Wireless Network Uplink Scheduling Method Based on Adaptive Grouping and Reinforcement Learning

Abstract

The present invention relates to a method for uplink scheduling in a wireless network based on adaptive grouping and reinforcement learning, comprising: S1. The wireless access point obtains the clients corresponding to all resource units, calculates the grouping information according to the adaptive grouping algorithm, and sends it to all clients through a negotiation mechanism to form multiple groups; S2. The wireless access point obtains the BSR information of each group during the BSR request phase; S3. The wireless access point performs RU resource scheduling using the uplink scheduling algorithm according to the received BSR information to obtain the RU allocation result; S4. The wireless access point sends the RU allocation result to each group through a trigger frame during the data transmission phase, and each group performs data transmission on the corresponding resource unit, and determines whether the data of all groups have been transmitted. If not, it will return to step S2. Compared with the prior art, the present invention has the advantages of stronger robustness, ensuring the priority and fairness of the system, and improving the overall performance of the system, etc.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and in particular to an uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning. Background Art

[0002] The Wi-Fi 6 wireless network (IEEE 802.11ax) introduces the Orthogonal Frequency Division Multiple Access (OFDMA) technology, which divides the entire channel into several specific subcarrier sets, and these subcarrier sets are called Resource Units (RUs). Different clients (STAs) can simultaneously upload data frames on their respective corresponding RU resources to achieve parallel uplink data transmission. Since the number of STAs is large, to avoid resource waste caused by conflicts generated by channel contention, IEEE 802.11ax designs an uplink multi-user scheduling access mechanism based on trigger frames. During the scheduling access process, the AP uniformly schedules RU resources to different STAs, thus avoiding channel contention and improving the spectrum utilization rate.

[0003] However, during the uplink multi-user scheduling access process of IEEE 802.11ax, the buffer status reports (BSRs) required by the wireless access point (AP) to schedule RU resources come from each associated STA. After the AP sends a buffer status report poll (BSRP) to notify each STA, since RU resource allocation has not been performed yet, all STAs with non-empty buffers can only feedback their BSRs through the UORA mechanism. Therefore, these STAs need to transmit their BSRs by competing to access idle RUs based on the UORA mechanism. As the total number of STAs increases, the collision probability of the random access process will become higher and higher, the transmission rate of the BSR will be greatly reduced, and even the AP may not receive any BSRs from any STA at all, unable to perform the next RU resource scheduling, resulting in a decrease in system throughput.

[0004] To solve the above problems, IEEE 802.11ax introduces the TWT mechanism during the uplink multi-user scheduling access process, which can not only reduce the unnecessary energy consumption of STAs through the sleep and wake-up mechanism, but also group STAs according to the set TWT time period, so as to achieve grouped resource scheduling for dense users and improve the overall performance of the system. Summary of the Invention

[0005] The purpose of the present invention is to provide an uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning to overcome the defects of low system throughput and poor user experience in the uplink scheduling access process of 802.11ax in a dense user environment existing in the above-mentioned prior art.

[0006] The object of the present invention can be achieved by the following technical solutions:

[0007] A method for uplink scheduling of a wireless network based on adaptive grouping and reinforcement learning, specifically including the following steps:

[0008] S1. The wireless access point obtains all client stations (STAs) corresponding to resource units within the local area network, groups the clients according to the adaptive grouping algorithm, obtains the grouping information and sends it to all clients through the negotiation mechanism, and the clients form multiple groups according to the grouping information;

[0009] S2. The wireless access point obtains the BSR (Buffer Status Report) information of each group during the BSR request phase;

[0010] S3. The wireless access point performs RU resource scheduling according to the received BSR information by using the uplink scheduling (USRL) algorithm to obtain the RU allocation result;

[0011] S4. The wireless access point sends the RU allocation result to each group through a trigger frame during the data transmission phase, and each group performs data transmission on the corresponding resource unit according to the RU allocation result, and determines whether the data of all groups has been completed. If not, it will go back to step S2.

[0012] The adaptive grouping algorithm in the step S1 includes the following steps:

[0013] S11. The wireless access point calculates the optimal grouping value according to the number of clients and the backoff parameter;

[0014] S12. Group the clients associated with the wireless access point according to the optimal grouping value;

[0015] S13. Determine whether the number of clients in the last group is within the preset variable grouping range. If not, allocate the clients in the last group to other groups;

[0016] S14. Determine whether the number of clients in each group after reallocation is within the variable grouping range. If not, dissolve all groups and group them according to the minimum value of the variable grouping range;

[0017] S15. Determine whether the number of clients in the last group is not within the variable grouping range. If so, split the clients in the last group into other groups.

[0018] Further, the variable grouping range is specifically [N min , N max , where N min is the minimum value of client grouping, and Nmax The maximum value for client grouping.

[0019] The adaptive grouping algorithm groups STAs within a Basic Service Set (BSS) based on the TWT mechanism. Compared with the ungrouped scheme, it reduces the number of STAs accessing at the same time, improves the BSR transmission rate during the BSR request phase, and thus significantly increases the system throughput in a dense user environment. In addition, compared with the fixed grouping scheduling access scheme, the present invention can adaptively adjust the grouping strategy as the number of accessing STAs changes, has stronger robustness, and can ensure a high system throughput in the face of different numbers of STAs.

[0020] The calculation formula for the optimal grouping value is as follows:

[0021]

[0022] Where, N op is the optimal grouping value, η op is the maximized BSR transmission rate. The calculation formula for the BSR transmission rate is as follows:

[0023]

[0024] Where, η(p) is the BSR transmission rate, p is the collision probability. The conditions that must be satisfied when maximizing the BSR transmission rate are as follows:

[0025]

[0026] Where, n ∈ [2, ∞), p ∈ [0, 1].

[0027] Step S2 specifically includes the following steps:

[0028] S21. The wireless access point broadcasts a BSRP (Buffer Status Report Poll) frame to all clients, starts the scheduling access process for the corresponding grouping in the current service cycle, and the clients in the remaining groupings remain in the sleep state;

[0029] S22. The STAs in the current grouping use the OFDMA-based uplink random access mechanism to perform BSR transmission;

[0030] S23. After receiving the BSR information of the current grouping, the wireless access point broadcasts an M-BA (Multi-STA Block ACK) frame within the group.

[0031] Furthermore, in step S23, the wireless access point transmits the list of clients that have successfully sent BSR frames to the clients within the group by broadcasting the M-BA frame.

[0032] The USRL algorithm performs RU resource scheduling based on the BSR information of the current packet. Compared with the scheme of evenly allocating channel resources to each STA, it can more effectively ensure the priority sorting and fairness among different STAs, and ensure that services with higher Quality of Service (QoS) requirements can preferentially obtain larger RU resources for data upload. For example, important information such as fire alarms, medical emergencies, and traffic accidents can be uploaded in a timely and accurate manner. On this basis, STAs with lower priorities can still obtain RU resource allocation, ensuring that these users will not starve, thus guaranteeing the fairness of the entire system.

[0033] The uplink scheduling algorithm in step S3 includes the following steps:

[0034] S31. According to the BSR information sent by the current packet, calculate the value of the data to be uploaded by each client in the packet and the size of the resource units required through a value function;

[0035] S32. Encode the uplink data frame of the current client into a knapsack instance vector;

[0036] S33. Input the initial knapsack instance sequence composed of the knapsack instance vectors of all clients into the pointer network trained by the wireless access point;

[0037] S34. The knapsack instance sequence output by the decoder of the pointer network, and the clients included therein are the clients to which the wireless access point will allocate RU resources. Only those STAs selected by the decoder of the pointer network can perform data transmission on the corresponding RU.

[0038] Furthermore, the specific process in step S4 is that when the jth group finishes data transmission, all STAs in the group will enter the sleep state, and the STAs of the next packet will be awakened by the BSRP trigger frame of the AP, thus starting the scheduling access process of the next packet.

[0039] Furthermore, the formula of the value function is as follows:

[0040]

[0041] where, v i is the value of the data to be uploaded by the ith client, n is the total number of clients, d i is the data volume of the uplink data frame of the ith client, q i is the QoS value corresponding to the service type of the ith client, h i is the number of time windows (T win ) that the ith client has waited due to not obtaining RU allocation.

[0042] Furthermore, the QoS value corresponding to the client service type is an integer greater than 0, and the specific value range is {1, 2, 3, 4, 5}.

[0043] Furthermore, the backpack instance vector is specifically c i =(w i , v i ), where w i is the size of the resource unit currently required by the i-th client, and the initial backpack instance sequence is specifically

[0044] The data transmission stage of the wireless access point in step S4 specifically includes the following steps:

[0045] S41. The wireless access point broadcasts the RU allocation result to the clients in the current group through a trigger frame;

[0046] S42. After the client successfully receives the trigger frame, it transmits a data frame on the specified resource unit;

[0047] S43. The wireless access point broadcasts an M-BA frame for data confirmation according to the received data frame;

[0048] S44. After the data confirmation is completed, the clients in the current group enter the sleep state and wait for the next scheduling access, and the wireless access point wakes up the next group.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The present invention will first group STAs based on the TWT mechanism using an adaptive grouping algorithm, which can dynamically adjust the grouping strategy as the number of STAs changes, and solves the problem of low BSR transmission rate in a dense user environment; then, within the service cycle corresponding to each group, the STAs are woken up and their BSR information is collected; finally, RU resource scheduling is achieved by using the USRL algorithm, thereby ensuring the priority and fairness of the system, improving the overall performance of the system, and enhancing the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a schematic flowchart of the present invention;

[0052] Figure 2 is a simulation result of the throughput change over time of two typical STAs under the AG-USRL scheme scheduling in an embodiment of the present invention;

[0053] Figure 3 is a comparison chart of the BSR transmission rates between the AG-USRL scheme and the ungrouped scheme in an embodiment of the present invention;

[0054] Figure 4 This is the simulation comparison result of the throughput varying with time for the time-varying situation of the typical STA QoS values under the scheduling of four schemes in the embodiments of the present invention;

[0055] Figure 5 This is a schematic diagram of the system average throughput and the number of STAs under the scheduling of three schemes in the embodiments of the present invention;

[0056] Figure 6 This is a comparison graph of the relationship between the system average throughput and the number of STAs for the fixed packet scheduling scheme and the AG-USRL scheme in the embodiments of the present invention. Detailed implementation manners

[0057] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented on the premise of the technical solution of the present invention, and detailed implementation manners and specific operation processes are given, but the protection scope of the present invention is not limited to the following embodiments.

[0058] Embodiment

[0059] As Figure 1 shown, a method for scheduling the uplink of a wireless network based on adaptive grouping and reinforcement learning specifically includes the following steps:

[0060] S1. The wireless access point obtains all the client stations (STAs) corresponding to the resource units within the local area network, groups the client stations according to the adaptive grouping algorithm, obtains the grouping information and sends it to all the client stations through the negotiation mechanism, and the client stations form multiple groups according to the grouping information;

[0061] S2. The wireless access point obtains the BSR information of each group during the BSR request phase;

[0062] S3. The wireless access point performs RU resource scheduling according to the received BSR information by using the uplink scheduling algorithm to obtain the RU allocation result;

[0063] S4. The wireless access point sends the RU allocation result to each group through a trigger frame during the data transmission phase, and each group performs data transmission on the corresponding resource unit according to the RU allocation result, and determines whether the data of all groups has been completed. If not, it will go back to step S2.

[0064] The adaptive grouping algorithm in step S1 includes the following steps:

[0065] S11. The wireless access point calculates the optimal grouping value according to the number of client stations and the backoff parameter;

[0066] S12. Group the client stations associated with the wireless access point according to the optimal grouping value;

[0067] S13. Determine whether the number of the last group is within a preset variable grouping range. If not, allocate the clients in the last group to other groups.

[0068] S14. Determine whether the number of each group after reallocation is within the variable grouping range. If not, dissolve all groups and perform grouping according to the minimum value of the variable grouping range.

[0069] S15. Determine whether the number of the last group is not within the variable grouping range. If so, split the clients in the last group into other groups.

[0070] In this embodiment, step S11 further includes calculating the probability τ that a STA can successfully pass the backoff stage and send a data frame according to the backoff parameter in the UORA mechanism. The formula is as follows:

[0071]

[0072]

[0073] W i = 2 i (W0 + 1) - 1, i = 1, 2,..., m - 1

[0074] where W0 represents the minimum value of the OFDMA contention window (OCW); p is the collision probability; m is the maximum backoff level, K represents the number of RUs used for UORA; W i is the expression of the OCW between the maximum value and the minimum value.

[0075] The variable grouping range is specifically [N min , N max , where N min is the minimum value of the client grouping, and N max is the maximum value of the client grouping.

[0076] The adaptive grouping algorithm groups STAs within a basic service set (BSS) based on the TWT mechanism. Compared with the ungrouped scheme, it reduces the number of STAs accessing at the same time, improves the BSR transmission rate in the BSR request stage, and thus greatly improves the system throughput in a dense user environment. In addition, compared with the fixed grouping scheduling access scheme, the present invention can adaptively adjust the grouping strategy with the change of the number of access STAs, has stronger robustness, and can ensure a higher system throughput in the face of different numbers of STAs.

[0077] The calculation formula for the optimal grouping value is as follows:

[0078]

[0079] Among them, N op is the optimal grouping value, and η op is the maximized BSR transmission rate. The calculation formula for the BSR transmission rate is as follows:

[0080]

[0081] Among them, η(p) is the BSR transmission rate, p is the collision probability, and the conditions that must be satisfied when maximizing the BSR transmission rate are as follows:

[0082]

[0083] Among them, n ∈ [2, ∞), p ∈ [0, 1].

[0084] In this embodiment, during the UORA process, when two or more STAs select the same RU after ending the backoff stage, a conflict will occur. Based on this, the collision probability is calculated, and the formula is as follows:

[0085]

[0086] Among them, n is the total number of STAs.

[0087] Step S2 specifically includes the following steps:

[0088] S21. The wireless access point broadcasts a BSRP frame to all clients, starts the scheduling access process for the corresponding grouping in the current service cycle, and the clients in the remaining groupings remain in the sleep state;

[0089] S22. The STAs in the current grouping use the OFDMA-based uplink random access mechanism for BSR transmission;

[0090] S23. After the wireless access point receives the BSR information of the current grouping, it broadcasts an M-BA frame within the group.

[0091] In step S23, the wireless access point passes the list of clients that have successfully sent BSR frames to the clients within the group by broadcasting the M-BA frame.

[0092] The USRL algorithm performs RU resource scheduling based on the BSR information of the current packet. Compared with the scheme of evenly allocating channel resources to each STA, it can more effectively ensure the priority sorting and fairness among different STAs, and ensure that services with higher Quality of Service (QoS) requirements can obtain larger RU resources for data upload first. For example, it enables important information such as fire alarms, medical emergencies, and traffic accidents to be uploaded in a timely and accurate manner. On this basis, STAs with lower priorities can still obtain RU resource allocation, ensuring that these users will not starve, thus guaranteeing the fairness of the entire system.

[0093] The uplink scheduling algorithm in step S3 includes the following steps:

[0094] S31. According to the BSR information sent by the current packet, calculate the value of the data to be uploaded by each client in the packet and the required resource unit size through the value function;

[0095] S32. Encode the uplink data frame of the current client into a knapsack instance vector;

[0096] S33. Input the initial knapsack instance sequence composed of the knapsack instance vectors of all clients into the pointer network trained by the wireless access point;

[0097] S34. After the pointer network encoding is completed, select the knapsack instances in the input sequence through the attention mechanism of the decoder, and finally output the knapsack instance sequence. The clients included in it are the clients to which the wireless access point will allocate RU resources. Only those STAs selected by the pointer network decoder can perform data transmission on the corresponding RU.

[0098] In this embodiment, in step S31, the BSR information successfully fed back by the M j th STA in the jth group and their transmission delays are used together to calculate the value of the data frame to be uploaded by each STA in the group. The higher the value, the higher the priority of being allocated RU resources in the data transmission stage.

[0099] The specific process in step S4 is that after the jth group completes data transmission, all STAs in the group will enter the sleep state, and the STAs in the next group will be awakened by the BSRP trigger frame of the AP, thus starting the scheduling access process of the next group.

[0100] The formula of the value function is as follows:

[0101]

[0102] Among them, v i is the value of the data to be uploaded by the ith client, n is the total number of clients, di is the data volume of the uplink data frame for the i-th client, q i is the QoS value corresponding to the service type of the i-th client, h i is the time window (T win ) number that the i-th client waits due to not getting RU allocation.

[0103] The QoS value corresponding to the client service type is an integer greater than 0, and the specific value range is {1, 2, 3, 4, 5}.

[0104] The backpack instance vector is specifically c i =(w i , v i ), where w i is the size of the resource unit currently required by the i-th client. The initial backpack instance sequence is specifically

[0105] The data transmission stage of the wireless access point in step S4 specifically includes the following steps:

[0106] S41. The wireless access point broadcasts the RU allocation result to the clients in the current group through a trigger frame;

[0107] S42. After the client successfully receives the trigger frame, it performs data frame transmission on the specified resource unit;

[0108] S43. The wireless access point broadcasts an M-BA frame for data confirmation according to the received data frame;

[0109] S44. After the data confirmation is completed, the clients in the current group enter the sleep state and wait for the next scheduling access, and the wireless access point wakes up the next group.

[0110] Specifically in implementation, the parameter settings are as follows: the channel bandwidth is 20 MHz, the minimum value W0 of OCW is 7, the number K of available RUs for UORA is 9, and the efficiency factor α is 0.95; and the TGax NLOS indoor channel model is adopted, and the LDPC channel coding method is used. For the pointer network model required for RU resource scheduling, the Actor-Critic algorithm is used for parameter training and is deployed to the AP side in advance. The performance improvement of this embodiment is analyzed by comparing this embodiment with the ungrouped polling algorithm, the ungrouped PRA algorithm, and the average allocation scheme based on adaptive grouping.

[0111] Such as Figure 2The figure shows the result display of the simulation experiment of this embodiment when the number of STAs in the BSS is 100. In the figure, two groups (Group A and Group C) are randomly selected for comparative analysis. In Group A and Group C, 2 representative STAs are respectively selected to display the simulation results of the throughput changing with time. The buffer data volumes of the 2 STAs in Group A are the same, and their parameter settings are as follows: the QoS value of STA1 is 1 and the MCS is 7; the QoS value of STA8 is 4 and the MCS is 7. The buffer data volumes of the 2 STAs in Group C are the same, and their parameter settings are as follows: STA 45 has a QoS value of 1 and an MCS of 4; STA 50 has a QoS value of 4 and an MCS of 4. It can be found from the figure that the average throughput of the STA with a smaller QoS value in both groups is higher, while the average throughput of the other STA with a larger QoS value within the group is lower. At the same time, STA1 in Group A finishes transmitting 3.4 s earlier than STA8, and STA 45 in Group C finishes transmitting 2.7 s earlier than STA 50 . This shows that the data transmission requests of the STAs with higher priorities in both groups can be preferentially satisfied, enabling them to finish data transmission earlier than the STAs with lower priorities within their own groups. In addition, the STAs with lower priorities in both groups do not show the phenomenon of "starvation", and their throughputs increase after the STAs with higher priorities finish transmitting, which fully demonstrates that the AG-USRL scheme of the present invention has good fairness guarantee and high spectrum utilization rate.

[0112] As Figure 3 shown is the comparison chart of the BSR transmission rate between the AG-USRL scheme and the ungrouped scheme. The BSR transmission rate is only related to the UORA mechanism and the number of STAs in the BSR request stage. It can be seen from the figure that as the number of STAs continuously increases, the proportion of STAs that can successfully transmit the BSR in the BSR request stage of the AG-USRL scheme after adaptive grouping basically remains at about 38%, while that of the ungrouped scheme gradually decreases. When the number of STAs is 100, the BSR transmission rate is only 1 / 15 of that of the AG-USRL scheme, and when the number of STAs reaches 180, the BSR transmission rate is even close to 0. That is to say, as the wireless local area network becomes denser, the number of STAs that can successfully access the AP in the BSR request stage of the AG-USRL scheme is much larger than that of the ungrouped scheme, ensuring the executability of the data transmission stage; relatively, the ungrouped scheme may not be able to perform RU resource scheduling and uplink data transmission due to being unable to receive the BSR information of any STA.

[0113] As Figure 4Shown is the result of adjusting the QoS value of a typical STA in the time domain and conducting an uplink scheduling access simulation experiment using the above four schemes. During the simulation, the QoS value of the typical STA service type remained 1 for the first 2 seconds, 4 from 2 seconds to 4 seconds, and then remained 2 until the end of the simulation. The parameters of all other STAs remained unchanged. It can be found from the figure that the AG-USRL scheme has the ability to adapt to changes in service types and can adjust the size of RU resource allocation in real time according to the service priority of the STA. It can also be found that the average allocation scheme based on adaptive grouping does not have the ability to guarantee priority and cannot make scheduling adjustments according to changes in service types, and the ungrouped scheme obviously does not have the corresponding function. This shows that compared with the other three schemes, the AG-USRL scheme has an additional service type tracking ability, can make corresponding adjustments in a timely manner, and has better RU resource scheduling performance.

[0114] As Figure 5 shown, to compare the overall performance of different scheduling access schemes, the number of STAs in the BSS was successively set to 20, 60, 100, 140, and 180 for the same uplink scheduling access simulation experiment. In the simulation experiment, the MCS value of all STAs was set to 7 to prevent the MCS from interfering with the simulation results. It can be found from the figure that when the number of STAs is 20, that is, when there are fewer users, the system average throughput of the ungrouped scheme and the AG-USRL scheme is about 64 Mbps, with little difference, and both can provide good uplink data transmission services for associated STAs. However, as the number of STAs in the BSS increases, the system average throughput of the two ungrouped schemes gradually decreases. When there are 100 STAs, it is only about 20 Mbps, which is 1 / 3 of the AG-USRL scheme. When the number of STAs reaches 180, the system average throughput of the ungrouped scheme is almost 0, and the entire network is severely affected by channel conflicts and it is difficult to complete scheduling access. As the number of STAs increases, the AG-USRL scheme can still keep the system's average throughput at about 64 Mbps and provide normal scheduling access services. In summary, the AG-USRL scheme reduces channel conflicts during BSR transmission through grouping, so it can still provide good uplink scheduling access services in a dense user environment.

[0115] As Figure 6As shown, the AG-USRL scheme is compared with the fixed packet scheduling access scheme to verify whether the present embodiment can dynamically adjust the number of packets and the packet size according to the change in the number of STAs in the BSS to ensure system performance. The number of groups in the fixed packet scheme is: 2, 4, 6, and the number of STAs in the simulation comparison experiment is: 20, 60, 100, 140, 180 in sequence. It can be found from the figure that when the number of STAs is 20, the average throughput of both the AG-USRL scheme and the fixed 2-group scheme is about 64 Mbps. However, as the number of STAs increases, the system average throughput of the fixed 2-group scheme becomes lower and lower. When there are 180 STAs, it cannot even reach 10 Mbps, which is far lower than the AG-USRL scheme. For the other two fixed packet schemes (4 groups and 6 groups), when the number of STAs is 20, they bring average throughputs of 48 Mbps and 39 Mbps respectively. When the number of STAs is too large, the system average throughput is lower than the AG-USRL scheme, and it can only approach or reach the AG-USRL scheme when there are 60 STAs. The above shows that the AG-USRL scheme provides a more flexible packet method for uplink transmission by using the adaptive packet algorithm, and its packet strategy can be adjusted according to the change in the number of STAs, reducing the impact on the system and improving the system throughput.

[0116] Thus, the 802.11ax uplink scheduling access scheme based on adaptive packet and reinforcement learning isolates the data transmission of different groups and provides a flexible packet method by using the adaptive packet algorithm based on the TWT mechanism, thereby improving the overall BSR transmission rate of the system and increasing the system throughput in a dense user environment compared with the ungrouped scheduling access scheme. In addition, compared with the fixed packet scheduling scheme, it can better adapt to the scenario where the number of STAs changes and has stronger robustness. And the USRL algorithm is used for RU resource scheduling in the data transmission stage, meeting the requirements of the system for priority and fairness guarantee and improving the overall performance of 802.11ax uplink scheduling access.

[0117] In addition, it should be noted that for the specific embodiments described in this specification, the names taken may be different. The above content described in this specification is only an example of the structure of the present invention. Any equivalent changes or simple changes made according to the structure, features, and principles conceived by the present invention are included in the protection scope of the present invention. Those skilled in the art of the present invention can make various modifications or supplements to the described specific examples or adopt similar methods, as long as they do not deviate from the structure of the present invention or exceed the scope defined by this claim book, they should belong to the protection scope of the present invention.

Claims

1. A method for uplink scheduling in a wireless network based on adaptive grouping and reinforcement learning, characterized in that, Specifically, it includes the following steps: S1. The wireless access point obtains the clients corresponding to all resource units in the local area network, groups the clients according to the adaptive grouping algorithm, obtains the grouping information and sends it to all clients through the negotiation mechanism, and the clients form multiple groups according to the grouping information; S2. The wireless access point obtains the BSR information of each group in the BSR request stage; S3. The wireless access point performs RU resource scheduling using the uplink scheduling algorithm according to the received BSR information to obtain the RU allocation result; the uplink scheduling algorithm in S3 includes the following steps: S31. According to the BSR information sent by the current group, calculate the value of the data to be uploaded by each client in the group and the size of the required resource units through the value function; The value calculation formula is: where v i is the value of the data to be uploaded by the i-th client, n is the total number of clients, d i is the data volume of the uplink data frame of the i-th client, q i is the QoS value corresponding to the service type of the i-th client, h i is the number of time windows (T win ) that the i-th client waits due to not receiving RU allocation; S32. Encode the uplink data frame of the current client into a knapsack instance vector; S33. Input the initial knapsack instance sequence composed of the knapsack instance vectors of all clients into the pointer network trained by the wireless access point; S34. The knapsack instance sequence output by the decoder of the pointer network, and the clients included therein are the clients to which the wireless access point will allocate RU resources; S4. In the data transmission stage, the wireless access point sends the RU allocation result to each group through the trigger frame, and each group performs data transmission on the corresponding resource units according to the RU allocation result, and determines whether the data of all groups has been completed. If not, it will go back to step S2.

2. The uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning according to claim 1, characterized in that, The adaptive grouping algorithm in step S1 includes the following steps: S11. The wireless access point calculates the optimal grouping value according to the number of clients and the backoff parameter; S12. Group the clients associated with the wireless access point according to the optimal grouping value; S13. Determine whether the number of the last group is within the preset variable grouping range. If not, allocate the clients in the last group to other groups; S14. Determine whether the number of each group after reallocation is within the variable grouping range. If not, dissolve all groups and group them according to the minimum value of the variable grouping range; S15. Determine whether the number of the last group is not within the variable grouping range. If so, split the clients in the last group into other groups.

3. The uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning according to claim 2, wherein The specific variable grouping range is [N min , N max , where N min is the minimum value of the client grouping, and N max is the maximum value of the client grouping.

4. An uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning according to claim 1, characterized in that, Step S2 specifically includes the following steps: S21. The wireless access point broadcasts a BSRP frame to all clients, starts the scheduling access process for the groups corresponding to the current service cycle, and the clients in the remaining groups remain in the sleep state; S22. The STA of the current group uses the OFDMA-based uplink random access mechanism to perform BSR transmission; S23. After the wireless access point receives the BSR information of the current group, it broadcasts an M-BA frame within the group.

5. The uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning according to claim 4, characterized in that In step S23, the wireless access point passes the list of clients that have successfully sent the BSR frame to the clients within the group by broadcasting the M-BA frame.

6. The uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning according to claim 1, characterized in that The QoS value corresponding to the client service type is an integer greater than 0, and the specific value range is {1, 2, 3, 4, 5}.

7. A method for uplink scheduling of a wireless network based on adaptive grouping and reinforcement learning according to claim 1, characterized in that The backpack instance vector is specifically c i =(w i , v i ), where w i is the size of the resource unit currently required by the i-th client, and the initial backpack instance sequence is specifically 8. The uplink scheduling method for a wireless network based on adaptive grouping and reinforcement learning according to claim 1, characterized in that The data transmission stage of the wireless access point in step S4 specifically includes the following steps: S41. The wireless access point broadcasts the RU allocation result to the clients of the current group through a trigger frame; S42. After the client successfully receives the trigger frame, it transmits a data frame on the specified resource unit; S43. The wireless access point broadcasts an M-BA frame for data confirmation based on the received data frame; S44. After the data confirmation is completed, the clients of the current group enter the sleep state and wait for the next scheduled access, and the wireless access point wakes up the next group.

Images