rates achievable by all devices included in the class.

[0019] Optionally, in an embodiment of the present application, the determination process of the linkage function is:

[0020] A distance function for determining the difference between user equipment is defined, and the linkage function is determined according to the distance function,

[0021] wherein the distance function is represented as:

[0022]

[0023] The link function is represented as:

[0024]

[0025] wherein, denotes the device U p as a signal device, the device U q the reciprocal of the signal-to-interference ratio of the interference, denotes the device U q as a signal device, the device U p the reciprocal of the signal-to-interference ratio of the interference, j p denotes the device U p the associated base station serial number, j q denotes the device U q the associated base station serial number, j q denotes the device U q the associated base station serial number, denotes the mth class in the clustering result, denotes the nth class in the clustering result, and d(p, q) denotes the difference between the device p and the device q.

[0026] Optionally, in an embodiment of the present application, the optimal non-overlapping clustering result is adjusted and optimized, comprising:

[0027] obtaining all classes with a height less than a preset height in the binary convergence tree to obtain a to-be-merged class set, wherein the preset height is a height corresponding to the optimal non-overlapping clustering result, and each element in the to-be-merged set is assigned a subscript in descending order of the corresponding height;

[0028] judging whether a current to-be-merged class in the to-be-merged class set satisfies a preset condition, if yes, merging the current to-be-merged class with the optimal non-overlapping clustering result, and updating the optimal non-overlapping clustering result using the merged optimal non-overlapping clustering result;

[0029] continuously updating the optimal non-overlapping clustering result until all classes in the to-be-merged class set are judged, and obtaining a final optimal non-overlapping clustering result as an optimal clustering result;

[0030] The preset condition is that: the current updated optimal non-overlapping clustering result and the current to-be-merged class have an empty intersection, all devices in the merged cluster use the same resource, and the resource utilization after merging is greater than before merging, wherein the resource utilization after merging is greater than before merging, including: the expected resource occupation amount after merging is greater than before merging or the expected resource occupation amount after merging is equal to before merging and the total reachable rate after merging is greater than before merging.

[0031] Optionally, in an embodiment of the present application, the wireless resource units are allocated to each class according to the optimal clustering result, and the allocation is stopped when the demands of all devices are met or the resource is exhausted, including:

[0032] A wireless resource unit is allocated to each class in the optimal clustering result, and the reachable rate of each user device after allocation is calculated, wherein if the user device exists in multiple classes, the reachable rate of the user device is the sum of the reachable rates of the user device in each class.

[0033] The rate demand of each user device is updated according to the reachable rate of each user device, and the current completed resource allocation is added to the resource allocation result list.

[0034] If the rate demands of all user devices after updating are all reduced to 0 or the resource is exhausted, resource scheduling is performed according to the current resource allocation result list, otherwise the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm is re-executed.

[0035] Optionally, in an embodiment of the present application, if the user device set has services with QoS demand greater than a threshold value, a small-scale fading compensation mechanism is applied to the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm, including:

[0036] When the user device set is classified using the hierarchical clustering algorithm and the optimal non-overlapping clustering result is adjusted and optimized, the sum of the reachable rates of all user devices after applying the fading protection boundary is taken as the total reachable rate, and the optimal transmission number is determined, and the product of the optimal transmission number and the expected maximum resource amount is taken as the expected resource occupation amount.

[0037] When the wireless resource units are allocated to each class according to the optimal clustering result, the wireless resource units allocated to each class are changed from 1 to the optimal transmission number, and the reachable rate of the user device after applying the fading protection boundary is taken as the reachable rate of the user device.

[0038] To achieve the above purpose, a second embodiment of the present application proposes a wireless network interference coordination and resource scheduling device based on a hierarchical clustering algorithm, including an acquisition module, a clustering module, a clustering optimization module, and a resource allocation module, wherein,

[0039] an acquisition module, configured to acquire a set of user equipment whose demands are not satisfied;

[0040] a clustering module, configured to cluster, according to a hierarchical clustering algorithm, devices in the set of user equipment into the same class if the devices interfere with each other less than a threshold value, and into different classes if the devices interfere with each other more than the threshold value, to obtain an optimal non-overlapping clustering result;

[0041] a clustering optimization module, configured to adjust and optimize the optimal non-overlapping clustering result to cause overlapping between different classes, to obtain an optimal clustering result;

[0042] a resource allocation module, configured to allocate a unit of wireless resource to each class according to the optimal clustering result, and stop the allocation if demands of all the devices are satisfied or resources are exhausted, to obtain a final resource allocation result.

[0043] Optionally, in an embodiment of the present application, the hierarchical clustering algorithm is a hierarchical agglomerative clustering algorithm, and the clustering module is specifically configured to:

[0044] cluster the set of user equipment according to differences between classes by using the hierarchical agglomerative clustering algorithm, to obtain a binary agglomerative tree, and set different heights for the binary agglomerative tree to obtain a set of clustering results, wherein the differences between the classes are determined by a linkage function, and each device in each class included in each clustering result uses the same resource, and different classes use different resources;

[0045] screen the set of clustering results according to expected resource occupation amounts and achievable total rates of each clustering result, to obtain the optimal non-overlapping clustering result, wherein the expected resource occupation amount is a sum of maximum resource amounts expected to be required by all the classes in the clustering result, and the achievable total rate is a sum of rates achievable by all the classes in the clustering result, and the sum of the rates achievable by each class is a sum of rates achievable by all the devices included in the class.

[0046] Optionally, in an embodiment of the present application, the clustering optimization module is specifically configured to:

[0047] obtain classes in the binary agglomerative tree whose heights are less than a preset height, to obtain a set of classes to be merged, wherein the preset height is a height corresponding to the optimal non-overlapping clustering result, and each element in the set of classes to be merged is assigned a subscript in descending order of the height corresponding to the element;

[0048] determine whether a current class to be merged in the set of classes to be merged meets a preset condition, and if so, merge the current class to be merged with the optimal non-overlapping clustering result, and update the optimal non-overlapping clustering result using the merged optimal non-overlapping clustering result;

[0049] continuously update the optimal non-overlapping clustering result until all the classes in the set of classes to be merged are determined, to obtain a final optimal non-overlapping clustering result as the optimal clustering result.

[0050] wherein, the preset condition is that: the current updated optimal non-overlapping clustering result and the current to-be-merged class intersection is empty, and all devices in the merged clustering use the same resource, and the resource utilization after merging is greater than before merging, wherein, the resource utilization after merging is greater than before merging, including: the expected resource occupation after merging is greater than before merging or the expected resource occupation after merging is equal to before merging while the total rate reachable after merging is greater than before merging.

[0051] Optionally, in an embodiment of the present application, the resource allocation module is specifically configured to:

[0052] allocate 1 unit of wireless resource to each class in the optimal clustering result, and calculate the reachable rate of each user equipment after allocation, wherein, if the user equipment exists in multiple classes, the reachable rate thereof is the sum of the reachable rates of the user equipment in each class;

[0053] update the rate requirement of each user equipment according to the reachable rate of each user equipment, and add the current completed resource allocation into the resource allocation result list;

[0054] if the rate requirement of all user equipments after updating is reduced to 0, or the resource is exhausted, perform resource scheduling according to the current resource allocation result list, otherwise, re-execute the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm.

[0055] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0056] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, including the appended drawings, wherein:

[0057] Figure 1 BMT schematic diagram obtained by the hierarchical clustering of the embodiments of the present application;

[0058] Figure 2 3GPP dual-strip model schematic diagram of the embodiments of the present application;

[0059] Figure 3 Flowchart of a wireless network interference coordination and resource scheduling method based on a hierarchical clustering algorithm provided in Embodiment One of the present application;

[0060] Figure 4 Performance comparison chart of the proposed scheme and the comparative scheme in the small-scale network of the embodiments of the present application;

[0061] Figure 5Performance comparison chart of the proposed scheme and the comparative scheme in a large-scale network of the embodiments of the present application;

[0062] Figure 6 Decoding error probability chart of the proposed scheme and the comparative scheme of the embodiments of the present application under different average transmission times;

[0063] Figure 7 Decoding error probability chart of the proposed scheme and the comparative scheme of the embodiments of the present application under different average transmission times;

[0064] Figure 8 Decoding error probability chart of the proposed scheme and the comparative scheme of the embodiments of the present application under different RU occupation amounts;

[0065] Figure 9 Structure schematic diagram of a wireless network interference coordination and resource scheduling device based on a hierarchical clustering algorithm provided by the embodiments of the present application. DETAILED DESCRIPTION

[0066] The embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0067] The following briefly describes the technologies used in the scheme of the present application:

[0068] Clustering algorithm: Clustering algorithm is a kind of unsupervised machine learning, which divides a data set into different classes or clusters according to a certain standard (such as distance), so that the similarity of data objects in the same cluster is as large as possible, and the difference of data objects not in the same cluster is as large as possible. That is, the data in the same class is as close as possible after clustering, and the data in different classes is as far as possible. The clustering method can be mainly divided into partition clustering, density-based clustering, and hierarchical clustering.

[0069] In the present application, mainly agglomerative hierarchical clustering (AHC) is used. There is a data point set S to be clustered, S = {s i| i = 1, 2,..., N}, define a "leaf" as a class containing only 1 data point, and a "root" as a class containing all data points. In AHC, each data point is initially considered as a leaf. In each step of the algorithm, the two most similar classes are combined into a new larger class, until all data points are contained in one class, i.e. a Binary Merge Tree (BMT) is formed, as shown in Figure 1 for an example of BMT when N = 6.

[0070] AHC relies on a non-negative linkage function to judge the dissimilarity between classes.

[0071] Step 1 initializes the remaining class set S to be all classes, i.e. all classes are leaves.

[0072] Let the remaining class set in the kth step be S The linkage function between class S and class S is denoted as d(S ). In the process of clustering, in each step, the two classes with the highest similarity (i.e. the smallest dissimilarity) will be found:

[0073]

[0074] and the two classes are combined into a new class S with height h . The remaining class set is updated as:

[0075]

[0076] This iteration continues until there is only one class (i.e. the root) left in the remaining class set, and the algorithm stops. Denote the set of all new classes (called nodes) generated by merging other classes as N = {N1,..., N N-1}, and the corresponding height set as h = {h1,..., h N-1}. Obviously, the height set is in ascending order (i.e. h1≤ h2≤...≤ h N-1 ). In addition, the height of a leaf is defined as 0.

[0077] From the above introduction, the linkage function directly determines the final result of AHC. Therefore, the linkage function should be reasonably defined according to the problem to be solved.

[0078] Wireless network interference and system model:

[0079] Figure 2 is a schematic diagram of the 3GPP dual-strip model, as Figure 2 ​As shown, in a small-scale wireless service hotspot area, such as an office, theater, or smart factory with a large number of sensors and industrial equipment, operators will deploy a large number of small base stations to address throughput or service QoS requirements. There are two rows of rooms on each side of the corridor, each row consisting of N... r The system consists of several rooms. Each room contains a small base station and q devices, therefore there are a total of R = 4N rooms. r A small base station and M=4N r There are q devices. In addition, the system has K allocable radio resource units (RUs). Let the set of small base stations be C = {C1, ..., C2}. R The set of devices is U = {U1, ..., U}. M}, any device U n The minimum rate requirement is R n, .

[0080] The proposed solution in this application is based on a wireless interference model for interference coordination and resource scheduling. Therefore, it is necessary to describe the interference model required by this application.

[0081] In the uplink direction of the mobile communication network, the serving device U m The uplink signal transmitted by the k-th RU is used to reach the home cell C. jm The base station's signal reception power is:

[0082]

[0083] in For the k-th RU device U m To the assigned community C jm The channel gain of a base station includes large-scale fading that is largely independent of the RU and small-scale fading that is related to the RU.

[0084] Similarly, the jamming device U transmits signals through the k-th RU. n Arrive at Community C jm The interference signal power of the base station is:

[0085]

[0086] in For the k-th RU device U n To Community C jm The channel gain of a base station also includes large-scale fading that is largely independent of the RU and small-scale fading that is related to the RU.

[0087] Therefore, on the k-th RU, the serving device U m The signal in the home cell C jmThe uplink signal-to-interference-plus-noise ratio (UL-SINR) at the base station is:

[0088]

[0089] in, To be compatible with service equipment U m The set of interference devices for the k-th RU is reused, σ 2 This represents noise power.

[0090] The above formula can be further transformed to obtain:

[0091]

[0092] in, For service equipment U m The signal-to-noise ratio (SNR) of the received signal at the k-th RU. For service equipment U m Interference device U that reuses the k-th RU n The signal-to-interference ratio (SIR) of the received signals.

[0093] In real-world wireless networks, small-scale fading conditions are neither available in real time nor predictable. Therefore, the proposed solution in this application uses only the statistical channel gain, which includes large-scale fading, to estimate network performance. (Corresponding to device U) m Statistical UL-SINR, SNR and U m with U n The SIRs between them are denoted as γ. m , and γ m, .

[0094] make The above equation simplifies to:

[0095]

[0096] Records and service equipment U m Let N be the set of interfering devices occupying the same wireless resource. m As can be seen from the above formula, the reciprocal of the statistical UL-SINR It can be provided by service equipment U m With each interfering device that occupies the same wireless resources U n Statistical SIR reciprocal between Service equipment statistical SNR reciprocal Linear superposition. Further, dense deployment of wireless networks is usually interference-limited, and the impact of noise can be negligible compared to interference. Therefore, the scheme proposed in this application will mainly rely on statistical SIR to coordinate and optimize interference. The interference model used in this application should also include statistical SIR information of all possible interfering device pairs, as well as statistical SNR information of each device.

[0097] It should be particularly noted that the composition of the interference device set may be different in different orthogonal resource sharing systems. For example, in Orthogonal Frequency Division Multiple Access (OFDMA) or Time Division Multiple Access (TDMA) systems, devices in the same cell do not interfere with each other, so the interference device set only includes devices that are not in the same cell as the serving device; but in Code Division Multiple Access (CDMA) and Non-Orthogonal Multiple Access (NOMA) systems, devices in the same cell can also interfere with each other, so the interference device set includes all devices in the system except the serving device. However, since the above derivation of the interference relationship is not specific to a particular orthogonal resource multiplexing method, by constructing different interference device sets, the interference coordination and resource allocation scheme proposed in this application can also be applied to various wireless systems using different orthogonal resource multiplexing methods. Similarly, the above interference relationship derivation also does not depend on the networking scenario shown in Figure 2 Therefore, the interference coordination and resource allocation scheme proposed in this application is also applicable to any wireless communication networking model.

[0098] To evaluate network performance, in scenarios where QoS requirements are not high and best-effort services are mainly used, the Shannon formula is usually used to calculate the instantaneous rate of any device U m The instantaneous rate that can be achieved in the kth RU is:

[0099]

[0100] where W (k) is the bandwidth of the kth RU.

[0101] In actual wireless networks, the bandwidth of each RU is usually the same, denoted as W. Therefore, excluding small-scale fading that cannot be predicted, this application uses statistical SINR to estimate system performance:

[0102] R m = min{W log2(1+m ), m,req}

[0103] Here, the QoS requirement rate is taken as the upper limit of the estimated rate to avoid allocating more resources than the device needs in the resource allocation process.

[0104] For scenarios with high QoS requirements, such as URLLC, reliability needs to be considered in particular. In this case, any device U m The instantaneous rate that can be achieved on the kth RU is:

[0105]

[0106] where, is the channel dispersion, τ is the duration of signal transmission, Q -1 is the inverse function of the Gaussian Q function, and ∈ is the decoding error probability. In this case, using statistical SINR to estimate system performance is not appropriate, because it will lead to real-time fluctuations in the decoding error probability, which cannot meet the QoS requirements. To compensate for the inability to estimate performance and difficulty in scheduling resources caused by small-scale fading, the scheme proposed in this application makes additional restrictions on the achievable rate to avoid performance fluctuations based on statistical SINR. The specific method will be described in detail in the "Small-scale fading compensation" section.

[0107] The wireless network interference coordination and resource scheduling method based on hierarchical clustering algorithm of the present application includes four parts: interference coordination, resource optimization, resource scheduling, and small-scale fading compensation.

[0108] Based on the interference model provided by the above wireless network user uplink interference modeling scheme, the "interference coordination" part of the scheme described in this application uses the hierarchical clustering algorithm to cluster devices according to the interference intensity relationship between devices provided by the interference model, so that devices with weak mutual interference can be divided into the same class and reuse the same resources, and devices with strong mutual interference can be divided into different classes and use orthogonal resources, to achieve the purpose of reducing network interference and reducing wireless resource occupation.

[0109] Since the hierarchical clustering algorithm is a non-overlapping clustering algorithm, the classification result obtained by the "interference coordination" part of the scheme described in this application is that each device belongs to only one class, i.e., only reuses the same set of wireless resources with one group of devices. However, in fact, each device has the ability to use multiple sets of wireless resources. Therefore, the "resource optimization" part of the scheme described in this application adjusts the classification result obtained by the "interference coordination" part, so that a device can use multiple different resources, to further improve the resource utilization efficiency of the system.

[0110] After the interference coordination and resource optimization are completed, the "resource scheduling" part of the scheme adjusts the classified results according to the adjusted classified results of the "resource optimization" part, and performs actual resource scheduling, including: allocating resources to each class in the optimal clustering result after "resource optimization" between devices, and integrating the allocation results into a resource allocation result list. If the QoS of all devices is satisfied or all resources are exhausted after allocation, the scheme ends running and obtains the final resource allocation result; otherwise, the above process is repeated for the devices whose QoS requirements are not met until the scheme ends running.

[0111] In addition, in the face of a scenario with high QoS requirements (such as URLLC), the scheme combines a small-scale fading compensation method to better guarantee the QoS requirements of each device.

[0112] As can be seen from the above introduction, the operation of the scheme does not depend on a specific networking scenario and multiple access mode, and can be adapted to a variety of common scenarios and business models in today's 5G network. Therefore, the scheme has a very broad application prospect. The applicable scenarios include but are not limited to: resource scheduling of public commercial 5G eMBB wireless networks, and grant-free resource scheduling of industrial 5G URLLC wireless networks.

[0113] The wireless network interference coordination and resource scheduling method and device based on the hierarchical clustering algorithm of the embodiments of the present application are described below with reference to the accompanying drawings.

[0114] Figure 3 A flowchart of a wireless network interference coordination and resource scheduling method based on a hierarchical clustering algorithm provided by Embodiment One of the present application.

[0115] As Figure 3 shown, the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm includes the following steps:

[0116] Step 301: Obtain a set of user devices whose requirements are not met;

[0117] Step 302: According to the hierarchical clustering algorithm, devices in the user device set that interfere with each other less than a threshold are classified into the same class, and devices that interfere with each other more than the threshold are classified into different classes, to obtain an optimal non-overlapping clustering result;

[0118] Step 303: Adjust and optimize the optimal non-overlapping clustering result to cause overlap between different classes, to obtain an optimal clustering result;

[0119] Step 304: According to the optimal clustering result, allocate a wireless resource unit to each class, and stop allocation when the requirements of all devices are met or resources are exhausted, to obtain a final resource allocation result.

[0120] The wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm of the embodiments of the present application, without the need to master accurate and complete channel measurement information, utilizes existing interference modeling technology to obtain the interference relationship between network devices, and with the help of hierarchical clustering algorithm, avoids and coordinates the interference and schedules resources according to the interference relationship, so as to reduce the consumption of wireless resources as much as possible while meeting the QoS of all devices.

[0121] Optionally, in an embodiment of the present application, the interference coordination part takes the hierarchical clustering algorithm as the core, classifies the user devices with less mutual interference into the same class, and allows the reuse of the same resource to improve the resource utilization rate; classifies the user devices with greater mutual interference into different classes, so that they occupy orthogonal resources to avoid strong interference from seriously affecting the system performance. Specifically, the hierarchical clustering algorithm used in the present application is the agglomerative hierarchical clustering algorithm AHC, and since AHC ultimately obtains the BMT as shown in Figure 1 Fig. 1, and according to the given height difference, the BMT can generate a plurality of different clustering results, the above classification of the user device set according to the hierarchical clustering algorithm includes:

[0122] The user device set is clustered according to the difference between the classes by the agglomerative hierarchical clustering algorithm to obtain a binary merge tree, and different heights are set for the binary merge tree to obtain a clustering result set, wherein the difference between the classes is determined by a linkage function, and each class contained in each clustering result uses the same resource, and different classes use different resources;

[0123] The clustering result set is screened according to the expected resource occupation amount and the achievable total rate of each clustering result to obtain an optimal non-overlapping clustering result, wherein the expected resource occupation amount is the sum of the maximum resource amounts expected to be required by all classes in the clustering result, the achievable total rate is the sum of the rates that can be achieved by all classes in the clustering result, and the sum of the rates that can be achieved by each class is the sum of the rates that can be achieved by all devices contained in the class;

[0124] Taking the dense deployment, OFDMA wireless system based on the 3GPP dual-strip model as shown in Figure 2 Fig. 2 as a typical system scenario, the method for determining and generating the optimal non-overlapping clustering result proposed by the present application is summarized as follows:

[0125] Step 1: initialize the clustering result as , indicate variable i = 1, expected minimum resource occupation amount E min = +∞, expected maximum total rate R max = 0, and the optimal clustering result is The optimal clustering result corresponds to the height

[0126] Step 2: let Each device contained in each class uses the same resource, and different classes use different resources.

[0127] Step 3: Calculate the achievable rate of each device in each class according to the rate estimation method in the wireless network interference and system model technology introduced above. For convenience of description, let Then, for any class The estimated achievable rate of each device in the class is {R m1 ,…,R mK}, and the required resource amount of the class is The achievable rate of the class is Therefore, the clustering manner The expected resource occupation amount that can be obtained is The total achievable rate is

[0128] Step 4: If E < E min , or E = E min and R > R max , update E min = E, R max = R,

[0129] Step 5: i = i + 1. If i < N and h i <+∞, return to Step 2; otherwise, the process ends, that is, the optimal clustering result is obtained, and the height corresponding to the result is

[0130] Optionally, in an embodiment of the present application, in order to enable the AHC to better solve the interference coordination problem, the present scheme needs to define the linkage function according to the characteristics of the interference coordination problem. Further, since the linkage function describes the difference between classes, and the difference between classes is determined by the difference between data points (each data point in the present application corresponds to a device) located in different classes, the present scheme first defines a function that describes the difference between data points, i.e., a distance function:

[0131]

[0132] That is, if the interference between two devices U p and U q is greater, the distance between them is also greater. If two devices cannot reuse resources due to the multiple access manner used by the system and other limitations, the distance between them is set to infinity to ensure that they will not be divided into the same class. ​

[0133] Based on the definition of the distance function, the link function can naturally be defined as:

[0134]

[0135] This linking function is also known as complete linkage. When using the complete linkage function, the BMT can be constructed quickly using the CLINK algorithm.

[0136] Optionally, in one embodiment of this application, the above-mentioned optimal non-overlapping clustering results are non-overlapping, that is, the intersection of any two classes in the clustering results is empty. However, in actual wireless networks, each device can occupy multiple resources simultaneously. Therefore, it is necessary to adjust and optimize the above-mentioned optimal non-overlapping clustering results to allow devices to occupy multiple resources simultaneously, so as to further improve the resource utilization of the system. Specifically, adjusting and optimizing the optimal non-overlapping clustering results includes:

[0137] Obtain all classes in the binary clustering tree whose height is less than the preset height to obtain the set of classes to be merged. The preset height is the height corresponding to the optimal non-overlapping clustering result. Each element in the set to be merged is assigned an index in descending order of its corresponding height.

[0138] Determine whether the current class to be merged in the set of classes to be merged meets the preset conditions. If it does, merge the current class to be merged with the optimal non-overlapping clustering result, and update the optimal non-overlapping clustering result using the merged optimal non-overlapping clustering result.

[0139] The optimal non-overlapping clustering result is continuously updated until all classes in the set of classes to be merged have been evaluated, and the final optimal non-overlapping clustering result is obtained as the optimal clustering result.

[0140] The preset conditions are: the intersection of the currently updated best non-overlapping clustering result and the current cluster to be merged is empty, all devices in the merged cluster use the same resource, and the resource utilization rate after merging is greater than that before merging. The resource utilization rate after merging is greater than that before merging includes: the expected resource consumption after merging is greater than that before merging, or the expected resource consumption after merging is equal to that before merging and the total achievable rate after merging is greater than that before merging.

[0141] by Figure 2 Taking the densely deployed OFDMA wireless system based on the 3GPP dual-strip model shown as a typical system scenario, let the optimal non-overlapping clustering result be... The proposed method for determining the optimal clustering result is summarized as follows:

[0142] Step 1: Organize the classes that can be used. With the BMT constructed in the process of generating the optimal non-overlapping clustering result, the process of resource optimization can be simplified. All classes in the BMT whose heights are less than (including the nodes whose heights are less than and all leaves) can be the objects that are merged into any class in in the adjustment process. Denote the set of these classes as S = {S1,..., S |S|}. The elements in the set are indexed in descending order of their corresponding heights, i.e., the higher the height, the smaller the index value.

[0143] Step 2: Initialize i = 1 and j = 1.

[0144] Step 3: Calculate the expected resource usage and the achievable rate of according to the method used in the process of generating the optimal non-overlapping clustering result, and initialize the optimal expected resource usage the optimal achievable rate and the optimal class

[0145] Step 4: Attempt to merge with S j in S to obtain If the following conditions are met simultaneously, S j is the optimal choice for the merging of :

[0146] 1)

[0147] 2) All devices in can use the same resource;

[0148] 3) The merging can improve the resource utilization. That is, the expected resource usage and the achievable rate of calculated according to the method used in the process of generating the optimal non-overlapping clustering result satisfy or both and

[0149] Step 5: If S j is the current optimal merging option for , update

[0150] Step 6: j = j + 1. If j ≤ |S|, return to Step 4. Otherwise, replace in with

[0151] Step 7: i = i + 1. If Go back to Step 3. Otherwise, is the optimal clustering result.

[0152] Optionally, in an embodiment of the present application, the wireless resource units are allocated to each class according to the optimal clustering result, and the allocation stops when all the demands of the devices are satisfied or the resource is exhausted, including:

[0153] allocating 1 RU to each class in the optimal clustering result, and calculating the achievable rates of each user device after the allocation, wherein if a user device exists in multiple classes, the achievable rate of the user device is the sum of the achievable rates of the user device in the classes;

[0154] updating the rate demands of each user device according to the achievable rates of the user devices, and adding the current completed resource allocation to the resource allocation result list;

[0155] if the rate demands of all the user devices after the update are all reduced to 0 or the resource is exhausted, performing resource scheduling according to the current resource allocation result list, otherwise re-executing the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm;

[0156] as shown in the dense deployment, OFDMA wireless system based on the 3GPP dual-strip model as a typical system scenario, the resource scheduling process is summarized as follows: Figure 2 Step 1: allocate 1 RU to each class in the optimal clustering result

[0157] , and estimate the achievable rates of the devices after this round of allocation in the manner introduced in the wireless network interference and system model part above If a device exists in multiple classes, the estimated achievable rate of the device is the sum of the estimated achievable rates of the device in the classes.

[0158] Step 2: update the rate demands of the devices. For a device U i , the rate demand is updated to and add the current completed resource allocation to the resource allocation result list. If after the update, the rate demands of all the devices are all reduced to 0 or the resource is exhausted, perform resource scheduling according to the current resource allocation result list; otherwise, re-perform the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm above.

[0159] ​Optionally, in an embodiment of the present application, it is mainly suitable for eMBB and other services with low QoS requirements. For URLLC and other services with high QoS requirements, only relying on the above process cannot compensate for the performance fluctuations caused by small-scale fading, thereby affecting the QoS satisfaction of such services.

[0160] Therefore, according to the characteristics of small-scale fading, the present application will give the best configuration of the two small-scale fading compensation methods: reserved fading protection boundary and multiple repeated transmission. At the same time, the present application will also give the mechanism of combining the two methods into the wireless network interference coordination and resource scheduling scheme based on the hierarchical clustering algorithm proposed in the present application to further expand the applicability of the wireless network interference coordination and resource scheduling scheme based on the hierarchical clustering algorithm proposed in the present application.

[0161] In the present application, small-scale fading analysis is mainly carried out for Nakagami-m channel fading model. Since it has been proved in many wireless communication system experiments that Nakagami-m channel model has good fitting for measured data, the model has good representativeness. The method of combining multiple repeated transmission is the maximum ratio combining (MRC) commonly used in today's wireless networks.

[0162] Fading protection boundary

[0163] The reserved fading protection boundary actually reduces the minimum SINR required to ensure decoding success by reducing the amount of data transmitted by users on resources, that is, a certain boundary is reserved for small-scale fading. In order to save resources while meeting QoS, it is necessary to reasonably determine the size of the protection boundary. In the present application, the expression of the best protection boundary value will be given.

[0164] In Nakagami-m fading channel, all useful signals and interference signals in the same resource obey Nakagami-m distribution. Let the useful signal where is Nakagami-m distribution, L is the number of signal transmissions, m x ∈[0.5, +∞) is the shape factor of Nakagami-m distribution, which depends on the channel characteristics, and the larger the value is, the more flat the fading is; Ω x is the average power of signal x.

[0165] Similarly, if there are J interference sources, the jth interference signal For analysis convenience, combine the J interference sources into an equivalent interference y, then and

[0166] Thus, the decoding error probability can be expressed as where γ is the instantaneous SINR, The minimum SINR required to ensure reliability is given by the statistical SINR calculated using the interference model divided by the guard margin value G.

[0167] Note that a densely deployed wireless network is usually interference-limited, i.e. the impact of noise is negligible, so that the SINR can be well estimated by the SIR. Thus, we have is the statistical SIR of the current useful signal. According to the above derivation, it is obvious that which can be directly obtained by the interference model. Let the decoding error probability requirement be no more than ε. After derivation, the expression of the guard margin value G can be obtained as

[0168]

[0169] where B(·,·) is the Beta function.

[0170] Similarly, if the current resource is used by only one device, there is no interference, so that the SINR degenerates to the SNR. In this case, the expression of the guard margin value G is

[0171]

[0172] Obviously, the value of G is related to L. Therefore, the guard margin value is denoted as G L . After applying the fading guard margin, the achievable rate of any device U m can be estimated by

[0173]

[0174] The value of L can be arbitrarily specified when the method is used alone, and it can also be determined in combination with the method introduced in the following multiple repeated transmission part.

[0175] Multiple repeated transmission

[0176] Multiple repeated transmission uses diversity and combining to avoid the system reliability reduction caused by the signal falling into deep fading. Too few transmission times can hardly guarantee the system reliability, while too many transmission times will affect the resource utilization efficiency of the system. Therefore, the present application will give a method for determining the optimal transmission times of a given class under the premise of guaranteeing QoS.

[0177] Step 1: initialize the optimal transmission times L * = 1, L = 1;

[0178] Step 2: calculate the achievable rate of any device The estimated average rate per resource occupation when L transmissions are used:

[0179]

[0180] Similarly, the average rate per resource occupation of any device U The estimated average rate per resource occupation when L+1 transmissions are used R m,L+1 ;

[0181] Step 3: The total achievable average rate of this class when L transmissions are used is obtained as follows:

[0182]

[0183] Similarly, the total achievable average rate of this class when L+1 transmissions are used is obtained as follows:

[0184] Step 4: The SINR lower bound when L transmissions are better than L+1 transmissions is calculated as follows:

[0185]

[0186] where,

[0187] Step 5: If the following two conditions are satisfied:

[0188] 1. The SINR lower bound is greater than the average rate per resource occupation of any device U 2. The total achievable average rate of this class when L transmissions are used is greater than the total achievable average rate of this class when L+1 transmissions are used

[0189] then L * = L is the optimal number of transmissions; otherwise, L = L+1, and return to Step 2.

[0190] The two small-scale fading compensation methods involved in this application can be applied separately, but in order to achieve better results, in this application, both will be applied to the aforementioned scheme at the same time, which will affect the estimation of the rate and resource occupation of the aforementioned scheme.

[0191] Specifically, after applying the small-scale fading compensation mechanism, the wireless network interference coordination and resource scheduling scheme based on the hierarchical clustering algorithm needs to be changed as follows:

[0192] Interference coordination part:

[0193] In Step 3 of generating the optimal non-overlapping clustering result, the expression given by the fading protection boundary part is used to estimate the rate R mk of any device U mk ;

[0194] In Step 3 of generating the optimal non-overlapping clustering result, the rate R When dealing with resource consumption, instead of using the method described above for multiple repeated transmissions, determine the optimal number of transmissions L first. * Instead, the following formula is used to estimate the amount of resources required for this type of resource:

[0195]

[0196] Resource optimization section:

[0197] Following the changes to the aforementioned interference coordination section, the methods for estimating achievable rates and resource consumption in steps 3 and 4 of that section are also changed accordingly.

[0198] Resource scheduling section:

[0199] In step 3 of this section, for any of the following categories The number of RUs allocated to it has been changed from 1 to L. * One, of which L * For the method determined using the above-described multiple repeated transmission section, The optimal number of transmissions;

[0200] In step 3 of this section, the achievable rate of the device is estimated using the expression given by the fading protection boundary section.

[0201] The following section describes the performance of the proposed hierarchical clustering algorithm-based wireless network interference coordination and resource scheduling scheme (hereinafter referred to as "HCSA"). In this section, existing schemes such as SGRA, StM, and TSDR are used as comparison schemes to demonstrate the superior performance of the proposed HCSA scheme compared to the closest and most effective existing schemes.

[0202] (1) Simulation parameters and scene settings.

[0203] To fully demonstrate the performance of the HCSA scheme proposed in this application, the simulation uses a URLLC scenario. Two systems of different scales are considered in the simulation:

[0204] 1) Small scale: The system has 16 cells, each with 3 devices, for a total of 48 devices;

[0205] 2) Large scale: There are 40 cells in the system, with 6 devices in each cell, for a total of 240 devices.

[0206] Each device has a latency limit of 1ms, corresponding to 700 allocatable RUs in the system. The required transmission rate for each device ranges from 288 to 720 kbps. The table below summarizes the detailed parameters used in the simulation.

[0207] Table 1 Simulation Parameters

[0208]

[0209] (2) Performance of the proposed scheme when the small-scale fading compensation mechanism is not applied.

[0210] This section does not consider small-scale fading to evaluate the performance of the HCSA scheme without applying a small-scale fading compensation mechanism. This approach is applicable to most wireless networks where average performance, rather than instantaneous performance, is critical. To assess the impact of interference relationship awareness on resource allocation algorithm performance, this section also simulates the SGRA algorithm for both known and unknown interference relationships.

[0211] Figure 4 For small-scale networks, performance graphs of the proposed scheme and comparative schemes are presented, from... Figure 4 It is evident that the HCSA scheme proposed in this application can save resources compared to the SGRA scheme with known interference relationships in small-scale networks, achieving an average saving of 14.99% and a maximum saving of 19.5% under different device rate requirements. In contrast, for the StM scheme, the HCSA scheme proposed in this application can save an average of 49.76% of resources, and up to 56.58%.

[0212] Furthermore, the availability of interference relationships significantly impacts algorithm performance. In cases where interference relationships are unavailable, the SGRA algorithm can save 10-30% of resources by utilizing the interference relationships provided by the interference model, a significant difference.

[0213] Figure 5 Performance graphs of the proposed scheme and comparative schemes are presented for large-scale networks. In large-scale networks, the resource savings of the proposed HCSA scheme are even more substantial. Figure 5 It can be seen that, compared to the SGRA scheme when the interference relationship is known, the HCSA scheme can save an average of 23.24%, and at most 31.5% of resource usage, under different equipment rate requirements. Compared to the StM scheme, it can save an average of 56.19% of resources.

[0214] from Figure 4 , Figure 5 The excellent performance of the HCSA scheme proposed in this application can be clearly demonstrated in the data.

[0215] (3) When applying the small-scale fading compensation mechanism, the performance of the proposed scheme is evaluated.

[0216] This section also includes simulations of small-scale fading to evaluate the performance of the HCSA scheme in the most demanding URLLC scenarios, in combination with the small-scale fading compensation mechanism.

[0217] In this part, small scale network, rate requirement R REQ Simulation is performed for the scenario of 288kbps.

[0218] Figure 6 With Figure 7 The decoding error probability of the proposed scheme and the comparative schemes under different average transmission times is shown. Among them, Figure 6 The decoding error probability of the proposed scheme and the comparative schemes under different average transmission times is shown. Among them, Figure 7 The decoding error probability of the proposed scheme and the comparative schemes under different average transmission times is shown. Among them,

[0219] From Figure 6 , Figure 7 It can also be seen that the HCSA scheme proposed in the present application has a great improvement compared with all the comparative schemes. Under the condition of meeting the decoding error probability limit, the average transmission times required by the HCSA scheme is less than 1.5 times, which is more than 60% less than the more than 4 times required by the SGRA scheme with a 3dB protection margin, and the effect of saving resources is very significant.

[0220] In addition, in order to show the effectiveness of the "reserved protection margin" small scale fading compensation method mentioned in the present application, the SGRA scheme is also used for comparison in this section. Under the condition of obtaining the same interference relationship, the introduction of a 3dB protection margin for the SGRA scheme can reduce the decoding error probability by at least one order of magnitude, which fully shows the effectiveness of the reserved protection margin in guaranteeing QoS.

[0221] Figure 8 The decoding error probability of the proposed scheme and the comparative schemes under different RU occupation amounts when all RUs are counted is shown, Figure 8 The relationship between the RU occupation amount and the decoding error probability of each scheme is more directly shown. Combined with the small scale fading compensation mechanism, the HCSA scheme proposed in the present application can use more than 55% less resources than the SGRA scheme with a 3dB protection margin, which is the best performance among the comparative schemes, under the condition of meeting the decoding error probability limit.

[0222] Undoubtedly, the HCSA scheme proposed in the present application and the small scale fading compensation mechanism show excellent adaptability, combined with the advantages of the HCSA scheme itself, thereby achieving outstanding performance in this part, and clearly showing the superiority of the scheme proposed in the present application.

[0223] In order to realize the above-mentioned embodiments, the present application further proposes a wireless network interference coordination and resource scheduling device based on a hierarchical clustering algorithm.

[0224] Figure 9 A structure schematic diagram of a wireless network interference coordination and resource scheduling device based on a hierarchical clustering algorithm is provided for an embodiment of the present application.

[0225] As shown in the figure, the wireless network interference coordination and resource scheduling device based on the hierarchical clustering algorithm comprises an acquisition module, a clustering module, a clustering optimization module, and a resource allocation module, wherein Figure 9

[0226] The acquisition module is configured to acquire a set of user equipment whose demands are not satisfied.

[0227] The clustering module is configured to group user equipment in the set of user equipment into the same class if the mutual interference of the user equipment is less than a threshold value, and into different classes if the mutual interference of the user equipment is greater than the threshold value, to obtain an optimal non-overlapping clustering result.

[0228] The clustering optimization module is configured to adjust and optimize the optimal non-overlapping clustering result to cause overlapping between different classes, to obtain an optimal clustering result.

[0229] The resource allocation module is configured to allocate a wireless resource unit to each class according to the optimal clustering result, to stop the allocation if the demands of all the user equipment are satisfied or the resource is exhausted, and to obtain a final resource allocation result.

[0230] Optionally, in an embodiment of the present application, the hierarchical clustering algorithm is a condensed hierarchical clustering algorithm, and the clustering module is specifically configured to:

[0231] perform clustering on the set of user equipment according to the differences between the classes by using the condensed hierarchical clustering algorithm, to obtain a binary agglomerative tree, and to obtain a set of clustering results by setting different heights for the binary agglomerative tree, wherein the differences between the classes are determined by a linkage function, each user equipment in each class included in each clustering result uses the same resource, and different classes use different resources.

[0232] perform screening on the set of clustering results according to an expected resource occupation amount and an achievable total rate of each clustering result, to obtain the optimal non-overlapping clustering result, wherein the expected resource occupation amount is the sum of the maximum resource amounts expected to be required by all the classes in the clustering result, and the achievable total rate is the sum of the rates that can be achieved by all the classes in the clustering result, and the sum of the rates that can be achieved by each class is the sum of the rates that can be achieved by all the user equipment included in the class.

[0233] Optionally, in an embodiment of the present application, the clustering optimization module is specifically configured to:

[0234] ​Obtaining all classes with height less than a preset height in the binary convergence tree, to obtain a set of classes to be merged, wherein the preset height is the height corresponding to the optimal non-overlapping clustering result, and each element in the set of classes to be merged is assigned a subscript in descending order of its corresponding height;

[0235] Determining whether a current class to be merged in the set of classes to be merged satisfies a preset condition, and if so, merging the current class to be merged with the optimal non-overlapping clustering result, and updating the optimal non-overlapping clustering result using the merged optimal non-overlapping clustering result;

[0236] Continuously updating the optimal non-overlapping clustering result until all classes in the set of classes to be merged are determined, and obtaining a final optimal non-overlapping clustering result as the optimal clustering result;

[0237] The preset condition is that the current updated optimal non-overlapping clustering result has an empty intersection with the current class to be merged, all devices in the merged clustering use the same resource, and the resource utilization rate after merging is greater than that before merging, wherein the resource utilization rate after merging is greater than that before merging includes that the expected resource occupancy after merging is greater than that before merging or the expected resource occupancy after merging is equal to that before merging while the total achievable rate after merging is greater than that before merging.

[0238] Optionally, in an embodiment of the present application, the resource allocation module is specifically configured to:

[0239] Assigning 1 unit of wireless resource to each class in the optimal clustering result, and calculating the achievable rate of each user device after the assignment, wherein if a user device exists in multiple classes, the achievable rate of the user device is the sum of the achievable rates of the user device in each class;

[0240] Updating the rate requirement of each user device according to the achievable rate of each user device, and adding the current completed resource allocation to a resource allocation result list;

[0241] If the rate requirements of all user devices after the update are all reduced to 0 or the resource is exhausted, performing resource scheduling according to the current resource allocation result list, otherwise re-executing the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm.

[0242] It should be noted that the foregoing explanation and description of the embodiment of the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm also applies to the wireless network interference coordination and resource scheduling device based on the hierarchical clustering algorithm of this embodiment, which will not be described here.

[0243] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" means that a particular feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. The illustrative appearances of the above- described terms in various places in the specification are not intended to exclude that the terms can be combined with one another in other embodiments or examples of the application, nor are they intended to exclude that terms described in one embodiment or example can be combined with terms described in another embodiment or example. Although the application has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the scope of the application. Any further variations to the specific aspects of the application described herein will be apparent to those skilled in the art and are considered to fall within the scope of the application as defined by the appended claims.

[0244] In addition, the terms "first", "second", etc. are used herein only to describe various steps in a method, process, or algorithm. Thus, the terms "first", "second", etc. are not intended to, and should not be construed to, refer to a ranking or order of importance of or relative importance of, or superiority or inferiority of, the various steps in the method, process, or algorithm, but instead are used merely to distinguish one step from another. In addition, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. Further, the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated items, including items that are the same as or different from each other.

[0245] Any process or method described in a flowchart or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions (or steps) of the process, and alternate implementations are possible. In some embodiments, the processes or methods described in this application can be tailored or varied by those skilled in the art to include more or less steps than those expressly described, and to use alternative arrangements, structures, or variations of the steps described herein without departing from the scope of the present application.

[0246] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a sequence of instructions to implement logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a computer- readable storage medium or a computer-readable signal medium. The computer- readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include the following: an electrical connection having one or more wires (electrical connections), a portable computer diskette (a magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and stored in a computer memory.

[0247] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. As such, in some embodiments, specifically configured hardware can be used to implement at least some of the functionality described herein. For example, if implemented in hardware, the hardware can include any or a combination of the following: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0248] Those of skill in the art would understand that information and signals can be represented using any of a variety of technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that can be referenced throughout the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0249] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0250] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for wireless network interference coordination and resource scheduling based on hierarchical clustering algorithm, characterized in that, The method comprises the following steps: obtaining a set of user equipment whose demands are not met; classifying the user equipment set into the same class according to a hierarchical clustering algorithm if the interference between the user equipment is less than a threshold value, and into different classes if the interference is greater than the threshold value, to obtain an optimal non-overlapping clustering result; adjusting and optimizing the optimal non-overlapping clustering result to cause overlapping between different classes, to obtain an optimal clustering result; allocating a unit of wireless resource to each class according to the optimal clustering result, and stopping the allocation when the demands of all the user equipment are met or the resource is exhausted, to obtain a final resource allocation result; wherein the hierarchical clustering algorithm is a hierarchical agglomerative clustering algorithm, and classifying the user equipment set according to the hierarchical clustering algorithm comprises: clustering the user equipment set according to the difference between the classes by the hierarchical agglomerative clustering algorithm, to obtain a binary agglomerative tree, and setting different heights for the binary agglomerative tree to obtain a set of clustering results, wherein the difference between the classes is determined by a linkage function, and each class contained in each clustering result uses the same resource, and different classes use different resources; screening the set of clustering results according to the expected resource occupation amount and the achievable total rate of each clustering result, to obtain the optimal non-overlapping clustering result, wherein the expected resource occupation amount is the sum of the maximum resource amount expected to be required by all the classes in the clustering result, and the achievable total rate is the sum of the rates that can be achieved by all the classes in the clustering result, and the sum of the rates that can be achieved by each class is the sum of the rates that can be achieved by all the user equipment contained in the class; the adjusting and optimizing of the optimal non-overlapping clustering result comprises: obtaining all the classes in the binary agglomerative tree whose heights are less than a preset height, to obtain a set of classes to be merged, wherein the preset height is the height corresponding to the optimal non-overlapping clustering result, and each element in the set of classes to be merged is assigned a subscript in descending order of the corresponding height; judging whether a current class to be merged in the set of classes to be merged meets a preset condition, and if so, merging the current class to be merged with the optimal non-overlapping clustering result, and updating the optimal non-overlapping clustering result using the merged optimal non-overlapping clustering result; continuously updating the optimal non-overlapping clustering result until all the classes in the set of classes to be merged are judged, to obtain the final optimal non-overlapping clustering result as the optimal clustering result; wherein the preset condition is that the intersection of the current updated optimal non-overlapping clustering result and the current class to be merged is empty, and all the user equipment in the merged clustering result use the same resource, and the resource utilization rate after the merging is greater than that before the merging, wherein the resource utilization rate after the merging is greater than that before the merging includes that the expected resource occupation amount after the merging is greater than that before the merging, or the expected resource occupation amount after the merging is equal to that before the merging while the achievable total rate after the merging is greater than that before the merging.

2. The method of claim 1, wherein, the determination process of the linkage function is: defining a distance function for determining the difference between the user equipment, and determining the linkage function according to the distance function, wherein the distance function is expressed as: the linkage function is expressed as: wherein indicates a device U p as a signal device, a device U q the reciprocal of the signal-to-interference ratio of the interference, indicates a device U q as a signal device, a device U p the reciprocal of the signal-to-interference ratio of the interference, j p indicates a device U p the associated base station number, j q indicates a device u q the associated base station number, j q indicates a device u q the associated base station number, indicates the mth cluster in the clustering result, denotes the nth cluster in the clustering result, d(p, q) denotes the difference between device p and device q.

3. The method of claim 1, wherein, The method comprises the following steps: allocating one unit of wireless resource to each class in the optimal clustering result, and calculating the achievable rate of each user equipment after the allocation, wherein if a user equipment exists in multiple classes, the achievable rate of the user equipment is the sum of the achievable rates of the user equipment in the multiple classes; updating the rate demand of each user equipment according to the achievable rate of each user equipment, and adding the current completed resource allocation to a resource allocation result list; if the rate demand of all user equipments after the updating is reduced to 0 or the resource is exhausted, performing resource scheduling according to the current resource allocation result list, otherwise re-executing the method.

4. The method of claim 2, wherein, if the user equipment set has services with QoS demand greater than a threshold, applying a small-scale fading compensation mechanism to the method, comprising: when classifying the user equipment set using the hierarchical clustering algorithm and adjusting and optimizing the optimal non-overlapping clustering result, taking the sum of the achievable rates of all user equipments after applying a fading protection boundary as the total achievable rate, and determining the optimal transmission times, and taking the product of the optimal transmission times and the expected maximum resource amount as the expected resource occupation amount; when allocating wireless resource units to each class according to the optimal clustering result, the wireless resource units allocated to each class are changed from one to the optimal transmission times, and the achievable rate of a user equipment after applying a fading protection boundary is taken as the achievable rate of the user equipment.

5. A wireless network interference coordination and resource scheduling apparatus based on hierarchical clustering algorithm, characterized in that, The method comprises an acquisition module, a clustering module, a clustering optimization module, and a resource allocation module, wherein: the acquisition module is configured to acquire a user equipment set with unmet demand; the clustering module is configured to classify devices with mutual interference less than a threshold in the user equipment set into the same class, and devices with mutual interference greater than a threshold into different classes, to obtain an optimal non-overlapping clustering result, using a hierarchical clustering algorithm; the clustering optimization module is configured to adjust and optimize the optimal non-overlapping clustering result to cause overlap between different classes, to obtain an optimal clustering result; the resource allocation module is configured to allocate wireless resource units to each class according to the optimal clustering result, and stop the allocation when the demand of all devices is met or the resource is exhausted, to obtain a final resource allocation result; wherein the hierarchical clustering algorithm is a condensation-type hierarchical clustering algorithm, and the clustering module is specifically configured to: cluster the user equipment set according to the differences between classes using the condensation-type hierarchical clustering algorithm, to obtain a binary agglomerative tree, and set different heights for the binary agglomerative tree to obtain a clustering result set, wherein the differences between classes are determined by a linkage function, and each class contained in each clustering result uses the same resource, and different classes use different resources. screening the set of clustering results according to expected resource occupation and reachable total rate of each clustering result, to obtain the optimal non-overlapping clustering result, wherein the expected resource occupation is the sum of maximum resource required by all classes in the clustering result, and the reachable total rate is the sum of rates reachable by all classes in the clustering result, and the sum of rates reachable by each class is the sum of rates reachable by all devices contained in the class; The clustering optimization module is specifically configured to: obtain all classes with a height less than a preset height in the binary convergence tree, to obtain a set of classes to be merged, wherein the preset height is a height corresponding to the optimal non-overlapping clustering result, and each element in the set of classes to be merged is assigned a subscript in descending order of the corresponding height; determine whether a current class to be merged in the set of classes to be merged meets a preset condition, if so, merge the current class to be merged with the optimal non-overlapping clustering result, and update the optimal non-overlapping clustering result using the merged optimal non-overlapping clustering result; constantly update the optimal non-overlapping clustering result until all classes in the set of classes to be merged are determined, to obtain a final optimal non-overlapping clustering result as the optimal clustering result; wherein the preset condition is that the intersection of the current updated optimal non-overlapping clustering result and the current class to be merged is empty, and all devices in the merged clustering result use the same resource, and the resource utilization rate after merging is greater than that before merging, wherein the resource utilization rate after merging is greater than that before merging includes that the expected resource occupation after merging is greater than that before merging or the expected resource occupation after merging is equal to that before merging while the reachable total rate after merging is greater than that before merging.

6. The apparatus of claim 5, wherein, The resource allocation module is specifically configured to: allocate one unit of wireless resource to each class in the optimal clustering result, and calculate the reachable rate of each user device after allocation, wherein if a user device exists in multiple classes, the reachable rate of the user device is the sum of the reachable rates of the user device in each class; update the rate requirement of each user device according to the reachable rate of each user device, and add the current completed resource allocation to a resource allocation result list; if the rate requirement of all user devices after updating is reduced to 0 or the resource is exhausted, perform resource scheduling according to the current resource allocation result list, otherwise, re-execute the wireless network interference coordination and resource scheduling method based on the hierarchical clustering algorithm.

Images