6G slice cache resource allocation device
By adopting a cache resource allocation algorithm based on evolutionary public goods game in the 6G communication network, the problem of imbalance in cache resource allocation between sliced virtual nodes is solved, and more efficient data transmission performance and lower network congestion are achieved.
Patent Information
- Application Number
- CN202311532511.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-17
- Publication Date
- 2025-05-30
AI Technical Summary
In 6G communication network, when multiple slice virtual nodes share the same underlying physical node, as the number of slice virtual nodes increases, cache resource allocation is unbalanced, resulting in data stacking loss, slice data transmission performance deteriorates, and network performance is affected.
A cache resource allocation algorithm based on evolutionary public goods game is proposed. By defining the cache resource difference and evolutionary game strategy, resource investment is made when the cache resources are sufficient, and investment is reduced when the resources are insufficient or the demand is large, so as to realize intelligent allocation of cache resources.
By intelligently allocating cache resources, data transmission performance is improved, network congestion is reduced, and overall network performance is improved.
Smart Images
Figure CN120075057A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of resource allocation, and particularly to a 6G slice cache resource allocation device. Background Art
[0002] The 6G communication network will support various diverse applications (such as unmanned driving, AR / VR, digital twin, and holographic communication, etc.), and these applications have different service quality requirements. To meet these requirements, the network slicing technology divides the 6G communication network into multiple slices, and each slice can be customized according to application requirements, so as to meet various service quality requirements. Therefore, the network slicing technology has become one of the key technologies of the 6G communication network.
[0003] Although network slicing can meet the diversity of 6G communication, however, when multiple slice virtual nodes share the same underlying physical node, with the increase in the number of slice virtual nodes carried on the same underlying physical node, due to the limited cache resources on the physical node, there is often an imbalance in resource allocation. When the cache resources are insufficient to accommodate the incoming data, it may lead to the accumulation and loss of data, and the data transmission performance on the slice will inevitably be negatively affected to a certain extent, resulting in a decline in network performance. Therefore, reasonable slice cache resource allocation is of great significance for improving network performance. Summary of the Invention
[0004] Aiming at the performance degradation problem caused by unreasonable cache resource allocation in slices, this paper proposes a 6G slice cache resource allocation device. The method of this device constructs a 6G network slice model based on cache resource competition, which depicts the virtual node mapping and resource sharing relationship, and then proposes a cache resource allocation algorithm based on evolutionary public goods game aiming at improving data transmission performance; this algorithm defines the cache resource difference, and selects the input cost according to the cache resource difference and the evolutionary game strategy, invests resources when the cache resources are sufficient, and reduces unnecessary investment to avoid resource waste in the case of insufficient cache resources or high demand, so as to realize the intelligent allocation of cache resources.
[0005] The cache resource allocation method for 6G slices described above includes the following steps:
[0006] 1) Establish a 6G network slice model based on cache resource competition.
[0007] 2) Establish a cache resource allocation algorithm based on evolutionary public goods game.
[0008] The method for establishing a 6G network slice model based on cache resource competition in the above step 1 is as follows:
[0009] In the 6G network slicing model based on cache resource competition, there are two layers, namely the slice layer and the underlying infrastructure network layer, with P = (L 0 , L, γ); at the slice layer, the layer is represented by L = {L 1 , L 2 , ..., L i , ..., L M} represents the entire slice, M is the total number of slice layers; each single slice L i = {V i , κ i} can represent the set of virtual nodes and edges in the slice layer; where V i The nodes in the slice layer are a set of virtual nodes, and the virtual nodes represent the virtual network functions (VNFs) of the slice; the set of virtual nodes in each slice layer is a subset of the set of physical nodes in the underlying infrastructure network; κ i It is the virtual link between each virtual node in the slice layer; in the underlying infrastructure network layer, layer L 0 Node is a set of physical nodes, where a physical node represents a physical device in the infrastructure network. 0 is the number of physical nodes;
[0010] In the above step 2, a method for establishing a cache resource allocation algorithm based on the evolutionary public goods game is as follows:
[0011] The present invention proposes a cache resource allocation algorithm based on evolutionary public goods game. The algorithm defines the cache resource difference and selects the investment cost according to the cache resource difference and the evolutionary game strategy. When the cache resources are sufficient, resources are invested. When the cache resources are insufficient or the demand is large, unnecessary investment is reduced to avoid resource waste, thereby realizing intelligent allocation of cache resources. The algorithm is defined as follows:
[0012] Define the resource difference ψ in the tth round of game i (t) is the total resource of the slice virtual node in round t-1 and the resource demand ξ in round t i (t); the resource difference is calculated as follows:
[0013]
[0014] In the formula, the total resources of the slice node in round t-1 are is the resource allocated to the node after the t-1th round of game, that is, the game income α i (t-1) and the remaining resources β in round t-1 i (t-1); the total resources of the node in the t-1 round of game are:
[0015]
[0016] Among them, the remaining resources are the resources of the uninvested cost remaining after the end of this round of game; if t = 1, the remaining resources of the previous round are 0, and the resources obtained in the previous round are the initial resources; if resource investment is made in this round of game, the remaining resources of this round are the resource requirements of this round; if no resource investment is made in this round of game, the remaining resources are the total resources of the previous round; the remaining resources of the t-th round of game are:
[0017]
[0018] Therefore, the resource difference formula can be:
[0019]
[0020] For the slice virtual node, the investment cost c i (t) in the t-th round of game is as follows:
[0021]
[0022] Slice virtual node V i If the cooperation strategy is adopted and the resource difference of the slice virtual node in this round of game is greater than 0, the resources α i (t) (i.e., α i，C (t)) that the slice virtual node can obtain are:
[0023]
[0024] Slice virtual node V i If the betrayal strategy is adopted or the resource difference of the slice virtual node in this round is less than 0, the resources α i (t) (i.e., α i，D (t)) that the slice virtual node can obtain are:
[0025]
[0026] It can be seen from this that the investment cost of the slice virtual node in this round of game is jointly determined by the current game strategy and the resource difference in this round. When the resource difference in this round is greater than 0 and the current slice virtual node adopts a cooperative game strategy, the slice virtual node will invest the excess resources into the shared pool to share with others; when the resource difference in this round is less than or equal to 0 or the current slice virtual node adopts a betrayal game strategy, the slice virtual node will choose not to invest the cost; after all slice virtual nodes in the game group complete the game, calculate the total resources in the shared pool, and then evenly distribute the total resources in the shared pool to all slice virtual nodes in this game group.
[0027] The present invention also provides a 6G slice cache resource allocation device, and the device includes:
[0028] A receiving module, configured to receive the resource requirements of nodes within each layer of slices.
[0029] A selection module, configured to determine the game strategy of slice nodes from a set of game strategies.
[0030] An allocation module, configured to determine the game revenue of slice nodes according to the resource difference and the game strategy.
[0031] This embodiment is a storage medium, in which at least one instruction is stored, and the at least one instruction is loaded and executed by a processor to implement the cache resource allocation method for 6G slices.
[0032] This embodiment is a device, which includes a processor and a memory. At least one instruction is stored in the memory, and the at least one instruction is loaded and executed by the processor to implement the cache resource allocation method for 6G slices. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings, where:
[0034] Figure 1 It is a 6G network slice model based on cache resource competition.
[0035] Figure 2 It is the change of the packet arrival rate under different resource allocation strategies.
[0036] Figure 3 The change of the packet arrival rate of different resource allocation strategies under different slice layers.
[0037] Figure 4 The change of the average network congestion degree under different resource allocation strategies.
[0038] Figure 5 The change of the packet arrival rate of different resource allocation strategies under different slice layers. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0039] The following further describes in detail the specific embodiments of the present invention with reference to the accompanying drawings.
[0040] In Figure 1 The established 6G network slice model based on cache resource competition includes two layers, namely the slice layer and the underlying infrastructure network layer, which is represented by P=(L 0 , L, γ); in the slice layer, this layer consists of L={L 1 , L 2 ,..., L i ,..., LM} represents the entire slice, where M is the total number of layers of the slice; each single-layer slice L i = {V i , κ i} can represent the set of virtual nodes and the connecting edges within the slice layer; among them, V i is the set of virtual nodes within the slice layer, and the virtual nodes represent the virtual network functions (VNFs) of the slice; the set of virtual nodes of each layer of the slice is a subset of the set of physical nodes of the underlying infrastructure network; κ i is the virtual link connecting the virtual nodes within the slice layer; at the underlying infrastructure network layer, the nodes 0 of layer L are the set of physical nodes, and the physical nodes represent the physical devices of the infrastructure network, where N 0 is the number of physical nodes;
[0041] Virtual nodes from multiple slices are deployed on the physical nodes of the underlying infrastructure network, that is, the virtual nodes in the slice will be mapped to the physical nodes in the infrastructure network. This means that the corresponding virtual nodes in multiple slice layers will share the resources on this physical node, and this resource sharing relationship is reflected by the mapping relationship established between the physical node and the virtual nodes deployed on it. For example: In the figure, the virtual nodes V 2 in slice layers L 3 , L M and L 1 all occupy the resources on the underlying physical node P 1 , which means that L 2 , L 3 , L M compete for the resources of the physical node P 1 . From this, it can be seen that when the same virtual nodes are deployed among different slices under the limited underlying infrastructure resources, these virtual nodes will share the resources of the physical nodes with the corresponding mapping relationships in the underlying infrastructure network, resulting in resource competition among slices. This competition may lead to serious congestion and inefficiency, thus affecting the performance of the entire network slice.
[0042] At Figure 2In the simulation, when comparing different data packet generation rates, the changes in the successful arrival rates of data packets under different resource allocation strategies are shown in Figures (a) and (b). In the figures, the horizontal axis represents the data packet generation rate, and the vertical axis represents the successful arrival rate of data packets. In addition, the blue, green, and pink curves in both figures represent Slice Layer 1, Slice Layer 2, and Slice Layer 3 respectively. The curves with circles and triangles in the figures represent the strategy proposed in the present invention, RANN, and RAND respectively. It can be seen from the figures that compared with the two strategies of RANN and RAND, the successful arrival rate of data packets in the three-layer slices under this strategy is the largest, and the data packet arrival rates under the two strategies increase as the data packet generation rate increases. This is because the strategy proposed in this paper takes into account the resource requirements of slice nodes within a layer and, through the idea of game theory, considers the competition relationship between slice nodes, enabling each slice node to decide how much resources to allocate according to the actual situation. It can achieve a higher data packet arrival rate when the data packet generation rate is relatively large. The RANN strategy, without considering the characteristics of nodes, has more cache resources in the slice layer with more node numbers. The resources within the layer are evenly distributed to each node. However, this method may cause an imbalance in resource allocation between high-degree nodes and low-degree nodes, thus affecting the caching effect and transmission success rate of data packets. The RAND strategy takes into account the node degree. When allocating the buffer size to nodes, high-degree nodes obtain more resources. This strategy can better adapt to the imbalance between nodes compared with the RANN strategy, but in some cases, it may lead to overcrowding of high-degree nodes while other nodes are idle, resulting in network congestion. In addition, from Figure 2 it can be seen from (a) that the average successful arrival rate of data packets of this strategy is 0.271. The average successful arrival rate of data packets of RANN is 0.258. From Figure 2 it can be seen from (b) that the average arrival rate of data packets of this strategy is 0.255, and the average arrival rate of data packets of RAND is 0.228. In summary, the overall data packet arrival rate of this strategy is increased by 4.79% and 10.5% compared with the latter respectively. Therefore, it is proved that adopting this strategy can improve the accuracy of the data packet arrival rate of slices.
[0043] In Figure 3In the simulation, at the same generation rate, the changes in the successful arrival rate of data packets under different resource allocation strategies are compared, as shown in the figure. In the figure, the horizontal axis represents the number of slice layers, and the vertical axis represents the successful arrival rate of data packets. In addition, the bar charts represent the successful arrival rate of data packets under the RAND strategy, the successful arrival rate of data packets under the RANN strategy, and the successful arrival rate of data packets under this strategy from left to right. It can be seen from the figure that when the number of slice layers increases from 3 to 5, the successful arrival rate of data packets under the RAND strategy, the RANN strategy, and the current strategy decrease from [0.1567, 0.1167, 0.1633] to [0.1240, 0.1241, 0.1360], respectively. This is because the total cache of the underlying infrastructure network and the generation rate of data packets remain unchanged. As the number of slice layers increases, the number of nodes increases accordingly. The average cache resources available to each node decrease, and more nodes need to store and transmit data packets in a limited cache space, which may lead to more intense resource competition. The competition for cache resources between nodes may affect the transmission of data packets, resulting in a decrease in the successful arrival rate. This may cause nodes to encounter insufficient buffering when storing and transmitting data packets, increasing the possibility of data packet loss. In addition, the improved public goods game method of this strategy takes into account information such as node resource requirements and network topology, helping nodes to make reasonable allocations when resources are limited, avoiding overly fierce resource competition between nodes, and thus reducing the possibility of data packet loss.
[0044] exist Figure 4 In the three-layer slice with a time step of 50, the change law of the network congestion degree when simulating and comparing different resource allocation strategies is shown in the figure. In the figure, the horizontal axis represents the generation rate of data packets, and the vertical axis represents the average congestion degree of the network. Among them, the circle, triangle and diamond curves in the figure represent the RAND strategy, RANN strategy and this strategy respectively. It can be seen from the figure that as the data packet generation rate increases, the average congestion degree of the network increases. This is because when the data packet generation rate increases, the number of generated data packets will also increase, while the cache size of the nodes in the layer remains unchanged, and the arrival speed of the data packets exceeds the cache processing capacity of the nodes, resulting in performance degradation. In other words, this indicator will increase when the data packet generation rate increases, indicating that the node cache is not enough to meet the needs of data packet arrival, and the performance may be affected. In addition, it can be seen from the figure that the average congestion degree of this strategy is 0.5109, while the average congestion degrees of the RAND strategy and the RANN strategy are 0.6781 and 0.6632 respectively. In summary, the overall network congestion degree of this strategy is improved by 24.6% and 22.9% respectively compared with the latter. Therefore, it is proved that the adoption of this strategy can reduce the congestion level of the slice and improve the data transmission performance of the slice.
[0045] exist Figure 5Among them, at the same generation rate, the simulation compares the change law of the network congestion degree under different resource allocation strategies, as shown in the figure. In the figure, the horizontal axis represents the number of slice layers, and the vertical axis represents the network congestion degree. In addition, the bar charts from left to right represent the network congestion degree of the slice under the RAND strategy, the network congestion degree under the RANN strategy, and the network congestion degree under this strategy respectively. It can be seen from the figure that when the number of slice layers increases from 3 to 5, the network congestion degrees of the RAND strategy, the RANN strategy, and this strategy increase from [0.3514, 0.3006, 0.2772] to [0.8287, 0.8752, 0.7893] respectively, indicating that a large number of nodes are in a congested state. This is because, with the physical node cache resources of the underlying infrastructure network remaining unchanged, as the number of slice layers increases, it means that the total cache resources of each layer of nodes decrease, and at the same time, more nodes in different slice layers will share the cache resources of the physical nodes in the underlying infrastructure network. In addition, compared with the other two strategies, the proposed strategy in this paper fluctuates up and down with the increase of the number of slice layers and maintains a relatively low network congestion. This is because this strategy allocates resources according to the resource requirements of each node, taking into account the personalized needs of the nodes and the mutual influence between nodes. Each node can conduct resource games with other nodes according to its own needs to obtain a more appropriate resource allocation. This method is more flexible and personalized, can better adapt to the characteristics of different nodes, and avoids the congestion problem caused by uneven resource distribution among nodes.
Claims
1. A caching resource allocation method for 6G slices is established, characterized by the following steps: 1) Establish a 6G network slice model based on caching resource competition; 2) Establish a caching resource allocation algorithm based on evolutionary public goods game.
2. The method for establishing a caching resource allocation method for 6G slices as claimed in claim 1, wherein in step 1), the method for establishing a 6G network slice model based on caching resource competition is as follows: The present invention establishes a 6G network slicing model based on cache resource competition, which includes two layers, namely the slicing layer and the underlying infrastructure network layer, represented by P=(L 0 , L, γ); in the slicing layer, this layer is represented by L={L 1 , L 2 ,..., L i ,..., L M} to represent the entire slice, and M is the total number of layers of the slice; each single-layer slice L i ={V i , κ i} can represent the set of virtual nodes and the connecting edges within the slicing layer; Wherein, V i The nodes in the slice layer are a set of virtual nodes, and the virtual nodes represent the virtual network functions (VNFs) of the slice; the set of virtual nodes of each slice layer is a subset of the set of physical nodes of the underlying infrastructure network; κ i is the virtual link connecting the virtual nodes within the slice layer; at the underlying infrastructure network layer, layer L 0 of the nodes is the set of physical nodes, and the physical nodes represent the physical devices of the infrastructure network, N 0 is the number of physical nodes.
3. The method for establishing a caching resource allocation method for 6G slices as claimed in claim 1, wherein in step 2), the method for establishing a caching resource allocation algorithm based on evolutionary public goods game is as follows: The present invention proposes a caching resource allocation algorithm based on evolutionary public goods game. This algorithm defines the caching resource difference, and selects the input cost according to the caching resource difference and the evolutionary game strategy. When the caching resources are sufficient, resources are invested. In the case of insufficient caching resources or high demand, unnecessary investment is reduced to avoid resource waste, thereby realizing the intelligent allocation of caching resources. The definition is as follows: Define the resource difference ψ of the t-th round of the game i The total resources of the (t - 1)-th round of the slice virtual node And the resource requirement ξ of the t-th round i (t) difference; the calculation formula of the resource difference is: Wherein, the total resources of the slice node in the (t-1)-th round is the resources obtained by the node through the game in the (t-1)-th round, that is, the game income α i (t-1) and the remaining resources β i (t-1) in the (t-1)-th round; the total resources of the node in the (t-1)-th round of the game are: Wherein, The remaining resources are the resources that have not been invested with cost after the end of this round of game; if t = 1, the remaining resources of the previous round are 0, and the resources obtained in the previous round are the initial resources; if resources are invested in this round of game, the remaining resources of this round are the resource demand of this round; if no resources are invested in this round of game, the remaining resources are the total resources of the previous round; the remaining resources of the t-th round of game are: Therefore, the resource difference formula can be: For the sliced virtual node, the input cost c i (t) in the t-th round of the game is as follows: Slice virtual node V i If the cooperation strategy is adopted and the resource difference of the slice virtual node in this round of game is greater than 0, then the resource α i (t) (i.e., α i，C (t)) is as follows: Slice virtual node V i If the betrayal strategy is adopted or the resource difference of the slice virtual node in this round is less than 0, the resource α i (t) (i.e., α i，D (t)) is as follows: It can be seen from this that the input cost of the slice virtual node in this round of game is jointly determined by the current game strategy and the resource difference in this round. When the resource difference in this round is greater than 0 and the current slice virtual node adopts a cooperative game strategy, the slice virtual node will invest the excess resources into the shared pool to share with others; when the resource difference in this round is less than or equal to 0 or the current slice virtual node adopts a betrayal game strategy, the slice virtual node will choose not to invest cost; after all slice virtual nodes in the game group complete the game, calculate the total resources in the shared pool, and then evenly distribute the total resources in the shared pool to all slice virtual nodes in this game group.
4. A 6G slice caching resource allocation device, the device comprises: a receiving module, configured to receive the resource demands of nodes in each layer of slice; a selection module, configured to determine the game strategy of the slice node from the game strategy set; an allocation module, configured to determine the game revenue of the slice node according to the resource difference and the game strategy.
5. A storage medium, characterized in that, at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the caching resource allocation method for 6G slices as claimed in any one of claims 1 to 3.
6. A device, characterized in that, the device includes a processor and a memory, and at least one instruction is stored in the memory, and the at least one instruction is loaded and executed by the processor to implement the caching resource allocation method for 6G slices as claimed in any one of claims 1 to 3.