Retransmission scheduling method based on cross-slice HARQ protocol

Abstract

The invention discloses a retransmission scheduling method based on a cross-slice HARQ protocol, and the method comprises the steps: constructing a scheduling model in a multi-slice scene of an RAN side network; on the basis of a scheduling model, a cross-slice retransmission protocol is designed by considering the limitation of a retransmission protocol in a network slice scene, and a double-layer optimization problem of throughput optimization based on the cross-slice protocol and on the premise of user guarantee is constructed. A user-level target rate and a slice-level uplink and downlink switching decision are optimized by deploying a flexible uplink and downlink switching design on an upper layer, then an adaptive retransmission parameter design and a cross-slice retransmission design are deployed based on an upper layer optimization decision, and real-time resource allocation is performed on each user according to a channel condition. According to the method, resource allocation decision optimization is achieved, the retransmission rate is reduced through protocol deployment and scheduling, the total network throughput can be improved on the premise that user guarantee is met, and retransmission scheduling based on the cross-slice HARQ protocol is completed. According to the invention, the optimization of the target rate decision, the uplink and downlink switching decision and the resource allocation decision can be realized, so that the retransmission rate is effectively reduced, and the throughput is improved under the condition of satisfying the user guarantee.

Description

Technical Field

[0001] This invention relates to a technology in the field of wireless communication, specifically a retransmission scheduling method based on the cross-slice HARQ protocol. Background Technology

[0002] Network slicing is an effective method for enabling diverse applications without building dedicated network infrastructure. It achieves end-to-end SLAs by allocating network resources individually. Network slicing includes core network slicing and radio access network (RAN) slicing. While core network slicing has successfully deployed software-defined networking and network virtualization technologies in upper-layer protocols, many challenges remain with RAN slicing. RAN slicing can be further divided into soft slices and hard slices. Soft slices in upper-layer protocols can introduce security and flexibility issues, while hard RAN slicing using isolated radio resources requires careful consideration of interactions with physical layer transmission schemes. To make hard RAN slicing more practical, recent work has focused on its interaction with the physical transmission environment. Nevertheless, existing hard RAN slicing schemes rarely consider interactions with many transport protocols, such as Hybrid Automatic Repeat Request (HARQ). Therefore, how to meet network performance (throughput) requirements under network slicing technology based on HARQ protocols while ensuring user guarantees (latency requirements) is a pressing issue. Summary of the Invention

[0003] This invention addresses the technical problems arising from the interaction between existing HARQ protocols and network slicing, including limitations imposed by stop-and-go protocols, the inability to retransmit across slices, and competition between retransmissions and uplink / downlink handover, which lead to throughput performance limitations while meeting user guarantees. It proposes a retransmission scheduling method based on a cross-slice HARQ protocol. This method implements a cross-slice HARQ protocol design, incorporates adaptive retransmission parameter adjustments to address stop-and-go protocol limitations while adhering to standards, designs cross-slice retransmission to resolve the inability to retransmit across slices, and designs flexible uplink / downlink handover to address uplink / downlink handover competition. This achieves optimization of target rate decisions, uplink / downlink handover decisions, and resource allocation decisions under the proposed protocol, effectively reducing the retransmission rate and improving throughput while meeting user guarantees.

[0004] This invention is achieved through the following technical solution:

[0005] This invention relates to a retransmission scheduling method based on a cross-slice HARQ protocol. It constructs a scheduling model in a multi-slice network scenario on the RAN side. Based on this model, it designs a cross-slice retransmission protocol considering the limitations of the retransmission protocol in the network slice scenario. It then constructs a two-layer optimization problem based on the cross-slice protocol, prioritizing user-level performance. By deploying flexible uplink / downlink switching design at the upper layer, it optimizes user-level target rates and slice-level uplink / downlink switching decisions. Furthermore, based on the upper-layer optimization decisions, it deploys adaptive retransmission parameter design and a cross-slice retransmission design. Real-time resource allocation is performed for each user according to channel conditions, resulting in optimized resource allocation decisions. Through protocol deployment and scheduling, the retransmission rate is reduced, improving the overall network throughput while meeting user-level performance requirements, thus completing the retransmission scheduling based on the cross-slice HARQ protocol. Technical effect

[0006] This invention considers the interaction between slicing and retransmission protocols. The cross-slice retransmission protocol, including adaptive retransmission parameter design, cross-slice retransmission design, and flexible uplink / downlink handover design, addresses the limitations of existing protocols such as the inability to cross slices and uplink / downlink handover restrictions. By combining the cross-slice protocol with RAN slices, the performance of existing protocols is enhanced, optimizing uplink / downlink handover decisions, target rate decisions, and resource allocation decisions during the scheduling process. By modeling a two-layer problem in the scheduling process, uplink / downlink handover decisions and target rate decisions are optimized at the upper layer, while resource allocation decisions are optimized at the lower layer. Corresponding cross-slice retransmission protocol designs are deployed for each layer, achieving throughput optimization under bidirectional user protection. Compared with existing technologies, this invention significantly improves throughput performance and is suitable for practical network deployments. Attached Figure Description

[0007] Figure 1 This is a flowchart of the present invention;

[0008] Figure 2 System model diagram;

[0009] Figure 3 A comparison chart of throughput and retransmission rate for different protocols under different signal-to-noise ratios;

[0010] Figure 4(A) is a comparison of throughput and user satisfaction under different numbers of users in the embodiment; Figure 4(B) is a comparison of latency under different numbers of users in the embodiment. Detailed Implementation

[0011] like Figure 1 As shown, this embodiment illustrates a retransmission scheduling method based on the cross-slice HARQ protocol for an Orthogonal Frequency Division Multiplexing (OFDMA) system, specifically including:

[0012] Step 1: Establish a model scenario in a multi-slice RAN network, then define the resource allocation process, uplink / downlink handover and retransmission process, and related scheduling parameters. Finally, provide system performance metrics, including:

[0013] 1.1) Model Scenario: This embodiment considers an Orthogonal Frequency Division Multiplexing (OFDMA) system, which has N RS There are several Radio Access Network (RAN) slices. In the nth slice, k ∈ U n Each User Equipment (UE) communicates with a Base Station (BS) via a downlink Incremental Redundancy (IR) protocol. Define N. S N T and N F These represent the number of subframes per time frame, the total number of time frames, and the number of subbands per network slice, respectively. The subframe and subband indices can be represented as t∈{1,…,N}. S N S +1,…,N S ×N T} and f∈{1,2,…,N F ×N RS Each RAN slice contains N S ×N F Resource blocks (RBs). Let S n ={(t,f)|t∈{1,…,N} S N S +1,…,N S ×N T},f∈{n×N F +1,…,(n+1)×N F} represents the radio resource block belonging to the nth slice. At the BS end, at the beginning of each time frame, the network slice scheduler manages the uplink / downlink handover decisions for slice n based on the base station environment. Target transmission rate decision for user k The network slice scheduler then matches and allocates RBs to different UEs in each subframe based on the real-time channel environment and upper-layer decisions, thus obtaining resource allocation decisions. The transmission and retransmission process is abstracted as The resulting equivalent capacity and A comparison between them. When the BS receives a successful transmission signal based on the HARQ protocol, the data packet will be cleared from the relevant uplink and downlink buffers. The specific parameter is set to N. RS =3, consider three slice scenarios, U n = 1 / 2 / 3, the number of users accessing the slice will vary in different experiments, N S =10 indicates the number of subframes contained in each frame, N T=1000, indicating that the total simulation time frames are 1000, N F =50RB(20M) means that each slice has 50 available resource blocks, and the total number of resource blocks in the system is 150.

[0014] 1.2) Resource allocation process: The set of RBs allocated to the k-th UE in the r-th time frame is the resource allocation decision. f k,m ∈{1,2,…,N F N RS Therefore, the equivalent capacity of the k-th UE in the r-th time frame can be calculated by the resource allocation decision. Where: m≥0 represents the index of the retransmission count, {m=0} represents the first transmission, and {m>0} represents the m-th retransmission; The signal-to-noise ratio (SNR) is 5-30 dB, and its purpose is to explore the effectiveness of the protocol under different channel conditions. It follows a Rayleigh distribution and represents the normalized channel fading coefficient.

[0015] 1.3) Uplink / Downlink Handover and Retransmission Process: Downlink of the nth network slice in the rth time frame. and uplink In the full-duplex mode of transmission, during downlink transmission, after soft combining, when the accumulated channel capacity exceeds the desired transmission rate, the acknowledgment (ACK) signal will be valid. In uplink transmission, by comparison and Determine the ACK / NACK signal, specifically: indicator function Right now express ,in: Indicates the target rate. This indicates the current retransmission round. After determining the uplink and downlink modes, the uplink and downlink transmission buffer states of the k-th user in the r-th time frame can be updated as follows: and Where: {α u (k)}=10 and It follows a Poisson distribution, representing the arrival rate of uplink / downlink packets to simulate the volume of traffic. Buffered packets follow a First-In-First-Out (FCFS) queuing rule.

[0016] 1.4) Performance metrics: For each user equipment k, consider the performance metric: the uplink and downlink bidirectional throughput of the k-th UE in the r-th time frame. and cumulative average uplink / downlink latency Where: T∈(0,N) T () indicates the total number of transmitted frames.

[0017] Step 2: An improved cross-slice protocol allows for adaptive retransmission parameter adjustment, selection of a new transmission process during retransmission, and enables data retransmission between different network slices. It also allows slices to switch between uplink and downlink services as needed, specifically including:

[0018] 2.1) Adaptive retransmission parameter design: Allows dynamic adjustment of user target rate decisions and overall resource allocation decisions That is, the location and quantity of resource blocks.

[0019] 2.2) Cross-slice retransmission design: The selectable range of retransmission resource blocks is expanded to... E n , where: E n The availability of free resources on slice n enables seamless resource sharing and coordination among slices for retransmission, enhancing resource allocation decision-making. The feasible domain.

[0020] 2.3) Flexible uplink / downlink handover design: Allows slices to switch between uplink and downlink services according to their needs, which, combined with designs 2.1 and 2.2, makes uplink / downlink handover decisions more flexible. It will no longer conflict with retransmission services.

[0021] like Figure 2 The diagram shows a comparison between the existing protocol and the cross-slice rewriting protocol of this embodiment: Design 2.1 is reflected in each column representing the granularity of 't', adaptive parameter selection in each time slot to optimize resource allocation decisions, and adaptive parameter optimization in each frame to optimize target rate decisions. Designs 2.2 and 2.3 are labeled in the figure. Therefore, the existing protocol effectively solves the problem of wasted resources in blank resource blocks, increasing resource utilization and throughput. Furthermore, the adaptive parameter design increases the feasible region, leading to a higher probability of successful transmission, a lower retransmission rate, and increased throughput.

[0022] Step 3: Construct a two-layer optimization problem based on the cross-slice retransmission protocol, specifically including:

[0023] 3.1) The upper-level problem is a throughput optimization problem that ensures user quality of service through uplink / downlink switching decisions, target rate decisions, and resource matching decisions: The constraints of the upper-level problem are This indicates the latency requirements for uplink and downlink transmission, where and This represents the latency threshold for the k-th UE. Furthermore, the lower-level problem also serves as a constraint on the upper-level problem.

[0024] In practical cases, it is necessary to... and Select parameters and improve the system's latency limits, setting them to... and Set the average latency thresholds for downlink and uplink respectively.

[0025] 3.2) The lower-level problem is in: The optimal solution set represents the lower-level problem, used to maximize the equivalent capacity of the k-th user equipment (UE). The constraints of the lower-level problem are as follows: This means that each resource block (RB) is allocated to only one UE in the resource allocation slot, ensuring resource isolation; This indicates that the initial transmission occurs within a slice n; This indicates that retransmissions can occur at any free resource block in any slice.

[0026] Step 4: Deploy the cross-slice retransmission protocol and perform resource scheduling, specifically including:

[0027] 4.1) Implement cross-slice protocol deployment and scheduling in the upper-layer problem by making uplink / downlink handover decisions based on the state in each time frame. and target rate decision The deployment of flexible uplink / downlink handover and adaptive retransmission parameter design across slice protocols is achieved. Specific decisions are made using reinforcement learning algorithms. Specifically, the base station selects its behavior based on its current state, outputs uplink and downlink probabilities using a softmax layer, and derives the uplink / downlink handover decision based on these probabilities. Decision-making using normalized target rate output from the tanh layer. To achieve dynamic decision-making, a reward function is used to ensure upper-layer latency constraints and throughput optimization, specifically including:

[0028] State space: vector in: It is the satisfaction level with the target rate. This represents the slice's idle resource rate. Data is normalized based on the different ranges of each parameter, so that the data range is [0,1].

[0029] Action space: consisting of the probabilities of uplink and downlink modes. Composition, represented as Actions include exploring actions using ∈-greedy as the decision output.

[0030] Reward function: in state s r Complete action a rLater received a reward Among them: [z] + =max{0,z},λ th , λ u and λ d These are the weighting coefficients for throughput and latency penalty, respectively. The upper-layer throughput optimization problem under cross-slice protocols, satisfying user guarantees, is solved through the design of the reward function, resulting in uplink and downlink decisions. and target rate decision Due to throughput η r (k) is also a normalized value, therefore the specific parameter value is λ. th =1,λ u =10 -2 , λ u =10 -2 .

[0031] 4.2) Implement cross-slice protocol deployment and scheduling in the lower-level problem. Given the upper-level decision parameters, resource allocation decisions are made by matching required resource blocks with available resource blocks: Adaptive retransmission parameters are deployed by mapping the number of resource blocks required by the user according to the target rate in each time frame, and cross-slice retransmission design is deployed by controlling the position and number of available resource blocks. The specific matching process uses a resource matching algorithm, taking the Hungarian algorithm as an example:

[0032] Transform the lower-level problem into a maximization problem. in: This represents the cost of allocating the (t,f)-th resource block (RB) to the k-th demand. By matching the required and available resource blocks to the user's needs, the optimal number and location of resource blocks are determined, leading to the resource allocation decision. The equivalent capacity C(r,m,k) can be obtained through resource allocation decisions.

[0033] 4.3) By comparing C(r,m,k) in the lower-level problem with C(r,m,k) in the upper-level problem... The retransmission feedback is used to implement the scheduling process, and a series of variables in the system are updated before the next round of scheduling is carried out.

[0034] like Figure 3 The diagram illustrates the performance of the cross-slice retransmission protocol of this invention at different signal-to-noise ratios (SNR) and the effect of each design step. When the SNR is 15 dB, this invention achieves throughput improvements of approximately 50.4%, 37.6%, and 23.2% compared to existing technologies. Furthermore, when the SNR exceeds 20 dB, the throughput under different protocols tends to be consistent under given parameters because retransmissions occur infrequently.

[0035] Figure 4(A) illustrates the impact of the number of users on overall throughput. As the number of users increases, overall throughput initially rises, peaking with 6 users, at which point user satisfaction is 99.67% under latency constraints. However, with further increases in the number of users, resource scarcity leads to a decrease in user satisfaction; with 15 users, user satisfaction drops to 96.68%, and latency significantly exceeds the constraints. The system achieves optimal performance with 150 available RBs per subframe, combined with 6 connected users. Furthermore, the system reaches its optimal performance when the RB service reaches 348 bits / t.

[0036] In summary, this invention can significantly improve total throughput by reducing the retransmission rate.

[0037] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.

Claims

1. A retransmission scheduling method based on cross-slice HARQ protocol, characterized in that, By constructing a scheduling model in a multi-slice network scenario on the RAN side, and considering the limitations of retransmission protocols in network slicing scenarios based on the scheduling model, a cross-slice retransmission protocol is designed. A two-layer optimization problem based on the cross-slice protocol and prioritizing user guarantees for throughput optimization is constructed. By deploying flexible uplink / downlink switching design at the upper layer, user-level target rate and slice-level uplink / downlink switching decisions are optimized. Then, based on the upper-layer optimization decisions, adaptive retransmission parameter design and cross-slice retransmission design are deployed. Real-time resource allocation is performed for each user according to channel conditions, resulting in optimized resource allocation decisions. Through protocol deployment and scheduling, the retransmission rate is reduced, and the total network throughput can be improved while meeting user guarantees, thus completing retransmission scheduling based on the cross-slice HARQ protocol.

2. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The aforementioned two-level optimization problem includes: The upper-level problem is a throughput optimization problem that ensures user quality of service through uplink / downlink switching decisions, target rate decisions, and resource matching decisions. The constraints of the upper-level problem are This indicates the latency requirements for uplink and downlink transmission, where and This represents the latency threshold for the k-th UE; the lower-level problem is also a constraint on the upper-level problem. The lower-level problem is in: The optimal solution set represents the lower-level problem, used to maximize the equivalent capacity of the k-th user equipment (UE). The constraints of the lower-level problem are as follows: This means that each resource block is allocated to only one UE in the resource allocation time slot, ensuring resource isolation; This indicates that the initial transmission occurs within a slice n; This indicates that retransmissions can occur at any free resource block in any slice.

3. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The aforementioned RAN-side network multi-slice scenario refers to: having N RS In an Orthogonal Frequency Division Multiplexing (OFDMA) system with radio access network (RAN) slices, in the nth slice, k∈U n Each User Equipment (UE) communicates with a Base Station (BS) via a downlink Incremental Redundancy (IR) protocol, defining N. S N T and N F These represent the number of subframes per time frame, the total number of time frames, and the number of subbands per network slice, respectively. The subframe and subband indices are denoted as t∈{1,…,N}. S N S +1,…,N S ×N T } and f∈{1,2,…,N F ×N RS Each RAN slice contains N S ×N F Each resource block (RBs) makes S n ={(t,f)|t∈{1,…,N} S N S +1,…,N S ×N T },f∈{n×N F +1,…,(n+1)×N F } represents the radio resource block belonging to the nth slice. At the BS end, at the beginning of each time frame, the network slice scheduler manages the uplink and downlink handover decisions for slice n based on the base station environment. Target transmission rate decision for user k The network slice scheduler then matches and allocates RBs to different UEs in each subframe based on the real-time channel environment and upper-layer decisions, thus obtaining resource allocation decisions. The transmission and retransmission process is abstracted as The resulting equivalent capacity and In the comparison between the two, when the BS receives a successful transmission signal based on the HARQ protocol, the data packet will be cleared from the relevant uplink and downlink buffers.

4. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The aforementioned construction of the scheduling model includes: a resource allocation process and an uplink / downlink handover and retransmission process, wherein: The resource allocation process refers to: the set of RBs allocated to the k-th UE in the r-th time frame as the resource allocation decision. Therefore, the equivalent capacity of the k-th UE in the r-th time frame can be calculated by the resource allocation decision. Where: m≥0 represents the index of the retransmission number. More precisely, {m=0} represents the first transmission, while {m>0} represents the m-th retransmission. This represents the signal-to-noise ratio (SNR) for user k communicating via slice n. Represents the normalized channel fading coefficient of RB relative to index (t,f); Uplink / downlink handover and retransmission process refers to: the downlink of the nth network slice in the rth time frame. and uplink In the full-duplex mode of transmission, during downlink transmission, after soft combining, when the accumulated channel capacity exceeds the desired transmission rate, the acknowledgment (ACK) signal will be valid. In uplink transmission, by comparison and Determine the ACK / NACK signal, specifically: indicator function Right now express in: Indicates the target rate. This indicates the current retransmission round. After determining the uplink and downlink modes, the uplink and downlink transmission buffer states of the k-th user in the r-th time frame can be updated as follows: and Where: {α u (k)} and {α d (k)} represents the uplink / downlink data packet arrival rate for user k, respectively.

5. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The aforementioned scheduling model, for each user equipment k, considers the performance metrics: the uplink and downlink bidirectional throughput of the k-th UE in the r-th time frame. and cumulative average uplink / downlink latency Where: T∈(0,N) T () indicates the total number of transmitted frames.

6. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The aforementioned adaptive retransmission parameter design refers to the fact that the cross-slice retransmission protocol allows for adaptive retransmission parameter adjustment, and allows for the selection of a new transmission process during retransmission.

7. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The cross-slice retransmission design refers to the cross-slice retransmission protocol allowing data retransmission between different network slices, promoting dynamic resource allocation across slices.

8. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The aforementioned flexible uplink / downlink switching design refers to the cross-slice retransmission protocol allowing slices to switch between uplink and downlink services according to their needs. Added uplink / downlink switching decision Flexibility.

9. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 1, characterized in that, The aforementioned protocol deployment and scheduling reduce the retransmission rate, specifically including: 1) Implement cross-slice protocol deployment and scheduling in the upper-layer problem. This involves flexible uplink / downlink handover and adaptive retransmission parameter design based on uplink / downlink handover decisions and target rate decisions made in each time frame according to the state. Specific decisions are obtained using reinforcement learning algorithms, including: State space: vector in: It is the satisfaction level with the target rate. Indicates the slice's idle resource rate; Action space: consisting of the probabilities of uplink and downlink modes. Composition, represented as / Actions include exploring actions using ∈-greedy as the decision output; Reward function: in state s r Complete action a r Later received a reward Among them: [z] + =max{0,z},λ th , λ u and λ d These are the weighting coefficients for throughput and latency penalty, respectively. The upper-layer throughput optimization problem under cross-slice protocols, satisfying user guarantees, is solved through the design of the reward function, resulting in uplink and downlink decisions. and target rate decision 4.2) Implement cross-slice protocol deployment and scheduling in the lower layer problem. Given the upper layer decision parameters, make resource allocation decisions by matching the required resource blocks and available resource blocks: deploy adaptive retransmission parameters by mapping the number of resource blocks required by the user according to the target rate in each time frame, and deploy cross-slice retransmission design by controlling the position and number of available resource blocks. The matching process is obtained using the Hungarian algorithm. 4.3) By comparing C(r,m,k) in the lower-level problem with C(r,m,k) in the upper-level problem... The retransmission feedback is used to implement the scheduling process, and a series of variables in the system are updated before the next round of scheduling is carried out.

10. The retransmission scheduling method based on cross-slice HARQ protocol according to claim 9, characterized in that, The Hungarian algorithm refers to: transforming the lower-level problem into... in: This represents the cost of allocating the (t,f) / -th resource block (RB) to the k-th demand. By matching the number and location of the resource blocks required by the user with the available resource blocks, the optimal resource block quantity and location are obtained, leading to a resource allocation decision. The equivalent capacity C(r,m,k) can be obtained through the resource allocation decision.

Images