Granted Patent B2

US 12,414,143 · App. 17/832,217 · Granted Sep 9, 2025

Inventors: Jagadishchandra Kodankiri Rai (Bangalore, IN); Godavarti Satya Venkata Uma Kishore (Bangalore, IN); Vimal Bastin Edwin Joseph (Bangalore, IN)

Assignee: SAMSUNG ELECTRONICS CO., LTD.

H04W72/542H04W16/14

Abstract

Claims (100)

1. A method for determining a dynamic spectrum sharing (DSS) policy across multiple radio access technologies (RATs), the method comprising:

determining, by a centralized node, a user equipment (UE) resource scaling factor for each of a plurality of UEs corresponding to at least one previous scheduling result based on at least one of: a UE learning rate of each UE, a previous radio resource requirements of each UE, or a UE spectral efficiency achieved by each UE for the at least one previous scheduling result;

determining, by the centralized node, radio resource requirements of each of the plurality of UEs based on the UE resource scaling factor of each UE;

determining, by the centralized node, cumulative radio resource requirements of the plurality of UEs based on the radio resource requirements of each UE; and

determining, by the centralized node, the DSS policy for scheduling radio resources based on the cumulative radio resource requirements of the plurality of UEs, and based on cell key performance indicator (KPI) metrics or a plurality of DSS scaling factors.

2. The method of claim 1 , wherein determining, by the centralized node, the UE resource scaling factor for each UE in the plurality of UEs comprises:

receiving a UE priority list depicting the plurality of UEs to be considered in determining the DSS policy from a radio access network (RAN);

enabling the RAN to allocate air resources to each of the plurality of UEs in the UE priority list based on UE KPI metrics and the cell KPI metrics monitored for at least one previous scheduling result of scheduling the radio resources;

creating a UE reward matrix for each UE by analyzing the UE KPI metrics and the UE priority list;

determining the UE spectral efficiency achieved by each UE for the at least one previous scheduling result from the UE reward matrix, wherein the UE spectral efficiency achieved by each UE is at least in terms of the performance parameters of the respective UE in the at least one previous scheduling result;

generating a UE learning queue by analyzing the UE KPI metrics and the UE spectral efficiency achieved for the at least one previous scheduling result;

determining the UE learning rate of each UE for the at least one previous scheduling result based on the generated UE learning queue; and

determining the UE resource scaling factor for each UE corresponding to the at least one previous scheduling result based on at least one of: the UE learning rate of each UE, the previous radio resource requirements of each UE, and the UE spectral efficiency achieved by each UE for the at least one previous scheduling result.

3. The method of claim 2 , wherein the UE priority list is determined based at least on an active UE list with non-zero data requirements and previously available performance parameters of each UE, wherein the previously available performance parameters of the UE include at least one of: the UE KPI metrics and policy parameters; and wherein the UE KPI metrics include at least one of: throughput, a reception/transmission (Rx/Tx) error, a scheduling frequency, link adaptation, power control metric, a multi-input multi-output, reference signals, data requirements, and mobility.

4. The method of claim 2 , further comprising:

updating the UE reward matrix for each UE based at least on spectral efficiency data

measured for each UE for each scheduling result,

wherein the UE reward matrix created for the UE includes:

a row index as a UE KPI index determined based on the UE KPI metrics monitored in the at least one previous scheduling result; and

a column index as a resource block size determined based on the previous resource requirements of the UE.

5. The method of claim 2 , wherein generating the UE learning queue includes:

analyzing the UE KPI metrics and the UE spectral efficiency achieved for the at least one previous scheduling result via an epsilon greedy method to generate the UE learning queue,

wherein the UE learning queue includes:

a row index as the UE KPI metrics; and

a column index as a spectral efficiency index determined based on the UE spectral efficiency achieved for the at least one previous scheduling result.

6. The method of claim 1 , wherein determining, by the centralized node, the radio resource requirements of each UE includes:

creating a shared radio resource allocation table among the plurality of UEs based on the UE KPI metrics, the current traffic requirements of each UE, and the individual UE resource scaling factor, wherein the shared allocation table indicates the total available radio resources per DSS cell, the radio resources allocated for each UE, and remaining available radio resources for allocation among UEs;

determining previous radio resource requirements of each UE from the created shared radio resource allocation table, wherein the previous radio resource requirements of the UE indicates radio resource requirements of the UE for the at least one previous scheduling result; and

determining the radio resource requirements of each UE by updating the previous radio resource requirements of each UE using the respective UE resource scaling factor determined based on the UE learning rate, wherein the radio resource requirements of each UE indicates the radio resources required by the UE for the multiple RATs, wherein a constructive heuristic based recursive method is used to determine the radio resource requirements of each UE at least by updating the previous radio resource requirements of each UE based on the respective UE resource scaling factor.

7. The method of claim 1 , wherein determining the DSS policy includes:

determining a DSS pattern that includes radio resource sharing information between the multiple RATs; and

determining a DSS pattern duration that includes a valid time duration applicable for the DSS policy selected in terms of continuous slots and/or radio frames for the given DSS pattern.

8. The method of claim 7 , wherein the radio resource sharing information comprises a slot-based (S-DSS) policy resource sharing and/or a frame-based (F-DSS) policy resource sharing, wherein the S-DSS policy resource sharing is in a frequency domain of each radio slot, wherein the F-DSS policy resource sharing is in a time domain of a radio frame.

9. The method of claim 7 , wherein determining the DSS pattern includes:

collecting the cell KPI metrics monitored for the at least one previous scheduling result and a current DSS policy, wherein the cell KPI metrics include at least one of: throughput, block error, and total radio link failure; and wherein the current DSS policy includes a current DSS pattern and a current DSS pattern duration;

creating a DSS reward matrix for the at least one previous scheduling result at least by analyzing the collected cell KPI metrics and the current DSS policy;

determining cell spectral efficiency achieved for the at least one previous scheduling result based on the created DSS reward matrix, wherein the cell spectral efficiency achieved for the at least one previous scheduling result indicates spectral efficiency achieved with respect to the cell KPI metrics for the at least one previous scheduling result;

generating a cell learning queue for the at least one previous scheduling result at least by analyzing the cell spectral efficiency achieved with respect to the cell KPI metrics for the at least one previous scheduling result;

determining a cell learning rate for the at least one previous scheduling result based on the generated cell learning queue;

creating a DSS pattern scaling factor table based on at least one of: a previously applied DSS pattern, the cell spectral efficiency achieved for the at least one previous scheduling result, a status of each of the plurality of UEs, and the cell learning rate;

determining a DSS pattern scaling factor for the at least one previous scheduling result based on the created DSS pattern scaling factor table, wherein the DSS pattern scaling factor is one of the plurality of DSS scaling factors;

using a constructive heuristic based recursive method to update the DSS pattern scaling factor determined for the at least one previous scheduling result based on the previously applied DSS pattern, the status of each of the plurality of UEs, the cell learning rate, and cell performance parameters;

storing the updated DSS pattern scaling factor in the DSS scaling factor table;

determining the DSS pattern by scaling the current DSS pattern using at least one of: the updated DSS pattern scaling factor, the cumulative radio resource requirements of the plurality of UEs, and a UE resource requirement ratio, wherein the UE resource requirement ratio indicates a ratio of the radio resources required by the UE for the multiple RATs; and

storing the determined DSS pattern in a DSS pattern table, wherein the DSS pattern table indicates various DSS patterns, wherein each DSS pattern is indicated by at least a pattern index.

10. The method of claim 9 , wherein:

the DSS reward matrix includes a row index as a cell KPI index determined based on the cell KPI metrics and a column index as a DSS policy index determined based on the current DSS policy;

the cell learning queue includes a row index as the cell KPI index and a column index as a spectral efficiency index calculated based on the cell spectral efficiency achieved for the at least one previous scheduling result; and

the DSS pattern scaling factor table includes a row index determined based on the previously applied DSS pattern and a column index determined based on the cell learning rate.

11. The method of claim 7 , wherein determining the DSS pattern duration includes:

creating a DSS pattern duration scaling factor table for the at least one previous scheduling result based on a previously applied DSS pattern duration, and the cell learning rate, wherein the DSS pattern duration scaling factor table includes a row index determined based on the previously applied DSS pattern duration, and a column index determined based on the cell learning rate;

determining a DSS pattern duration scaling factor from the DSS pattern duration scaling factor table for the at least one previous scheduling result, wherein the DSS pattern duration scaling factor is one of the plurality of DSS scaling factors;

updating the DSS pattern duration scaling factor determined for the at least one previous scheduling result based on at least one of: the cell learning rate, the cell spectral efficiency, the status of each of the plurality of UEs, and the scaled DSS pattern; and

determining the DSS pattern duration by scaling the current DSS pattern duration using the updated DSS pattern duration scaling factor.

12. The method of claim 1 , further comprising:

determining, by the centralized node, the DSS policy at least by following multiple RAT scheduling coordination and restriction, wherein the multiple RAT scheduling coordination and restriction includes at least one of: shared resource coordination, interference reduction, and communicating radio metrics between the multiple RATs with an absence of a direct interface.

13. The method of claim 1 , wherein:

the multiple RATs operating in a same frequency band share a same spectrum between each other with an absence of spectrum re-farming; and

the multiple RATs operate as coherent to standard co-existence mechanisms to reduce interference with each other, wherein the standard co-existence mechanisms include rate matching around reference signals to align subcarrier positioning.

14. A centralized node comprising:

a processing circuitry;

memory storing instructions, that, when executed by the processing circuitry, cause the centralized node to:

determine a user equipment (UE) resource scaling factor for each of a plurality of UEs corresponding to at least one previous scheduling result based on at least one of: a UE learning rate of each UE, a previous radio resource requirements of each UE, or a UE spectral efficiency achieved by each UE for the at least one previous scheduling result;

determine radio resource requirements of each of the plurality of UEs via at least the UE resource scaling factor of each UE;

determine cumulative radio resource requirements of the plurality of UEs based on the radio resource requirements of each UE; and

determine the DSS policy for scheduling radio resources based on

the cumulative radio resource requirements of the plurality of UEs, and based on cell key performance indicator (KPI) metrics or a plurality of DSS scaling factors.

15. The centralized node of claim 14 , wherein the processing circuitry is further configured to:

receive a UE priority list depicting the plurality of UEs to be considered in determining the DSS policy from a Radio Access Network (RAN);

enable the RAN to allocate air resources to each of the plurality of UEs in the UE priority list based on UE KPI metrics and the cell KPI metrics monitored for at least one previous scheduling result of scheduling the radio resources;

create a UE reward matrix for each UE by analyzing the UE KPI metrics and the UE priority list via a reinforcement learning (RL) based Q-learning module comprising circuitry;

determine the UE spectral efficiency achieved by each UE for the at least one previous scheduling result from the UE reward matrix, wherein the UE spectral efficiency achieved by each UE is at least in terms of the performance parameters of the respective UE in the at least one previous scheduling result;

generate a UE learning queue by analyzing the UE KPI metrics and the UE spectral efficiency achieved for the at least one previous scheduling result;

use the RL based-Q learning module to determine a UE learning rate of each UE for the at least one previous scheduling result based on the generated UE learning queue; and

determine the UE resource scaling factor for each UE corresponding to the at least one previous scheduling result based on at least one of: the UE learning rate of each UE, the previous radio resource requirements of each UE, and the UE spectral efficiency achieved by each UE for the at least one previous scheduling result.

16. The centralized node of claim 15 , wherein the processing circuitry is further configured to:

create a shared radio resource allocation table among the plurality of UEs based on the UE KPI metrics, the current traffic requirements of the UE, and the individual UE resource scaling factor, wherein the shared allocation table indicates at least the total available radio resources per DSS cell, the radio resources allocated for each UE, and remaining available radio resources for allocation among UEs;

determine the previous radio resource requirements of each UE from the created shared radio resource allocation table, wherein the previous radio resource requirements of the UE indicates radio resource requirements of the UE for the at least one previous scheduling result;

determine the radio resource requirements of each UE at least by updating the previous radio resource requirements of each UE based on the respective UE resource scaling factor determined based on the UE learning rate, wherein the radio resource requirements of each UE indicates the radio resources required by the UE for the multiple RATs.

17. The centralized node of claim 14 , wherein the processing circuitry is further configured to determine the DSS policy at least by:

determining a DSS pattern that includes radio resource sharing information between the multiple RATs; and

determining a DSS pattern duration that includes a valid time duration applicable for the DSS policy selected in terms of continuous slots or radio frames for the given DSS pattern.

18. The centralized node of claim 17 , wherein the processing circuitry is further configured to:

collect the cell KPI metrics monitored for the at least one previous scheduling result and a current DSS policy, wherein the cell KPI metrics include at least one of:

throughput, block error, and total radio link failure; and wherein the current DSS policy includes a current DSS pattern and a current DSS pattern duration;

create a DSS reward matrix for the at least one previous scheduling result by analyzing the collected cell KPI metrics and the current DSS policy via the RL module;

determine cell spectral efficiency achieved for the at least one previous scheduling result based on the created DSS reward matrix, wherein the cell spectral efficiency achieved for the at least one previous scheduling result indicates spectral efficiency achieved with respect to the cell KPI metrics for the at least one previous scheduling result;

generate a cell learning queue for the at least one previous scheduling result by analyzing the cell spectral efficiency achieved with respect to the cell KPI metrics for the at least one previous scheduling result;

determine a cell learning rate for the at least one previous scheduling result based on the generated cell learning queue;

create a DSS pattern scaling factor table based on at least one of: a previously applied DSS pattern, the cell spectral efficiency achieved for the at least one previous scheduling result, a status of each of the plurality of UEs, and the cell learning rate;

determine a DSS pattern scaling factor for the at least one previous scheduling result based on the created DSS pattern scaling factor table, wherein the DSS pattern scaling factor is one of the plurality of DSS scaling factors;

use a constructive heuristic based recursive technique to update the DSS pattern scaling factor determined for the at least one previous scheduling result based on the previously applied DSS pattern, the status of each of the plurality of UEs, the cell learning rate, and cell performance parameters;

store the updated DSS pattern scaling factor in the DSS scaling factor table;

determine the DSS pattern by scaling the current DSS pattern using at least one of, the updated DSS pattern scaling factor, the cumulative radio resource requirements of the plurality of UEs, and a UE resource requirement ratio, wherein the UE resource requirement ratio indicates a ratio of the radio resources required by the UE for the multiple RATs; and

store the determined DSS pattern in a DSS pattern table, wherein the DSS pattern table indicates various DSS patterns, wherein each DSS pattern is indicated by a pattern index.

19. The centralized node of claim 17 , wherein the processing circuitry is further configured to:

create a DSS pattern duration scaling factor table for the at least one previous scheduling result based on a previously applied DSS pattern duration, and the cell learning rate, wherein the DSS pattern duration scaling factor table includes a row index determined based on the previously applied DSS pattern duration, and a column index determined based on the cell learning rate;

determine a DSS pattern duration scaling factor from the DSS pattern duration scaling factor table for the at least one previous scheduling result, wherein the DSS pattern duration scaling factor is one of the plurality of DSS scaling factors;

update the DSS pattern duration scaling factor determined for the at least one previous scheduling result based on at least one of: the cell learning rate, the cell spectral efficiency, the status of each of the plurality of UEs, and the scaled DSS pattern; and

determine the DSS pattern duration at least by scaling the current DSS pattern duration using the updated DSS pattern duration scaling factor.

Assignments (1)

Priority Claims (2)

Continuity (2)

References Cited (25)