A method for identifying neighboring cells based on limited information.

Abstract

By reducing electricity consumption and carbon dioxide emissions, we aim to mitigate the negative environmental impact of our operations while simultaneously improving our end-user revenue. [Solution] A method for resolving neighboring cell ambiguity includes the steps of receiving a user equipment (UE) measurement report and identifying which combination of neighboring cells would have resulted in the set of physical cell IDs (PCIs) in the UE measurement report by identifying the cells in each combination that have a common area of ​​geometrical radio coverage overlap with the UE's serving cell.

Description

[Technical Field]

[0001] [Related applications]

[0001] This application claims the interests of U.S. Patent Provisional Application No. 63 / 725,866, entitled “Construction of Tripartite Graphs and Algorithms for Offloading Ues and Putting Serving Cells in Sleep Mode,” U.S. Patent Provisional Application No. 63 / 725,857, entitled “Determining Identity of Neigbor Cells Based on Limited Information,” and U.S. Patent Provisional Application No. 63 / 725,851, entitled “Detecting Radio Coverage Overlap of Two or More Cells,” all of which are incorporated herein by reference in their entirety.

[0002]

[0002] Embodiments of the present disclosure relate to wireless communications, and more specifically, embodiments disclosed herein relate to cellular communications in which cells are identified to enter a low-power consumption state and user equipment (UEs) being served by those cells are moved to other cells. [Background technology]

[0003]

[0003] Today, mobile network operators are considering power consumption and carbon dioxide consumption because the energy used to power mobile networks contributes to greenhouse gas emissions and climate change. In addition to the environmental impact, excessive power consumption can also result in increased operating costs for network operators. Therefore, mobile network operators can use reducing their own power consumption and carbon dioxide emissions to mitigate the adverse impact of their operations on the environment and, on the other hand, also improve their ultimate revenue.

[0004]

[0004] One of the options people are considering to reduce power usage is to turn off base stations during low-traffic periods. However, if a mobile network operator puts a cell to sleep, the mobile network operator may create coverage holes, which will dissatisfy many of its customers. One solution to address coverage holes is to attempt to increase the coverage of those cells by tilting the antennas of those cells upward to cover the coverage holes. There are several problems associated with this approach. First, even after tilting, coverage holes may still exist. Second, tilting the antennas of nearby cells may cause more interference in other nearby cells (that are not put to sleep). Additionally, handovers between cells may be affected, which can result in an increased number of failed handovers due to tilting again. All of these problems may become even more pronounced if a mobile network operator wants to put several cells to sleep simultaneously to save significant power.

SUMMARY OF THE INVENTION

[0005]

[0005] Methods and apparatuses for constructing a three - part graph are disclosed. In some embodiments, a method includes receiving a measurement report wirelessly transmitted from a user equipment (UE), and creating a three - part graph including first, second, and third sets of nodes in response to the measurement report, wherein the first set of nodes represents a plurality of cells, the third set of nodes represents a plurality of UEs receiving coverage from the plurality of cells, each node in the second set of nodes represents one or a group of one or more of the plurality of cells, in the three - part graph, one or more cells in the group are connected to one or more UEs, and cell coverage for one or more UEs can be obtained from at least one cell in the group if one or more cells in the group are shown as on in the three - part graph, and identifying cell coverage for a plurality of UEs based on the three - part graph using a set of active cells from the plurality of cells.

[0006]

[0006] Other aspects and advantages of the embodiments will become apparent from the following detailed description, which, taken in conjunction with the accompanying drawings that illustrate, by way of example, the principles of the described embodiments.

[0007]

[0007] The described embodiments and their advantages can be best understood by reference to the following description taken in conjunction with the accompanying drawings. These drawings do not limit in any way any changes in shape and details that can be made by those skilled in the art to the described embodiments without departing from the spirit and scope of the described embodiments.

Brief Description of the Drawings

[0008] [Figure 1A] FIG. is a flowchart of some embodiments of a process for constructing and using a three - part graph. [Figure 1B]This is a flowchart of several other embodiments of the process for constructing and using a three-part graph. [Figure 2A] This figure shows an example of a coverage problem. [Figure 2B] This figure shows an example of a coverage problem. [Figure 3] This is a flowchart of several embodiments of the process for creating a three-part graph and using that graph to identify which cells should be in a low-power consumption state (e.g., sleep state). [Figure 4A] This figure shows an example of a three-part graph and an exemplary reduction based on PCI from a measurement report (MR). [Figure 4B] This figure shows an example of a three-part graph and an exemplary reduction based on PCI from a measurement report (MR). [Figure 4C] This figure shows an example of a three-part graph and an exemplary reduction based on PCI from a measurement report (MR). [Figure 4D] This figure shows an example of a three-part graph and an exemplary reduction based on PCI from a measurement report (MR). [Figure 4E] This figure shows an example of a three-part graph and an exemplary reduction based on PCI from a measurement report (MR). [Figure 4F] This figure shows an example of a three-part graph and an exemplary reduction based on PCI from a measurement report (MR). [Figure 4G] This figure shows an example of a three-part graph and an exemplary reduction based on PCI from a measurement report (MR). [Figure 5] This figure shows an example of a tripartite graph. [Figure 6A] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6B] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6C] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6D] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6E] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6F] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6G] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6H] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6I] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 6J] This figure shows an example of a coverage algorithm applied on a three-part graph. [Figure 7A] This figure shows an example of a tripartite graph. [Figure 7B] This figure shows an example of an extended tripartite graph. [Figure 7C] These figures show the original three-part graph and the extended three-part graph, respectively. [Figure 7D] These figures show the original three-part graph and the extended three-part graph, respectively. [Figure 8A] This figure shows an example of how to process a tripartite graph. [Figure 8B] This figure shows an example of how to process a tripartite graph. [Figure 8C] This figure shows an example of how to process a tripartite graph. [Figure 8D] This figure shows an example of how to process a tripartite graph. [Figure 8E] This figure shows an example of how to process a tripartite graph. [Figure 8F] This figure shows an example of how to process a tripartite graph. [Figure 8G] This figure shows an example of how to process a tripartite graph. [Figure 9A] This figure shows separate embodiments in which a set of likely neighboring ECIs is identified based on a sector list. [Figure 9B]This figure shows separate embodiments in which a set of likely neighboring ECIs is identified based on a sector list. [Figure 10] This diagram shows an example of coverage overlap. [Figure 11A] This figure shows omni-coverage modeling using convex polygons. [Figure 11B] This figure shows omni-coverage modeling using convex polygons. [Figure 11C] This figure shows omni-coverage modeling using convex polygons. [Figure 12A] This figure shows an example of using a linear constraint to represent a circular sector. [Figure 12B] This figure shows an example of using a linear constraint to represent a circular sector. [Figure 13A] This figure shows an example of how one non-convex region can be represented as two convex regions. [Figure 13B] This figure shows an example of how one non-convex region can be represented as two convex regions. [Figure 14] This diagram shows an example of how sector coverage is represented. [Figure 15] This diagram illustrates an exemplary network where serving cells overlap in coverage due to two cells having the same PCI. [Figure 16A] This figure shows a first example of a cell-by-cell embodiment. [Figure 16B] This figure shows a first example of a cell-by-cell embodiment. [Figure 17A] This figure shows a second example of a cell-by-cell embodiment in which neighboring cells are identified based on bandwidth limitations. [Figure 17B] This figure shows a second example of a cell-by-cell embodiment in which neighboring cells are identified based on bandwidth limitations. [Figure 17C] This figure shows a second example of a cell-by-cell embodiment in which neighboring cells are identified based on bandwidth limitations. [Figure 18A] This figure shows examples of embodiments for each UE. [Figure 18B]This figure shows examples of embodiments for each UE. [Figure 19A] This figure shows another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 19B] This figure shows another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 19C] This figure shows another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 20A] This figure shows another example of an embodiment where neighboring cells are identified per UE. [Figure 20B] This figure shows another example of an embodiment where neighboring cells are identified per UE. [Figure 20C] This figure shows another example of an embodiment where neighboring cells are identified per UE. [Figure 21A] This figure shows yet another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 21B] This figure shows yet another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 21C] This figure shows yet another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 22A] This figure shows another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 22B] This figure shows another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 22C] This figure shows another example of an embodiment where neighboring cells are identified per UE based on bandwidth limitations. [Figure 23] This is a data flow diagram of several embodiments of data preprocessing operations. [Figure 24] This figure shows several embodiments of generating a raw graph dictionary for use when creating a three-part graph, and using that information to support reducing the power of serving cells and offloading the UEs of those reduced-power serving cells. [Figure 25] This is a data flow diagram of several embodiments of the process for creating a three-part graph. [Figure 26] This figure shows several embodiments of a set of algorithmic objects used to create a tripartite graph. [Figure 27] This is a data flow diagram of several embodiments of the process for identifying which cells to turn off while offloading any of the UEs that are serviced by those cells. [Figure 28A] This figure illustrates an example of using the metrics described above to determine which cells will be turned on and which can be put into a low-power state, while still maintaining coverage for the UE. [Figure 28B] This figure illustrates an example of using the metrics described above to determine which cells will be turned on and which can be put into a low-power state, while still maintaining coverage for the UE. [Figure 29] This figure illustrates an example of using the metrics described above to determine which cells will be turned on and which can be put into a low-power state, while still maintaining coverage for the UE. [Figure 30] These are block diagrams of several embodiments of a base station. [Figure 31] This is a data flow diagram of several embodiments of the process for identifying cell coverage for UE. [Figure 32] This is a data flow diagram of several embodiments of the process for performing neighbor cell ambiguity resolution. [Figure 33] This is a data flow diagram of several other embodiments of the process for identifying cell coverage for UE. [Modes for carrying out the invention]

[0009]

[0043] The following description provides many details to give a more thorough explanation of the present invention. However, it will be apparent to those skilled in the art that the present invention can be carried out without these specific details. In other cases, well-known structures and devices are shown in block diagrams rather than in detail, in order to avoid complicating the present invention.

[0010]

[0044] For the purposes of this specification, it should be noted that a cell in a mobile network represents a specific geographical area covered by a base station. Each cell is covered by a base station, and the coverage area of ​​each cell may fluctuate and change over time. These changes can be due to several reasons, such as base station antenna parameters, including transmit power, tilt of the electronic antenna, and corrections to the azimuth transmission direction, or changes in the cell environment (e.g., a new building). For user equipment (UEs), a cell is divided into a serving cell and neighboring cells. A serving cell is the cell currently providing services to the UE, in which case the UE is currently communicating with and through the base station of this cell (e.g., for calls and data transmissions). A neighboring cell is a cell with overlapping coverage with the serving cell and may become a new serving cell. This change may occur when the UE moves to a new location. This change may also occur for load balancing purposes. For example, suppose a serving cell is congested with traffic, while neighboring cells have very little traffic. In such a case, the Random Access Network (RAN) controller can be aware of this situation and signal the UEs in the congested cell to provide measurement reports in the bandwidth where the low-load cells operate (or, in some cases, on several bandwidths with potentially low-load cells). Those UEs return the reports, and based on these reports, one or more UEs with a sufficiently strong signal to the low-load cells are moved to that cell. Note that these UEs may have an even lower signal-to-noise ratio (SINR) to the new cell, but because that cell is low-load, they can be given a much larger portion of the resources (e.g., bandwidth) for communication. In addition, the UEs remaining in the loaded cells also gain more resources because those heavily loaded cells are offloading some of their load.

[0011]

[0045] Today, there are several green initiatives aimed at reducing power consumption. More recently, some of these initiatives have focused on mobile networks. With this in mind, mobile networks are evaluated over various periods (e.g., the following periods) to determine whether it is possible to reduce, or potentially minimize, active network resources for the purpose of reducing or minimizing power consumption, provided that the traffic demand across the entire Random Access Network (RAN) is met. To that end, in some embodiments, power savings are achieved by putting cells into sleep mode (e.g., turning off cells and putting them into a low-power consumption state) and offloading any UEs that were being served by those sleeping cells to other nearby active neighboring cells, thereby reducing power. In some embodiments, at different times of the day, a network operator could put certain cells in the network to sleep while still ensuring a high quality of experience (QoE) for UEs in the network.

[0012]

[0046] The technology disclosed herein addresses the problem of "generalized set coverage," which must be solved to identify which cells can be put into sleep mode. This would allow all UEs in those cells to be offloaded to nearby active cells (and thus none of those UEs would fail due to the applied energy-saving mode). In other words, the problem of reducing power by turning off cells and offloading the UEs served by those cells to other nearby active cells involves solving the coverage problem. In some embodiments, the coverage problem involves finding a few cells in the network (e.g., a minimum number of cells) that can provide full coverage for the UEs in the network. The technology disclosed herein leverages measurement reports from UEs in each serving cell to identify neighboring cells to which each reporting UE can be offloaded. In some embodiments, the solution to the coverage problem can be applied to the RAN during periods of low traffic.

[0013]

[0047] For complete coverage, identification of neighboring cells can be performed based on measurement reports (MRs) sent by UEs to their serving cells. If each UE specifies accurate identification information for its neighboring cells in its measurement report, the problem can be transformed into a "set covering" problem, where the goal would be to select the minimum number of cells that cover all UEs. These standard set covering problems are well known and occur in a wide range of applications. These problems can be represented by bipartite graphs. Heuristic algorithms exist for solving these problems on bipartite graphs.

[0014]

[0048] However, while MR includes information that helps identify neighboring cells (e.g., PCI, bandwidth information, received RSRP, etc.), UE measurement reports only report the physical cell ID (PCI) and not the precise identification information of each cell. PCI is assigned by the RAN operator and is assumed to be "locally unique" on its respective frequency band. The existence of only partial information regarding the identification of neighboring cells makes standard "set covering" methods insufficient.

[0015]

[0049] While identifying cells capable of providing full coverage, in some embodiments, assuming the most recent MR, the network finds the minimum number of cells capable of providing full coverage while still meeting the current traffic demand. The techniques disclosed herein include a method for mapping problems associated with partial neighbor cell information to a three-part graph and algorithms for propagating on these graphs that are capable of ensuring coverage.

[0016]

[0050] In some embodiments, determining what constitutes complete coverage can also be based on sector list data that provides information about the deployed access points. In some embodiments, the sector list data includes the longitude and latitude information of the access points, frequency band information, and antenna parameters (e.g., height, tilt, orientation, beamwidth, etc.). In some embodiments, the sector list data also includes RAN parameters (e.g., PCI, cell ID (ECI), sector reference, etc.). In some embodiments, one goal is to provide coverage for all UEs in the MR. For each UE in the MR, at least one neighboring cell or the serving cell itself should remain active. By collecting MRs over a sufficiently long period (e.g., one week), the network can ensure with a high probability that all UEs in the network can be covered by covering the MR. In fact, 3GPP provides provisions for periodic UE reporting across the entire bandwidth used by network operators to provide network access to their UEs. These MRs effectively provide a uniform sampling of the offloading options available to UEs across the entire coverage space. If MRs have been collected over a sufficiently long period and across all typical types of traffic (e.g., morning, evening, weekday, weekend), finding a coverage set for all MRs guarantees a very high probability that the UE will have coverage with a limited set of coverage cells active.

[0017] A three-part graph that enables offloading UEs to nearby cells and putting serving cells into a low-power consumption state.

[0051] As discussed above, the coverage problem can be addressed by obtaining information from the UE's MR and using that information to construct a ternary graph. In some embodiments, the ternary graph includes multiple nodes, which represent cells, cell groups, and UEs. Once the ternary graph is constructed, a coverage algorithm is run on that ternary graph to identify which cells can be put into a low-power state (e.g., sleep state) and which cells can be active in order to provide coverage for UEs in the mobile network.

[0018]

[0052] Figure 1A is a flowchart of several embodiments of the process for constructing and using a three-part graph. This process can be performed by processing logic that may include hardware (e.g., circuits, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0019]

[0053] Referring to Figure 1A, this process begins when the processing logic receives a measurement report (MR) transmitted wirelessly from the UE (processing block 101). In some embodiments, each measurement report is used to identify what may be referred to herein as a “UE instance.” Each UE instance corresponds to an offloading option for the UE at a given location in the network. In some embodiments, such a UE instance is described as a subset of lines from a measurement report communicated by a single UE within a sufficiently short period of time (e.g., within 1 to 3 seconds). Such a UE instance enumerates the serving cell, neighboring PCI (and optionally the E-UTRA absolute radio frequency channel number (EARFCN)), as well as reference signal received power measurement (RSRP) and reference signal received quality measurement (RSRQ) for the serving cell and neighboring PCI.

[0020]

[0054] In response to the MR, the processing logic creates a three-part graph (processing block 102). In some embodiments, the three-part graph includes first, second, and third sets of nodes. The first set of nodes represents cells in the network. The third set of nodes represents UEs that receive wireless coverage from the cells represented by the first nodes. Each node in the second set of nodes represents a group of one or more cells from the cells represented by the first set of nodes, and is connected between one or more cells in that group and the UEs that those cells are capable of providing wireless coverage.

[0021]

[0055] After creating a three-part graph, the processing logic identifies cell coverage for UEs using a set of active cells from the cells represented by a first set of nodes in the three-part graph (process 103). That is, the processing logic identifies a set of cells in the three-part graph that can cover the UEs (provide wireless coverage to those UEs).

[0022]

[0056] In some embodiments, this process also includes generating preferred states for the MR's serving cells (processing block 104). In some embodiments, the preferred states may include active or sleep states for each of the cells servicing the UE in the network. An active state indicates a cell that will remain active when the network attempts to reduce power consumption by reducing the power of other cells in the network, while a sleep state indicates a cell that will be put to sleep or into a low-power state when the network attempts to reduce power consumption.

[0023]

[0057] In some embodiments, once cell coverage for a UE is identified (from a ternary graph of a set of active cells represented by a first set of nodes in the ternary graph), the process also includes offloading the set of UEs from their connection to a serving cell that will be powered down to one or more other nearby active cells (processing block 105). In some embodiments, nearby active cells are cells identified as overlapping in radio coverage with one or more serving cells of the offloaded UEs.

[0024]

[0058] This process may also include processing logic putting the serving cell into a low-power state after offloading the UEs it was servicing to one or more other nearby active cells (processing block 106). In some embodiments, before powering off a given cell X, the cell has the ability to ask each of its UEs to provide a measurement report of type “event A4”. This is a series of measurements per UE for each of the frequency bands specified by cell X. In each measurement, in some embodiments, the UE specifies the band being measured, the neighboring PCIs that the UE sees, and the intensity / interference levels of those neighboring PCIs (e.g., RSRP, RSRQ measurements). In some embodiments, each such measurement also specifies the RSRP and RSRQ levels of the serving cell. Based on this information, and assuming that the coverage algorithm specifies the covering set of cells that will remain on, at least one of these cells is reported by the respective UE. Cell X then proceeds to move the UE to one of these cells in the coverage set. The same procedure is currently followed regarding offloading UEs from busy cell X, for example, when cell X is congested and some of its neighboring cells have lighter coverage. Cell X collects event A4 reports and offloads some of its UEs to nearby (less congested) cells where these UEs have sufficiently strong RSRP / RSRQ values.

[0025]

[0059] In some embodiments, determining cell coverage to a UE using a set of active cells based on a tripartite graph involves running a coverage algorithm on the tripartite graph. In some embodiments, the coverage algorithm includes a sleep selection algorithm for determining which cells should be in a low-power state. In some embodiments, the coverage algorithm includes an incremental algorithm in which candidate cell group nodes that are in the off state are changed to the on state one at a time (creating a new state on the tripartite graph), one or more metrics are calculated for the candidate cell group nodes when they are on, the impact of the candidate cell group nodes on turning on the cell groups associated with the candidate cell group nodes is determined, one or more metrics are compared for all candidate nodes, and one or more candidate cell group nodes are turned on based on the comparison of the metrics. The algorithm may also include performing another iteration of these operations on the tripartite graph using candidate nodes that have good metrics when they are on. These operations can be summarized as follows: In iteration n: 1. Given: The state of the tripartite graph (identified from the set of on / off cells) 2. For each cell group node p that is in the OFF state: i. Turn on candidate node p. ii. Obtain a new (candidate) state for the tripartite graph. iii. Calculate a metric for candidate node p by comparing the new state with the old state. 3. Turn on the candidate node p that has the highest metric. 4. Update the status of the three-part graph. 5. Proceed to iteration n+1.

[0026]

[0060] In some embodiments, the metric includes a criterion indicating whether additional UE nodes are covered if the cell group associated with the candidate node is turned on. In some other embodiments, the metric includes a criterion indicating whether additional UE nodes are covered if the cell group associated with the candidate node is turned on, divided by the extra cells that are turned on. In some embodiments, turning on one or more candidate nodes based on a comparison of metrics includes turning on one candidate node that has a better metric compared to the metrics of the candidate nodes. Therefore, in some embodiments, the metric can be one of the following:

number

[0027]

[0061] The algorithm described above involves turning on the candidate node p with the highest metric, but in some other embodiments, there are variations in which multiple nodes with sufficiently high metrics are turned on. For example, in various embodiments, the following are all valid options: A) The N nodes with the highest metric are turned on for a period of time specified by the network (NW) engineer. B) All nodes with metrics within a predefined percentage of the node with the highest metric are turned on, and consequently, the number of nodes turned on in this case potentially changes with each iteration. C) Step B is performed first to identify M nodes within a predefined percentage of the highest metric. If M > N, the best N nodes are retained; otherwise, all M nodes are retained. D) The remaining cells are specified by the NW engineer to be at most a% of the remaining (still inactive) cells. After performing option C, the cells with the highest metrics are retained, and the same number of additional selected cells (from performing C) are retained without having a set to turn on more than a% of the remaining cells.

[0028]

[0062] After applying the coverage algorithm to the tripartite graph, the deployment of cells to the on or off state is based on the state of the node in the tripartite graph. A cell node is covered if the corresponding cell is in the on state. A cell group node is in the on state if all of its neighboring cells are covered and any set of such cells in any of its likely groups is in the on state and capable of serving the UE. A UE node is covered if at least one of its neighboring cell groups is in the on state. In other words, the calculations on the tripartite graph are adjusted to resolve the immediate coverage issues, including uncertainty in which neighboring cells broadcast the PCI values ​​reported by each UE, with the cell group node being the middle node in the tripartite graph, and each such node being connected to one or more cells (left node). A cell group (the middle node) is active only if and only if all cells connected to it by edges are also active, which means that a UE reporting a PCI value associated with a given cell group can only be "covered" by that PCI if all likely nearby cells associated with that PCI value are turned on. These likely cells are all cells broadcasting this PCI value that could potentially have led to that PCI measurement by a UE reporting that PCI value in a given serving cell.

[0029]

[0063] While incremental algorithms can be used, other greedy or non-greedy algorithms can also be applied to ternary graphs. These may include faster algorithms. Other optimizations can be used, such as using smaller graphs and / or parallelizing the process and / or running the process on a graphics processing unit (GPU).

[0030]

[0064] In some embodiments, the faster algorithm used is a multiple candidate selection algorithm. In such an algorithm, turning on one or more candidate nodes based on a comparison of metrics includes turning on a set of cell group candidates, the set of cell groups including cell groups that have a superior metric (e.g., maximum impact) compared to the metrics of other cell nodes, and one or more additional criteria. These additional criteria may include cell groups that have a metric that is in a predetermined mathematical relationship with the superior metric, a predetermined percentage of cells in the cell group, and a predetermined number of cells in the cell group. These operations can be summarized as follows: In iteration n: Given: The state of the tripartite chart (identified from the set of on / off cells) For each cell group node p among the off cell group nodes, • Turn on candidate node p. • Obtain a new (candidate) state for the tripartite graph. • Calculate a "metric" for candidate node p by comparing the new state with the old state. A subset of candidate nodes is turned on based on the metrics of those candidate nodes and potentially additional criteria (such as those mentioned above). Update the status of the tripartite graph. Proceed to iteration n+1. This process ends when all UEs (the nodes on the right) are covered. The resulting set of covered cells (the nodes on the left) corresponds to the set of coverage cells, i.e., the set of cells that, if always kept on, would guarantee coverage for all UEs.

[0031]

[0065] Figures 28A-28B and 29 illustrate an example of using the metrics described above to determine which cells will be turned on and which can be put into a low-power state, while still maintaining coverage for UEs. In the examples shown in Figures 28A-28B and 29, the algorithm begins by initializing the set of coverage cells as an empty set. Turning on the middle node A in Figures 28A-28B requires turning on the one cell on the left (cell A), which covers the three UEs on the right (UE1, 2, and 3). Therefore, the metric for turning on the middle node A is equal to (3 covered UEs) / (1 cell) = 3. Turning on the middle nodes A and C instead requires turning on the two cells on the left (cells A and C), which covers the four UEs on the right (UE1, 2, 3, and 4). Therefore, the metric for turning on the middle nodes A and C is (4 covered UEs) / (2 cells) = 2. Therefore, depending on the selected metric, it would be preferable to turn on the middle node A rather than the middle nodes A and C, because it provides a higher metric. In some embodiments, it is preferable to turn on the middle node A rather than the middle nodes A and C, because turning on A covers 3 UEs per activated cell, whereas in the case of nodes A and C, 2 UEs are covered per activated cell. In some embodiments, the algorithm applies the same metric calculation to each middle node in Figures 28A and 28B. Investigation reveals that the algorithm would choose to activate the middle node Z. Indeed, Z has the highest metric value of 4 among all the middle nodes in Figures 28A and 28B. Turning on the middle node Z requires turning on one cell on the left (Z), which covers the four UEs on the right (UE3, 4, 5, and 6).Therefore, turning on Z will result in the largest number of UEs covered per activated cell among all options. When you select the middle node Z, the algorithm adds the associated cell on the left (i.e., cell Z) to the cell's coverage set and proceeds to the next step to cover more UEs.

[0032]

[0066] Figure 29 shows the next step in the algorithm, namely after node Z is turned on and UE1, 2, 3, and 4 are covered. As shown in Figure 29, only two UEs, UE1 and 2, remain uncovered. The middle node A requires turning on one cell on the left (cell A), covering both of the remaining uncovered UEs on the right (UE1 and 2), resulting in a metric of 2, equal to (two covered UEs) / (one turned-on cell). Nodes A,C and A,B,C result in lower metrics; turning on any of these nodes covers the two remaining UEs, but doing so requires turning on several cells. Nodes B,C, on the other hand, do not cover any of the remaining uncovered cells, and therefore have a metric of 0. As a result, the algorithm selects the middle node A and adds the corresponding cell on the left (i.e., cell A) to the coverage set. Since there are no more UEs (nodes on the right) that remain uncovered after applying the second algorithmic step (shown above in Figure 29), the algorithm terminates by outputting the cell coverage set {A,Z}.

[0033]

[0067] Figure 1B is a flowchart of several embodiments of the process for constructing and using a three-part graph. This process can be performed by processing logic that may include hardware (e.g., circuits, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0034]

[0068] Referring to Figure 1B, this process is similar to the process in Figure 1A, with the exception that, after receiving the measurement report transmitted wirelessly from the UE, the process includes identifying a list of cells which are options for use in providing coverage for each UE in the measurement report, and converting each neighbor PCI reported in the measurement report in the list of cells into one or more sets of neighbor cells (processing block 101A). In some embodiments, a three-part graph is then created using the one or more sets of neighbor cells resulting from the conversion, and the rest of the process continues as in Figure 1A.

[0035]

[0069] Figures 2A and 2B illustrate the coverage problem. Referring to Figure 2A, serving cell S provides coverage for UE201, 202, and 203. Cell X is potentially capable of providing wireless coverage to UE202 and UE203, cell Z is potentially capable of providing coverage to UE202, and cell Y is potentially capable of providing wireless coverage to UE201. Cells X and Y have a PCI equal to 10, while cell Z has a PCI equal to 15. As stated above, PCI does not precisely identify cells, and therefore there is ambiguity regarding which cells are neighbors of serving cell S.

[0036]

[0070] As shown in the table in Figure 2A, neighboring PCI for UE201-203 is a cell with a PCI equal to 10, and includes a cell group containing either X or Y. UE202 also has a neighboring PCI equal to 15, and includes a cell group containing only cell Z. Neighboring PCI is provided in the actual measurement report (MR) sent to serving cell S by UE201-203. Using this information, the three-part graph 210 includes nodes 221-224 associated with cells X, Y, Z, and S, respectively. The three-part graph 210 also includes nodes 241-243 representing UE203, 201, and 202, respectively. Cell groups are identified using nodes 231-233. Node 231 represents cell group X or Y, node 232 represents cell group Z, and node 233 represents the cell group containing serving cell S.

[0037]

[0071] Using this three-part graph 201, a coverage algorithm is applied to determine which set of cells can cover UE201-203, thereby enabling the possibility of putting serving cell S into a low-power state while still maintaining wireless coverage for all UE201-203. In this example, it is possible to determine whether UE201-203 can receive wireless coverage from cells X, Y, and Z, thereby enabling serving cell S to be put into a low-power state.

[0038]

[0072] While the three-part graph 210 in Figure 2A represents a limited set of nodes, in reality, three-part graphs can be considerably larger when considering all cells, all cell groups, and all UEs in a mobile network. For example, as shown in Figure 2B, some three-part graphs 211 may contain 45,000 cells, 60,000 cell groups, and 8.8 million UEs. This makes determining coverage even more difficult.

[0039]

[0073] Because the actual number of cells, UEs, and potential cell groups is very large, in some embodiments, the processing logic identifies neighboring cells that could potentially be offloading options for each UE in the measurement report and translates each reported neighbor PCI into a list of one or more likely neighboring cell IDs (ECIs) for cells to offload. In some embodiments, the associated neighboring EARFCN (indicating the neighboring frequency band) is also reported by the UE along with each neighboring PCI. It is possible to broadcast a given (neighboring) PCI value and use additional frequency band information to restrict the selection of likely neighbors to only those cells operating on the frequency band specified by the neighboring EARFCN value. For the purposes of this specification, it should be noted that cell groups can be identified as (PCI,{ECI}) pairs, and these terms can be used interchangeably. The processing logic then constructs a ternary graph using the list of likely neighboring ECIs for each UE in the MR.

[0040]

[0074] Figure 3 is a flowchart of several embodiments of a process for creating a three-part graph and using that graph to identify which cells should be placed in a low-power consumption state (e.g., sleep state). This process can be performed by processing logic that may include hardware (e.g., circuitry, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0041]

[0075] Referring to Figure 3, this process begins with the processing logic receiving an input file (processing block 301). In some embodiments, the input file includes measurement reports from the UE, Mobility Management Entity (MME) files (including pre-ECI1 to ECI1, etc.), a sector list (including ECI and PCI in sectors), and input from a network (NW) engineer. In some embodiments, the input file includes only a subset of these items. In some embodiments, the sector list provides deployment information for each cell, including ECI, PCI, EARFCN, frequency band, and antenna deployment parameters, such as the latitude and longitude of the access point, the azimuth and elevation angles of the antenna, beamwidth, and transmit power. These parameters can be used to convert neighboring PCIs to likely neighboring cells. In some other embodiments, a subset of these parameters is used to convert neighboring PCIs to likely neighboring cells. In some embodiments, the MME file can be used to capture handovers from source cells to destination cells over the course of a day and further improve the list of likely neighbors, and if a list of likely UEs served by some cell S (and corresponding to some neighbor PCI value P) has two cells X and Y, and if there are many handovers between cell S and cell Z (which also has a PCI value equal to P), then it can be assumed (and the processing logic can act on that assumption) that cell Z is also a neighbor of S, and therefore the list of likely neighbor cells having PCI value P can be extended to {X, Y, Z}. In some embodiments, these inputs include specifying parameters related to the coverage sector overlap identification algorithm (see, for example, Figures 11A to 14 and Figures 16B to 25), which cells and bandwidths to keep on and which to keep off (Figure 26), which graphs (standard / extended) and algorithms to use (see, for example, Figures 6A to 8G), and so on.Using the input file, the processing logic limits the scope of the process to cells within the coverage area of ​​the measurement report (processing block 302). In this method, the module identifies all cells that are potentially offloading options for each UE in the MR.

[0042]

[0076] Next, the processing logic performs PCI-to-ECI disambiguation (processing block 303). As part of PCI-to-ECI disambiguation, the processing logic translates each reporting neighbor PCI into a list of one or more likely ECIs. These ECIs identify neighbor cells to potentially offload UEs from the serving cell. After performing PCI-to-ECI disambiguation, the processing logic creates a tripartite graph using the list of likely neighbor ECIs for each UE in the report (processing block 304). After creating the tripartite graph, the processing logic executes a sleep selection algorithm or other algorithm (e.g., other greedy algorithms) on the tripartite graph (processing block 305). In some embodiments, the sleep selection algorithm is a coverage algorithm executed on the tripartite graph, aimed at providing coverage for all UEs in the report with a minimum number of active cells.

[0043]

[0077] After executing the coverage algorithm, the processing logic outputs a list of recommended states for all MR serving cells (processing block 306). In some embodiments, the recommended states include sleep state and active state to indicate, respectively, that a serving cell will transition to a sleep state (low power consumption state) or an active state, in which case the serving cell continues to serve UEs with coverage. In some embodiments, the process also includes offloading a set of UEs that have connections to serving cells that will be put into a sleep state to one or more other nearby active cells (processing block 307), and after offloading the UEs to one or more other nearby active cells, putting the serving cells that are indicated to be put into a low power consumption state (e.g., sleep state) into a low power consumption state (e.g., sleep state) (processing block 308).

[0044]

[0078] Given the number of cells and UEs in a typical wireless communication system, the tripartite graph can be very large. Figure 4A shows the initial tripartite graph with 45,000 cells, 50,000 cell groups, and 8.8 million UEs. There are several ways to reduce the size of the tripartite graph to make it easier to apply coverage algorithms to the tripartite graph with the aim of identifying the cells to which UEs should be offloaded, while also reducing the power of the serving cells of those UEs.

[0045]

[0079] In some embodiments, the three-part graph can be reduced by integrating UEs that have the same offload option indicated by PCI. For example, Figure 4B shows a table that provides ordered PCIs from the MR. Referring to Figure 4B, there are five UE indices shown with the same serving cell and a list of PCIs contained in the MR for each of the UEs identified by the UE index. For example, the first UE identified by UE index 1 in row 1 has a serving cell identified by the number 35592707 and has three neighboring cells with PCIs 13, 459, and 479.

[0046]

[0080] To reduce the initial three-part graph to three-part graph 2 in Figure 4A, UEs with the same neighboring PCI can be combined into a single group. In this case, the UEs identified by indices 1 and 5 in Figure 4B have the same neighboring PCI. Therefore, these two UEs can be grouped together. Figure 4C shows a table illustrating the grouped together. As a result shown in three-part graph 2, the number of cells is still equal to 45,000, but the number of cell groups is equal to 60,000, and the number of UEs has been reduced to 800,000 (down from 8.8 million).

[0047]

[0081] In some embodiments, by taking into account that the order of neighboring PCIs in the MR is not important, the three-part graph 2 can be further reduced to three-part graph 3 in Figure 4A. For example, in Figure 4D, the original MR provided an ordered set of PCIs. Referring to Figure 4D, the UE identified by UE index 1 has neighboring cells identified by PCIs 13, 459, and 479. Similarly, the UE identified by UE index 3 has neighboring PCIs identified by PCI numbers 459, 13, and 479. Finally, the UE identified by UE index 5 has neighboring cells identified using PCI numbers 13, 459, and 479. Therefore, UEs 1, 3, and 5 all have the same neighboring cells based on PCIs in the table (although their neighboring cells are not in the same order). In this case, the graph can be reduced by ignoring the order of PCIs and combining those UEs into a single group having the same neighboring cells.

[0048]

[0082] In some embodiments, these neighboring PCIs are in decreasing order of value, but this is not mandatory. One advantage of having neighboring PCIs in decreasing (or increasing) order for each UE instance is that this ordering allows software to more quickly check whether two UE instances are equivalent in terms of likely neighbors. For example, consider two PCI tuples, (100,300,250) and (250,100,300). Visually, these tuples correspond to the same set of neighboring PCI values. Sorting all tuples in increasing [or decreasing] order results in the same tuple (100,250,300) [or (300,250,100)] for both sets, and therefore those tuples can be immediately declared as equivalent.

[0049]

[0083] An example of the resulting table is shown in Figure 4E, where the UE group identified by UE group index 1 has three rows and neighboring cells identified by PCI 479, 459, and 13. As shown in the resulting tripartite graph 3 in Figure 4A, the number of cells is still 45,000, but the number of cell groups has been reduced to 40,000 (down from 60,000 in tripartite graph 2), and the total number of UEs has been reduced to 500,000 (down from 800,000 in tripartite graph 2).

[0050]

[0084] The three-part graph can be further reduced to a basic level, as shown in three-part graph 4 in Figure 4A. Figure 4F shows a table showing a sample section of the compressed MR that produces three-part graph 3. As shown, the UE group identified by UE group index 1 and the UE group identified by UE group index 3 have some of the same neighboring cells, as shown by PCI. In this case, UE group index 1 has neighboring cells with PCI equal to 479, 203, and 13, while UE group index 3 has neighboring cells identified by PCI 479 and 13. Since 479, 203, and 13 include 479 and 13, the UE group index having 479, 203, and 13 can be mapped to the group having neighboring cells identified by PCI numbers 479 and 13. In other words, the two groups can be merged. Figure 4G shows the reduced chart with the merged groups.

[0051] Coverage algorithm

[0085] While applying the coverage algorithm on a tripartite graph, the state of the tripartite graph becomes such that any given on / off deployment of a cell corresponds to the state of a node in the tripartite graph. In some embodiments, to obtain a state, a cell node is covered if the corresponding cell is on. A cell group node is on if all of its neighboring cells are covered. This assumes that any such set of any likely group is on and capable of servicing the UE. A UE node is covered if at least one of its neighboring cell groups is on. An exemplary tripartite graph is shown in Figure 5. Referring to Figure 5, cells X and Z are on, cell group Z is on, and UE501 is covered because its cell group Z is on.

[0052]

[0086] There are several coverage algorithms that can be applied on a tripartite graph to obtain state information that can be used for cell on / off deployment. For example, there is a family of greedy algorithms that can be applied on a tripartite graph. In some embodiments, one of the greedy algorithms, referred herein as an incremental algorithm, can be applied on a tripartite graph. In some other embodiments, a multiple candidate selection algorithm can be used. In some further embodiments, other algorithms can be used on a tripartite graph to obtain state information that can be used for deployment decisions.

[0053]

[0087] Figures 6A to 6J show examples of coverage algorithms applied on a three-part graph. These coverage algorithms are executed by processing logic that may include hardware (e.g., circuits, dedicated logic, memory), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0054]

[0088] Referring to Figure 6A, the three-part graph 600 includes cells A-C and X-Z along with UE1-6. UE1-3 were initially serviced by cell A, while UE4-6 were initially serviced by cell X. The three-part graph 600 is generated from MRs from two serving cells A and Z. Cell groups are represented by nodes and include cell groups A;A,C;A,B,C; and B,C. Cell groups also include cell groups X,Y,Z;X,Y; and cell group Z. For the left graph edges and the nodes on those graph edges, to turn on a cell group, all cells connected to the cell group by the edge are turned on. For the right graph edges, any edge from a cell group to a UE covers a UE.

[0055]

[0089] In some embodiments, all cells in a three-part graph are initially off (e.g., in sleep mode), and the application of the coverage algorithm begins by turning on a single cell group. In this case, cell groups A and C, represented by node 601, are turned on. Figure 6B shows where a single cell group node is turned on in a three-part graph.

[0056]

[0090] Regarding turning on cell groups A and C, the first step is to identify which cells need to be turned on in order to turn on cell groups A and C. The processing logic identifies which cells to turn on in order to turn on cell groups A and C. In this case, to turn on cell groups A and C, the processing logic turns on cells A and C. Figure 6C shows that cell groups A and C, represented by node 601, are turned on, and cell A, represented by node 602, and cell C, represented by node 603, are turned on. After turning on cells A and C, the processing logic identifies which other cell groups to turn on, due to the fact that turning on cells A and C may turn on additional cell groups. If cells A and C are on, the processing logic turns on cell groups A and A,C (i.e., not just cells A and C). In other words, after turning on cell groups A and C, and thereby turning on cells A and C, the processing logic also turns on cell group A. Note that since cell B is still off, cell groups A, B, C and B, C are not on. Figure 6D shows cell group A and cell groups A, C being turned on.

[0057]

[0091] After the additional cell groups are turned on, the processing logic determines which UEs are covered by the turned-on cell groups. Since cell groups A and A,C are turned on, these cell groups cover UE1, 2, 3, and 4. This is shown in Figure 6E.

[0058]

[0092] After identifying which UEs are covered, the processing logic removes the covered UEs and their edges from the tripartite graph because they are no longer needed. In this example, UE1, 2, 3, and 4 are already covered, and therefore their nodes are no longer needed and can be removed from the graph. Similarly, all edges of UE1, 2, 3, and 4 are no longer needed and can be removed. Figure 6F shows the result after the covered UEs and their edges have been removed from the tripartite graph.

[0059]

[0093] As described above, after removing the covered UEs and their edges, the processing logic removes the on cell group nodes and their edges as they are no longer needed. Since cell groups A and A,C are no longer needed, the edges from cell groups A and A,C are no longer needed and can be removed. Figure 6G shows a three-part graph without on cell group nodes and edges.

[0060]

[0094] After deleting the on cell group nodes and edges, the processing logic deletes them as they are no longer needed. In this example, cells A and C are no longer needed, nor are the edges of cells A and C. Figure 6H shows the deletion of on cell nodes and edges from the tripartite graph. After deleting the on cell nodes and edges, the processing logic merges cell groups connected to the same cell. Since cells A and C are on, cell groups A,B,C and B,C become cell group B. Therefore, the processing logic merges these two cell group nodes and renames the resulting node to B. Cell groups A,B,C and B,C are both connected to a UE to cell B and are therefore merged. Figure 6I shows the merging of cell groups connected to the same cell.

[0061]

[0095] After merging cell groups connected to the same cell, the processing logic merges cell group nodes connected to the same cell. These cell groups can be renamed to reflect the cells that will turn them on. Note that in some embodiments, merging is not applied. When cell nodes B and C are turned on, nodes A, B, and C are also turned on because cells A and C are already on. Note that merging produces a simpler graph but requires computation. Figure 6J shows the resulting three-part graph after merging cell group nodes connected to the same cell. As shown in Figure 6J, the edges for cell groups X, Y, Z and X, Y have been removed from their connections to UE5 and UE6 (cell group X, Y has already had its edges removed).

[0062]

[0096] As a result, this process demonstrates that cells A, B, and C can cover all of UE1-6, and that cells X, Y, and Z can be put into a low-power consumption state (e.g., sleep state).

[0063]

[0097] In some other embodiments, an extended ternary graph is used. In an extended ternary type of graph, a UE can be covered by a cell group only if and only if the UE has an edge to a cell group. For example, in this case, the original ternary graph, shown as ternary graph 701 in Figure 7A, can be extended by processing logic to generate an extended graph 702. In some embodiments, the original ternary graph directly represents the reported PCI, and accordingly, each connection between the UE and the middle node represents either 1a) coverage provided by one of the neighboring PCIs reported by the UE (in which case the middle node represents a group of likely neighboring cells that have that PCI), or 1b) coverage provided by the UE's serving cell (in which case the middle node represents a group containing only serving cells), but the original graph does not necessarily represent all coverage options for each UE. For example, a UE can be provided with coverage by its serving cell X, but it can also be provided with coverage by any other middle node representing a group of cells containing X. In fact, such a middle node would correspond to the neighbor PCI reported to X in the surrounding cells (and in this case, the reported neighbor PCI would coincide with the PCI of X). In contrast, in some embodiments, the extended graph arises from further post-processing and represents all coverage options per UE (in terms of which group of cells can cover each UE). In such cases, a simpler coverage algorithm 703 can be applied to identify the power saving allocation and state 704, in contrast to applying a typical coverage algorithm 705 to the original three-part graph 701 to identify the power saving allocation 704.

[0064]

[0098] An extended ternary graph simplifies the design of the coverage algorithm. Figure 7B shows an example of an extended ternary graph. In Figure 7B, in the left graph, cells A and C are the cells required to turn on cell group A,C. In the right graph, UE1, 2, 3, and 4 are the UEs that are covered when cell group A,C is turned on. Figures 7C and 7D show the original ternary graph and the extended ternary graph, respectively. To initialize the extended ternary graph, an edge is added for each cell group to incorporate the extended ternary graph. In this case, cell group A covers UE1, 2, and 3, and A or C also covers UE1, 2, and 3.

[0065]

[0099] Using an extended graph, the processing logic turns on a single cell group. In some embodiments, it is assumed that all cells are initially in sleep mode. Figure 8A shows an example where a single cell group is turned on. Referring to Figure 8A, the processing logic turns on cell group A,C. Accordingly, the processing logic identifies which cells must be turned on. An example is shown in Figure 8B, where the processing logic turns on cell group A,C, and therefore turns on cells A and C. At this point, the processing logic identifies which UEs are covered by turning on cell group A,C. Since nodes A and C are turned on, UEs 1, 2, 3, and 4 are covered (see Figure 8C).

[0066]

[0100] It should be noted that the operation to identify which UEs are covered can be performed independently of the operation to identify which additional cells must be turned on in response to the cell group being turned on. In other words, these operations can be performed in reverse order. Furthermore, these two initial operations—identifying which cells must be turned on in response to the cell group being turned on, and then identifying which UEs are covered—are the same as those performed on a standard three-part graph. However, these operations are performed sequentially in a standard graph, but not in an extended graph.

[0067]

[0101] After identifying which UEs are covered, the processing logic removes the covered UEs and their edges from the tripartite graph. In this example, UE1, 2, 3, and 4 are already covered, and therefore their nodes are no longer needed and can be removed from the graph. Similarly, all edges of UE1, 2, 3, and 4 are no longer needed and can be removed. Figure 8D shows the tripartite graph after removing the nodes representing UE1, 2, 3, and 4 and their edges.

[0068]

[0102] After removing the covered UE nodes and their edges, the processing logic removed the covered cells and their edges. In this example, cells A and C are no longer needed, and therefore their edges have been removed. Figure 8E shows an example of a three-part graph with cells A and C and their edges removed.

[0069]

[0103] Next, the processing logic merges cell group nodes that share the same edge. This step removes further redundancy by eliminating cell nodes that share the same edge connections (left and right). In this example, cell groups A, B, C and B, C share the same set of edges. Therefore, it is sufficient to track these nodes and one of their edges. Figures 8F and 8G show examples of three-part graphs before and after merging cell group nodes that share the same edge.

[0070]

[0104] In some embodiments, the extended graph can also be represented as a ternary graph formed from a combination of two different standard bipartite graphs (used to represent standard set coverage problems), in which case the source node in each case is the middle node. That is, the extended graph can be created by combining two separate bipartite graphs, each describing a standard set coverage problem. 1. The graph on the right, where the source nodes are a second set of nodes (i.e., cell groups) and the destination nodes are a third set of nodes (i.e., UEs). Such a bipartite graph describes a standard set coverage problem. A destination node (UE) is on (covered) only if and only if the source nodes connected to it by edges are on. 2. The left graph, where the source nodes are again a second set of nodes (i.e., cell groups) and the destination nodes are a first set of nodes (i.e., individual cells). Note that this graph is a horizontally inverted bipartite subgraph of the left graph. A destination node (cell) is on if and only if the source node (corresponding to the cell group containing this cell) to which it is connected by an edge is on. With regard to algorithms applied to these graphs, in some embodiments, two "combined" coverage set algorithms are performed on these graphs. If the algorithm requires calculating the impact of a metric on turning on any given middle node (which is the source node for both graphs), the algorithm first finds the nodes that this middle node covers on the left (cells) and the right (UE), and uses the total number of nodes covered on the left and right to calculate the desired metric as described above.

[0071]

[0105] Therefore, in this style, using a regular or extended ternary graph yields equivalent output. However, extended ternary graphs are usually preferred because they typically offer a significant reduction in algorithm runtime (1 / 60th in some experiments) at the expense of a reasonable increase in memory footprint (e.g., 2 to 3 times).

[0072] Identifying neighboring cells

[0106] In some embodiments, the process of determining which cell to offload a UE to and which cell to de-power is based on identifying which active cells are neighbors of the serving cell currently servicing the UE to be offloaded. A three-part graph can be constructed using all cells in the mobile network based on PCI in the MR from the UE, and then processed using a coverage algorithm to identify the active neighbor cells from which to offload the UE from the serving cell to be de-powered. However, since PCI does not precisely identify each cell, the three-part graph can be very large. See Figure 2B.

[0073]

[0107] One method for reducing the size of a tripartite graph is to reduce the cells used in the tripartite graph to a set of likely neighboring cells. These likely neighboring cells will be identified by a Cell ID (ECI). To identify likely neighboring cells (e.g., the ECI of a cell), a PCI-to-ECI disambiguation process is performed. In some embodiments, disambiguating a set of PCIs involves identifying which combinations of neighboring cells may have resulted in the PCI combination in the UE MR.

[0074]

[0108] In some embodiments, ECI disambiguation is performed cell by cell (cell-specific). In this case, the processing logic, as described above, generates (serving cell, neighbor PCI) pairs from the MR, and these pairs are input to the disambiguation process. Using these inputs, the processing logic generates a set of likely neighbor ECIs. In some embodiments, the processing logic uses these input pairs, along with a sector list, to identify a set of likely neighbor ECIs based solely on PCI values. For example, PCI ECIAll ECIs with =p are identified as a set of likely neighbor ECIs. In some other embodiments, the processing logic uses these input pairs along with the sector list to generate a set of likely neighbor ECIs based solely on the sector list. For example, in some embodiments, PCI ECI All ECIs having =p and overlapping sectors with the serving cell are identified as a set of likely ECIs for the cell. In some other embodiments, the processing logic uses these input pairs along with the sector list, other MR, and MME data to identify a set of likely neighboring ECIs based on the sector list and other network (NW) data. For example, a set of likely ECIs is PCI ECI Both ECIs can have =p. Figure 9A shows these separate embodiments.

[0075]

[0109] In some embodiments, ECI disambiguation is performed per UE (UE-specifically). In this case, from the MR, the processing logic generates tuples of (serving cell, neighbor PCI1, neighbor PCI2, ...), and these tuples are input to the disambiguation process. Using these inputs, the processing logic generates a set of likely neighbor ECIs. In some embodiments, the processing logic uses these input tuples along with a sector list to identify a set of likely neighbor ECIs based solely on PCI values. For example, in this case, the set of likely neighbors includes at least one set of ECIs of the form {E1, E2...}, in which case Ek is PCI ECI = has pk. In some other embodiments, the processing logic uses these input tuples along with the sector list to generate a set of likely neighbor ECIs based solely on the sector list. For example, in this case, the set of likely neighbors includes at least one set of ECIs of the form {E1, E2...}, in which case Ek is PCI ECI=pk has a common coverage area with all sectors and serving cells of Ek. In some other embodiments, the processing logic uses these input tuples along with sector lists, other MR, and MME data to identify a set of likely neighboring ECIs based on sector lists and other network (NW) data. For example, in this case, the set of likely neighbors includes one or more sets of ECIs in the form {E1, E2, ...}, in which case Ek is PCI ECI The ECIs in {E1, E2...} all have =pk and are in the same frequency band. Figure 9B shows these separate embodiments.

[0076]

[0110] In some embodiments, the process of performing neighboring cell disambiguation includes identifying wireless coverage overlap between two or more cells. These two or more cells may include the UE's serving cell and one or more cells in neighboring cell combinations based on the reported PCI. In some embodiments, identifying wireless coverage overlap includes constructing geometric coverage areas for each cell and for neighboring cell combinations. Wireless coverage overlap may be coverage overlap between two or more sectors. In some embodiments, constructing geometric coverage areas for neighboring cell combinations includes constructing one or more convex polygons representing the geometric coverage area for each cell in the neighboring cell combination. In some embodiments, constructing one or more convex polygons representing the geometric coverage area for each cell in the neighboring cell combination is performed using sector list information in one or more propagation models.

[0077]

[0111] More specifically, in some embodiments, the coverage overlap and likely ECI identification is performed as a "per (serving) cell" option having the goal of selecting a single set of likely ECIs for each neighboring PCI, p value reported by a UE in serving cell X, which can be used to offload any UE group in serving cell X reporting neighboring PCI, p. For the per cell option, the inputs and outputs are as follows. Input: · X: Serving ECI · p: Neighboring PCI value reported by a UE in X · Y p Set of all ECIs having PCI = p Output: · Z p =Cov_overlap(X,Y p ): Subset of Y containing only cells that overlap with X in coverage p . In this case, the size of the tripartite graph can be reduced to its basic state without any loss of "information". This loss of information can be illustrated with an example. Assume that in a certain cell, some UEs report PCI(10,15) and other UEs report (10,15,20). Using the cell-by-cell option, each PCI number is resolved by considering only the overlap in coverage with the serving cell, resulting in 10→X1,X2, 15→Y1,Y2, and 20→Z. Therefore, the neighbor PCI list (10,15) is resolved to become the most likely neighbor set list ({X1,X2},{Y1,Y2}), and the neighbor PCI list (10,15,20) is resolved to become the most likely neighbor set list ({X1,X2},{Y1,Y2},{Z}). Therefore, the neighbor PCI list (10,15,20) can be removed from the graph because it is covered by the neighbor PCI list (10,15). Also, the only way a coverage algorithm can cover a UE instance that has the neighbor PCI list (10,15,20) is if the coverage algorithm also covers the UE instance that reported the neighbor PCI list (10,15). However, there is a difference when using per-UE ambiguity resolution. Here, the neighbor PCI list (10,15) can be resolved to become a highly probable neighbor set list ({X1,X2},{Y1}), while the neighbor PCI list (10,15,20) can be resolved to become a highly probable neighbor set list ({X1},{Y1},{Z}). In this case, covering a UE instance that has the neighbor PCI list (10,15,20) does not cover a UE instance that has the neighbor PCI list (10,15). The per-cell option offers a good trade-off between algorithmic complexity and power efficiency.

[0078]

[0112] In some other embodiments, the identification of overlapping coverage and likely ECIs is performed as a “per-UE-group” option with the goal of selecting a set of likely ECIs per PCI for each neighboring UE group n in serving cell X, which has neighboring PCIs p1, p2, ..., and it can be used to offload only specific UE groups in serving cell X. With respect to the per-UE option, the inputs and outputs are as follows: input: ·X: Serving ECI ·n: UE Group Index • {p1, p2, ...}: Neighbor PCI values ​​reported by UE in X ·{Y p1 ,Y p2 ,...}: All ECIs that have PCIs named p1, p2,... output: · Cov_overlap(X,Y p1 ,Y p2 ,...): A subset of overlapping combinations with Serving ECI X This option, in principle, allows for a smaller, more likely ECI set, and consequently, improved power savings. In some embodiments, a basic graph is used for this option. However, for better power savings, it is best to work with the three-part graph 3 in Figure 4A.

[0079]

[0113] In some embodiments, coverage overlap is identified based on circular sectors. This sector overlap method is based on geometric regions. In this case, the geometric coverage of cells is represented as a combination of the omni-coverage region (e.g., a circle) and the region within the circular sector. Figure 10 shows an exemplary overlap.

[0080]

[0114] In some embodiments, sector overlaps are identified using linear constraints via omni-coverage modeling. In this case, each coverage region is represented as a combination of convex regions, and each convex region can be described by a set of linear constraints. Figures 11A–11C illustrate omni-coverage modeling using convex polygons. Referring to Figure 11B, a circular coverage is shown, as used in Figure 10. However, Figures 11A and 11C show circles approximated as separate convex polygons in their respective cases, and each of these convex polygons can be represented by linear constraints. More specifically, the region contained by each convex polygon can be represented as the intersection of half-planes, with defining lines along the segments of that polygon. Each such half-plane can be represented by a single linear constraint. In the case of Figure 11A, the display is coarse because Dθ is equal to 30°, while in the case of Figure 11C, the display is not so coarse because Dθ is equal to 10°. It should be noted that these techniques are not limited to using 10° and 30°, and other Dθ values ​​can be used.

[0081]

[0115] In some embodiments, at least a portion of the cell coverage region can be represented by one or more geometric sectors, each of which can be identified using linear constraints via sector coverage modeling. In this case, the coverage region is a combination of convex regions, where each convex region can be described by a set of linear constraints. In this case, circular sectors can be represented using linear constraints. Figures 12A and 12B show examples of the use of linear constraints to represent circular sectors. In the cases of Figures 12A and 12B, the representation has a Dθ equal to 10°. It should be noted that these techniques are not limited to using 10°, and other Dθ values ​​can be used.

[0082]

[0116] Regarding wide sector modeling, it should be noted that sectors can have a non-convex shape. This can be advantageous for wide beam widths. In some embodiments, a non-convex sector is represented as a combination of two convex regions. Figures 13A and 13B show an example where one non-convex region is represented as two convex regions 1 and 2.

[0083]

[0117] In some embodiments, an algorithm for identifying sector overlaps via linear constraints is provided such that X1, X2, X3 (and possibly X4...) are [X n PCI value p n Identify whether it overlaps with [having]. n Assuming the coverage area is as follows:

number

number

number

number

number

number

number

[0084]

[0118] In some embodiments, the linear program returns one (and only one) of the following as its solution: Case 1: The pair of values ​​(x,y) that yields the maximum value of f(x,y) among all (x,y) pairs that satisfy all linear constraints. * ,y * Case 1: The problem is not feasible because there are no (x,y) pairs that satisfy all linear constraints, and therefore no solution is returned. Thus, in order to identify the overlap of geometric sectors, it is only necessary to know whether the problem returns Case 1 or Case 2. Thus, the choice of f(x,y) is not important, and if the returned solution corresponds to Case 1 (the problem is feasible), there is a common region of coverage overlap among all checked geometric sectors, while if the returned solution corresponds to Case 2, there is no coverage overlap among all checked geometric sectors.

[0085]

[0119] In some embodiments, the algorithm tests all coverage area triplets of three cells in some order. If the algorithm finds coverage overlap in a triplet (i.e., it is a viable LP), the algorithm stops and outputs an indication that coverage overlap was detected. If the algorithm examines all triplets and finds no overlap in any triplets (i.e., all tested LPs are not viable), the algorithm outputs an indication that no overlap was detected.

[0086]

[0120] It should be noted that the geometric sector is not an accurate representation of actual wireless coverage. Therefore, in some embodiments, the geometric coverage region is cautious but not overly cautious. That is, in some embodiments, the UE is at the limit of actual wireless coverage when the received power is between -120 and -125 dBm, and the geometric sector is parameterized via a parameter Prx_min, intended to represent the minimum RX power at the UE that guarantees coverage. For example, creating a sector using Prxmin 15 to 20 dB below -125 dBm is an indirect way to expand the coverage and ensure that any point of actual wireless coverage is included by the geometric coverage sector. It should also be noted that other algorithms may be used to solve the linear constraint.

[0087]

[0121] In some embodiments, the algorithm for identifying sector overlap by linear constraints uses one or more additional algorithmic parameters. For example, in some embodiments, the following parameters are used to identify cell coverage areas. • Sector list information: Latitude and longitude of the access point (AP), beam direction, beam width, transmit (Tx) power, antenna height, etc. • PL Model: Any suitable PL model can be used. In some embodiments, such a PL model can be a PL model from the 3GPP specification. In some other embodiments, the PL model may depend on the deployment scenario (urban macro, urban micro, rural, etc.). • Parameters specified by the user: R nom : Nominal omni-coverage radius for each cell (e.g., 0.25 km, or other settings with separate values ​​for each frequency band)

number

number

number

number

number

number

[0088]

[0122] In some embodiments, the algorithm executes the following loop:

number

number

number

number

number

number

number

number

number

[0089]

[0123] Neighbors may not be detected (for example, the sector coverage algorithm may not find a likely ECI). This occurs when there is no option with overlapping coverage. In some embodiments, in this case, option 1 is as follows: PCI value = for each ECI with a neighboring PCI: Set the sector beam width (or serving cell and given ECI) to 360°. If an overlap is detected, the given ECI is a highly probable ECI. If a likely ECI cannot be found,

number

number

[0090]

[0124] Figures 15–22C illustrate several exemplary embodiments of PCI-to-ECI disambiguation disclosed herein. Figure 15 shows an exemplary network in which a serving cell is overlapped in coverage by two cells X1 and X2 having PCI=10. UE group A includes {a1, a2}, where {a1, a2} are UEs served by the serving cell and are UEs in the wireless coverage area where the serving cell and cells X1 and X2 overlap. This exemplary network also includes that the serving cell is overlapped in coverage by three cells Y1, Y2, and Y3 having PCI=15. UE group B includes {b1, b2, b3}, where {b1, b2, b3} are UEs served by the serving cell and are UEs in the wireless coverage area where the serving cell and cells Y1, Y2, and Y3 overlap. UE group C has 10 and 15 neighboring PCIs, {c 21 ,c 23} includes {c 21 ,c 23} is a UE that is serviced by a serving cell, and is located in a region where there is coverage overlap between the serving cell and cells Y1 and X2, and overlap between serving cell 1501 and cells Y3 and X2.

[0091]

[0125] Figures 16A and 16B show a first example of a cell-by-cell embodiment. Figures 16A and 16B include careful disambiguation (configuration C0) and cell-by-cell disambiguation (configuration C1). Careful disambiguation (configuration C0) is shown in Figure 16A, and for UE group A, all cells X having PCI=10. k For UE group B, all cells Y have PCI=15. k And for group C, all cells X have PCI=10. k , and all cells Y having PCI=15 kThis is how it is specified. After applying cell-by-cell ambiguity, neighboring cells are specified as {X1,X2} for UE group A, as {Y1,Y2,Y3} for UE group B, and as {X1,X2} and {Y1,Y2,Y3} for group C, as shown in Figure 16B.

[0092]

[0126] Figures 17A to 17C show a second example of a cell-by-cell embodiment in which neighboring cells are identified based on bandwidth limitations. Figures 17A to 17C include a first careful disambiguation (configuration C0), a second careful disambiguation based on bandwidth F (configuration C2) in which neighboring cells are identified based on bandwidth F limitations, and a cell-by-cell disambiguation (configuration C3). The first careful disambiguation (configuration C0) is the same as in Figure 16A, and as shown in Figure 17A, neighboring cells are all cells X with PCI=10 for both UE groups A and C. k And for both UE groups B and C, all cells Y with PCI=15 k The second careful resolution of ambiguity based on bandwidth F (configuration C2) is shown in Figure 17B, and for UE group A, all cells X have PCI=10 and bandwidth F. k For UE group B, cell Y has PCI=15 and bandwidth F. k And with respect to UE group C, all cells X have PCI=10 and bandwidth F. k , as well as all cell Y having PCI=15 and bandwidth F k It is specified as follows. After applying cell-by-cell ambiguity, as shown in Figure 17C, the neighboring cell specified for UE group A is cell {X2}, the neighboring cell specified for UE group B is cell {Y2,Y3}, and the neighboring cells specified for UE group C are cell {X2} and cell {Y2,Y3}.

[0093]

[0127] Figures 18A and 18B show a first example of an embodiment for each UE. Figures 18A and 18B include careful disambiguation (configuration C0) and per-UE disambiguation (configuration C4). Careful disambiguation (configuration C0) is shown in Figure 18A, and for UE group A, all cells X having PCI=10. k For UE group B, all cells Y have PCI=15. k And for group C, all cells X have PCI=10. k , and all cells Y having PCI=15 k It is specified as follows. After applying ambiguity resolution for each UE, neighboring cells are specified as cells {X1,X2} for UE group A, as cells {Y1,Y2,Y3} for UE group B, and as cells {X2} and {Y1,Y3} for group C, as shown in Figure 18B.

[0094]

[0128] Figures 19A–19C show a second example of a per-UE embodiment in which neighboring cells are identified based on bandwidth limitations. Figures 19A–19C include a first careful disambiguation (configuration C0), a second careful disambiguation based on bandwidth F (configuration C2) in which neighboring cells are identified based on bandwidth F limitations, and a per-UE disambiguation (configuration C3). The first careful disambiguation (configuration C0) is the same as in Figure 16A, and as shown in Figure 19A, neighboring cells are all cells X with PCI=10 for UE group A. k For UE group B, all cells Y have PCI=15. k And for group C, all cells X have PCI=10. k , and all cells Y having PCI=15 k The second careful resolution of ambiguity based on bandwidth F (configuration C2) is shown in Figure 19B, and for UE group A, all cells X have PCI=10 and bandwidth F. k For UE group B, cell Y has PCI=15 and bandwidth F.k And with respect to UE group C, all cells X have PCI=10 and bandwidth F. k , as well as all cell Y having PCI=15 and bandwidth F k It is specified as follows. After applying ambiguity resolution for each UE, as shown in Figure 19C, the neighbor cell specified for UE group A is cell {X2}, the neighbor cell specified for UE group B is cell {Y2,Y3}, and the neighbor cells specified for UE group C are cell {X2} and cell {Y2,Y3}.

[0095]

[0129] Figures 20A to 20C show a second example of an embodiment in which neighboring cells are identified per UE. Figures 20A to 20C include careful disambiguation (configuration C0), cell-by-cell disambiguation in which neighboring cells are identified (configuration C2), and per UE disambiguation (configuration C4). Careful disambiguation (configuration C0) is the same as in Figure 16A, and as shown in Figure 20A, neighboring cells are identified as all cells X having PCI=10 with respect to UE group A. k For UE group B, all cells {Y1, Y2, Y3} are included, and for group C, all cells X have PCI=10. k , and all cells Y having PCI=15 kThe neighboring cells are specified as follows. Cell-specific disambiguation (configuration C1) is performed as shown in Figure 20B, specifying neighboring cells as {X1,X2} for UE group A, as {Y1,Y2,Y3} for UE group B, and as {X1,X2} and {Y1,Y2,Y3} for UE group C. After applying UE-specific disambiguation, as shown in Figure 20C, the neighboring cells specified for UE group A are {X1,X2}, the neighboring cells specified for UE group B are {Y1,Y2,Y3}, and the neighboring cells specified for UE group C are {X2} and {Y1,Y3}. Comparing Figures 18A-18B with Figures 20A-20C, it becomes clear that both approaches return the same UE-specific configuration C4 as their output. However, the approaches in Figures 20A-20C have even lower runtime requirements. This is because obtaining configuration C4 from configuration C1 requires testing far fewer cell combinations (by searching for geometric sector overlaps) compared to obtaining configuration C4 from configuration C0.

[0096]

[0130] Figures 21A–21C show a third example of a per-UE embodiment in which neighboring cells are identified based on bandwidth limitations. Figures 21A–21C include careful disambiguation (configuration C0), per-cell disambiguation (configuration C2) in which neighboring cells are identified based on bandwidth F limitations, and per-UE disambiguation (configuration C6). Careful disambiguation (configuration C0) is the same as in Figure 16A, and as shown in Figure 21A, neighboring cells are all cells X with PCI=10 for UE group A. k For UE group B, this refers to cells {Y1, Y2, Y3}, and for group C, all cells X with PCI=10. k , and all cells Y having PCI=15 kThe neighboring cells are specified as follows. Cell-specific disambiguation (configuration C3) is performed as shown in Figure 21B, specifying neighboring cells as {X1,X2} for UE group A, as {Y2,Y3} for UE group B, and as {X2} and {Y2,Y3} for UE group C. After applying UE-specific disambiguation, as shown in Figure 21C, the neighboring cell specified for UE group A is cell {X2}, the neighboring cell specified for UE group B is cell {Y2,Y3}, and the neighboring cells specified for UE group C are cell {X2} and cell {Y3}. Careful examination of Figures 20A-20C and 21A-21C reveals that in cases where bandwidth information is also available for PCI-to-ECI disambiguation, Figures 21A-21C provide the same type of runtime benefits as Figures 20A-20C.

[0097]

[0131] Figures 22A–22C show a second example of a per-UE embodiment in which neighboring cells are identified based on bandwidth limitations. Figures 22A–22C include careful disambiguation (configuration C0), per-cell partial disambiguation (configuration C7) in which neighboring cells are identified based on bandwidth F limitations, and per-UE disambiguation (configuration C4). Careful disambiguation (configuration C0) is the same as in Figure 16A, and as shown in Figure 22A, neighboring cells are all cells X with PCI=10 for UE group A. k For UE group B, all cells Y have PCI=15. k And for group C, all cells X have PCI=10. k , and all cells Y having PCI=15 k Specify as follows. Cell-specific ambiguity resolution (configuration C7) is as shown in Figure 22B, with neighboring cells being {X1,X2} for UE group A, and all cells Y with PCI=15 for UE group B. k And for UE group C, cells {X1,X2} and all cells Y having PCI=15. kIt is specified as follows. After applying ambiguity resolution for each UE, as shown in Figure 22C, the neighboring cells specified for UE group A are cells {X1, X2}, the neighboring cells specified for UE group B are cells {Y1, Y2, Y3}, and the neighboring cells specified for UE group C are cells {X2} and cells {Y1, Y3}.

[0098] Typical flowchart

[0132] Figure 23 is a data flow diagram of several embodiments of a data preprocessing operation. In some embodiments, the data preprocessing in Figure 23 is performed prior to creating a ternary graph and identifying which neighboring cells should be used to offload the UE for the purpose of reducing the power of existing serving cells. This process can be performed by processing logic that may include hardware (e.g., circuitry, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0099]

[0133] Referring to Figure 23, MR serving cells are obtained using MR files 1-N (processing block 2301). That is, MR serving cells are identified using MR files 1-N. In some embodiments, a serving cell list is created (processing block 2302). In some embodiments, the serving cell list is used to identify neighboring rectangles and to identify and process duplicate ECIs (processing block 2303). This rectangle limitation can be used for runtime efficiency. Typically, the algorithm described herein is run over an area spanning thousands of serving cells at a time. In central Tokyo, for example, the locations of access points for these cells may span a 10km x 10km rectangular area (specified by minimum / maximum longitude and minimum / maximum latitude). If any cell in the Tokyo area is tested as a potential neighbor candidate, approximately several hundred thousand cells would have to be tested. However, neighboring cells are known to be close together. Therefore, the neighbor search is initially limited to a neighboring rectangular area centered on, but slightly larger than, a 10km x 10km rectangle. In some embodiments, the neighboring rectangular region is defined between 20km and 25km on each side, but regions of other sizes can be used. This limits the potential neighbor candidates to tens of thousands of cells, saving much of the processing time and resources unnecessarily wasted by searching for neighbors in candidate cells that are clearly not neighbors.

[0100]

[0134] As part of data preprocessing, the processing logic creates an active ECI sector list (processing block 2311). In some embodiments, the processing logic creates an active ECI sector list based on pre-ECI to ECI information in the MME file, and ECI (processing block 2321), PCI-enabled information in the sector list file (processing block 2322). With respect to active ECI, in some embodiments, the hardware corresponding to one access point corresponds to a single ECI, with the exception of a few cases where a) the hardware is scheduled to be replaced, b) several new cell configurations are scheduled to be tested, or c) scheduled maintenance is scheduled to be performed. In these special cases, the same cell can be described by two different ECIs, each representing a separate set of hardware components, so that there is only one set of hardware active in the network at any given time. In some embodiments, these two ECIs can be assigned two different PCI values ​​by the network operator. Two ECIs corresponding to the same cell can be identified by checking the sector reference numbers of those ECIs, which are truly unique for each cell (and thus the same for both ECIs). Using the available data (e.g., serving ECIs in MR, MME data, which list ECI destination / handover pairs), it is possible to identify which ECI is active at any given time and use the PCI value of that active ECI in the PCI-to-ECI (or more precisely, to a set of likely neighboring cells) resolution process. Using the entire sector list, the processing logic creates information for the entire cell (e.g., cell information for the entire Tokyo area in the example above) (processing block 2312).The processing logic can create neighboring cell and neighboring sector list information by restricting cell information to nearby MRs based on the serving cell list (processing block 2301). In some embodiments, the output of restricting cell information to nearby MRs is a neighboring cell information dictionary and a neighboring sector list (processing block 2304).

[0101]

[0135] Figure 24 illustrates several embodiments of generating a raw graph dictionary for use when creating a three-part graph, and using that information to support reducing the power of serving cells and offloading the UEs of those reduced-power serving cells. This process can be performed by processing logic that may include hardware (e.g., circuitry, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0102]

[0136] Referring to Figure 24, an MR dictionary is created using MR files 1 to N (processing block 2401). In some embodiments, for the purposes of this specification, the dictionary is a Python dictionary or a similar "information carrying object", X. A Python dictionary has the form X[key]=value with respect to a given set of keys. In some embodiments, each key can be a number, a list of numbers, a word such as "ECI", a list of words, etc. In the case of MRs, the keys of the MR dictionary contain the serving cell ECI, and the corresponding "value" in each case would be a "combined" MR for a given serving cell. A combined MR for a given cell would, for example, take all the MR files, purify them, and make a single "basic" MR for that serving cell. With this in mind, in some embodiments, the flowcharts described herein correspond to embodiments of flowcharts in the Python programming language.

[0103]

[0137] In some embodiments, the MR dictionary is created based on MR files 1 to N and a neighboring sector list, such as the neighboring sector list generated in Figure 23. The MR dictionary contains a single curated description of all MRs.

[0104]

[0138] Using the MR dictionary, the processing logic creates a raw graph dictionary (processing block 2402). In some embodiments, the processing logic creates a raw graph dictionary based on the MR directory and neighbor cell information, for example, the neighbor cell information generated in Figure 23. In some embodiments, the output of creating the raw graph dictionary is a dictionary of cell information, a dictionary of PCI information, and a dictionary of UE information. These dictionaries contain the graph / node dependencies of the ternary graph. In some embodiments, the ternary graph is a careful ternary graph.

[0105]

[0139] Figure 25 is a data flow diagram of several embodiments of a process for creating a three-part graph. This process can be performed by processing logic that may include hardware (e.g., circuits, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0106]

[0140] Referring to Figure 25, the processing logic uses dictionaries of cell information, PCI information, and UE information as input, along with an MME file (with pre-ECI to ECI conversion information) and an NRT file (with ECI to destination ECI conversion information) to create a set of likely ECIs (processing block 2501). In some embodiments, creating a set of likely ECIs involves generating one or more of the following: a sector overlap conflict graph, an MME conflict graph, an NRT conflict graph, and a combined conflict graph. A conflict graph has two sets of nodes: left nodes and right nodes. Conflicts are represented by edges connecting the left nodes to the right nodes. For the purposes of this specification, in cell-by-cell ambiguation, the left nodes are serving cells, and all right nodes are potential neighbor cell candidates. In the case of a sector overlap conflict graph, a conflict represents a sector overlap between the left node (serving cell) and the right node (neighbor candidate). Therefore, conflicts mean likely neighbors. For example, if there is a node S (serving cell) and a PCI value P is reported by a UE in that cell, then X1, X2, ..., X NAssume that all nodes with the ECI are possible neighbor candidates with PCI value = p. If the sector overlap method is performed and overlap is identified only between (S, X2) and (S, X4), then there will be an edge between S and {X2, X4} in the graph, but no edge between S and the rest of Xn. Similarly, using the MME conflict graph, if there are handovers only between S and X2 and between S and X3 in the MME graph, then there will be an edge between S and {X2, X3}, but no edge between S and the rest of Xn. The joined graph "collects" all edges. In this example, the joined graph will have an edge between S and {X2, X3, X4}, but no edge between S and X n There will be no edges between the remaining ones. Therefore, the likely set of neighbors corresponding to PCI value = P will be {X2, X3, X4}. In some embodiments, the output of creating the likely ECI set is a set of ECI sets, which include the set of neighbor ECIs associated with each UE group in MR.

[0107]

[0141] The processing logic updates the graph dictionary of cell information (processing block 2502), PCI information, and UE information using the most likely ECI set, and then combines them into a three-part graph (processing block 2503).

[0108]

[0142] Figure 26 shows several embodiments of a set of algorithmic objects used to create a three-part graph. This process can be performed by processing logic that may include hardware (e.g., circuits, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0109]

[0143] Referring to Figure 26, the processing logic initializes the ternary graph to a predetermined algorithmic state (processing block 2601). In some embodiments, the predetermined algorithmic state is that all ECIs are set to the off position. As part of the algorithm, the processing logic updates the algorithmic state based on a list of cells that are on and a list of bandwidths that are on (e.g., 800 MHz) (processing block 2602). Furthermore, the processing logic also obtains a list of cells that must remain off and excludes certain nodes, such as neighbor PCIs and neighbor ECI groups (processing block 2603). This results in a PCI node exclusion list. In some embodiments, this list specifies middle nodes in the ternary graph (e.g., cell group nodes) that cannot be turned on by the algorithm. Turning on such middle nodes means that cells from the list of off cells must be turned on. In some embodiments, these middle nodes and their edges (to the left and right nodes) are removed from the graph. In one embodiment, the left node corresponding to a cell that must remain off is also removed from the ternary graph.

[0110]

[0144] Figure 27 is a dataflow diagram of several embodiments of a process for identifying which cells to turn off while offloading any of the UEs serviced by those cells. This process can be performed by processing logic that may include hardware (e.g., circuitry, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0111]

[0145] Referring to Figure 27, in some embodiments, this process includes the processing logic selecting an algorithm to execute and preparing its inputs (processing block 2701). In some embodiments, selecting an algorithm (e.g., a coverage algorithm) and preparing the inputs is based on the ternary graph, the algorithm state (e.g., several ECIs being on), the PCI node exclusion list, the selection of the algorithm to be used, and its parameters. Using these inputs, the processing logic selects an algorithm and then executes it (processing block 2702). In some embodiments, the algorithm can be a baseline greedy coverage algorithm, such as the greedy algorithm described above. In some other embodiments, the algorithm can be a parameter-based greedy coverage algorithm (processing block 2703). It should be noted that the techniques described herein are not limited to greedy coverage algorithms, and other non-greedy coverage algorithms can be used. The result of applying the coverage algorithm to the ternary graph is output as the result (processing block 2704). In some embodiments, the output includes an algorithmic state indicating which ECIs will be turned on or off. This information can be used to offload UEs from their current serving cell to other nearby neighboring cells, thereby allowing their current serving cell to be put into a low-power consumption state (e.g., sleep state).

[0112]

[0146] In some embodiments, greedy coverage algorithms that can be used, such as those described above, are given below. Such algorithms can be executed as part of the flowchart in Figure 27 in relation to the three-part graph 2 in Figure 4A. Initialization: E ← Set of ECIs turned on (for example, 45,000 nodes) P ← A set of (PCI,{ECI}) pairs that are ON (for example, 60,000 nodes) U ← A set of uncovered UE groups (e.g., 800,000 nodes) No_steps←0 While there are uncovered UE groups (i.e., U is not empty), do the following: No_steps ← No_steps + 1 Initialization: best_criterion_gain ← 0 (no gain)

number

number

[0113]

[0147] From a high-level description standpoint, in some embodiments, the greedy coverage algorithm includes both an upper and lower bound. Examples of upper and lower bounds are as follows: Upper limit [careful resolution of ambiguity]: Let E be the number of ON ECIs |E| at any given time point. |E| represents the worst-case scenario. If the algorithm is executed with any less cautious PCI → {likely ECI} assignment, a set E' is expected where |E'| ≤ |E|. lower limit: At any given step of the algorithm, this is equal to No_steps (see the algorithm). When the algorithm is executed with any less cautious PCI → {likely ECI} assignment, a set E' is expected that will satisfy the following with a high probability. |E'|≧No_steps The reason is that at every step of the algorithm, No_steps is increased by 1. More UEs are covered, This is because the minimum number of cells required to achieve that additional UE coverage is 1 (which is the same as the increase in No_steps). This is a "rough" lower limit. It is possible to create examples where the resulting E’ with |E’| < No_steps is accompanied by some cells / cell groups / UEs. However, creating such examples is really difficult, and this is less likely to occur, especially in large graphs.

[0114]

[0148] The greedy algorithm may have a high computational load. However, there are many options for speeding up, including parallelizing the code for execution on a graphics processing unit (GPU), using more efficient coding of the greedy algorithm, and modifying the algorithm itself. One option related to using more efficient coding of the greedy algorithm involves identifying which cell group metrics remain unchanged from one algorithm step to the next, and therefore, those cell group metrics do not need to be recalculated.

[0115]

[0149] In some embodiments, for example, algorithms with low complexity, such as those described above, are used. The use of algorithms with low complexity may include speeding up greedy selections, but some power savings provided by the assignment may be lost. The use of algorithms with low complexity may also include the use of multiple candidate selection coverage algorithms. For example, at every step of a greedy selection, one or more cell groups can be selected, and which cell group is selected depends on conditions specified by appropriately adjusted parameters. For example, the algorithm being executed can select the N best cell group nodes at a time. In some embodiments, the value of N can be changed (e.g., decreased) during subsequent iterations of the algorithm.

[0116]

[0150] In some embodiments, the incremental algorithm is as follows: In iteration n, the algorithm calculates a metric for each candidate cell group node that is still off, in which case the cell group node is off if one or more of its cell nodes are off. The algorithm selects the cell group node with the highest metric in iteration n and turns on all cell nodes associated with the selected candidate cell group node.

[0117]

[0151] In some embodiments, a parameterized family of multiple candidate selection coverage algorithms can be used. In some embodiments, a parameterized family of algorithms based on a greedy design may include the following: [Parameters: α, β, N] (For example, a=95%, b=10%, N=10) In iteration n: The algorithm calculates a metric for each candidate cell group that is still off (similar to how an incremental algorithm would work). The algorithm selects the cell group candidate with the highest metric (similar to how an incremental algorithm would work), and potentially additional cell group candidates. X n This represents a selected set of candidate cell groups during iteration n. m best Let (n) represent the highest metric calculated during iteration n. Selected set X n The following conditions must be met: X n Each candidate metric in this context is at least α% of the best metric. X n It has a maximum of N elements (i.e., a maximum of N cell group nodes), and, X n This has up to β% of the candidate cell groups that are considered for selection (i.e., are still in the off state). The algorithm is X n Turn on all cell nodes associated with the selected candidate cell group.

[0118]

[0152] Other algorithmic options exist. For example, the right bipartite portion of the graph can be used (i.e., the cell group → UE group portion of the graph). In this case, a greedy algorithm can be efficiently coded on a bipartite graph. However, this option can lead to a significant performance decrease because the bipartite portion of the graph, which includes cells and cell groups, is completely ignored.

[0119]

[0153] Figure 31 is a dataflow diagram of several embodiments of a process for identifying cell coverage for a UE. This process can be performed by processing logic that may include hardware (e.g., circuitry, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0120]

[0154] Referring to Figure 31, this process includes the processing logic receiving a measurement report transmitted wirelessly from a user device (UE) (processing block 3101). In some embodiments, this process includes the processing logic outputting a recommended state for the serving cell of the measurement report, the recommended state including an active state or a sleep state (processing block 3102).

[0121]

[0155] This process also includes the processing logic creating a three-part graph containing first, second, and third sets of nodes in response to the measurement report, where the first set of nodes represents multiple cells, and the third set of nodes represents multiple UEs that receive coverage from multiple cells (processing block 3103). In some embodiments, each node in the second set of nodes represents a group of one or more cells from multiple cells, and in the three-part graph, there are connections between the cells in the group and one or more UEs, and cell coverage for one or more UEs can be obtained from at least one cell in the group if one or more cells in the group are shown as ON in the three-part graph. In some embodiments, a three-part graph is a graph representing a combination of first and second two-part graphs, the first graph having a second set of nodes and a third set of nodes, wherein a first node from the third set of nodes representing a UE is connected to the first node by an edge, and is turned on only when a second node from the second set of nodes representing a cell group is turned on to indicate that coverage is available; the second graph having a second set of nodes and a first set of nodes, wherein a third node from the first set of nodes representing a cell is connected to a fourth node by an edge, and is turned on only when a fourth node from the second set of nodes representing a cell group is turned on.

[0122]

[0156] Based on a tripartite graph, the processing logic identifies cell coverage for multiple UEs using a set of active cells from multiple cells (processing block 3104). In some embodiments, the processing logic identifies cell coverage for multiple UEs using a set of active cells from multiple cells, based on the tripartite graph, by running a coverage algorithm on the tripartite graph. In some embodiments, the coverage algorithm includes incrementally turning on candidate nodes that are off one at a time, calculating metrics for the candidate nodes when they are on to determine the impact of turning on the cell groups associated with the candidate nodes, comparing the metrics for the candidate nodes with respect to all candidate nodes, turning on one or more candidate nodes based on the comparison of the metrics, and running another iteration of the coverage algorithm on the tripartite graph using candidate nodes that have better metrics when they are on. In some embodiments, the metrics include criteria indicating whether additional UE nodes are covered when the cell groups associated with the candidate nodes are on. In some other embodiments, the metric includes a criterion indicating whether additional UE nodes are covered, which is calculated by dividing the number of extra cells that are on by the number of cells that are on when the cell group associated with the candidate node is on.

[0123]

[0157] In some embodiments, turning on one or more candidate nodes based on a comparison of metrics includes turning on one candidate node that has a superior metric compared to the metrics of other candidate nodes. In some embodiments, turning on one or more candidate nodes based on a comparison of metrics includes turning on a set of cell group candidates, the set of cell groups including cell groups that have a superior metric compared to the metrics of other candidate nodes, cell groups that have a metric within a predefined mathematical relationship with the superior metric, a predefined percentage of cells in a cell group, and one or more of a predefined number of cells in a cell group.

[0124]

[0158] In some embodiments, the processing logic identifies cell coverage for a plurality of UEs based on a three - part graph using a set of active cells from a plurality of cells by executing a plurality of coverage algorithms on the three - part graph, the coverage algorithms identifying candidate nodes in a second set of nodes that cover cells of a second graph and UEs on a first graph, calculating a metric for the candidate nodes when in an on state using the total number of covered nodes, and identifying the impact of turning on cell groups associated with the candidate nodes.

[0125]

[0159] In some embodiments, the processing logic can identify a list of cells that are an option for use in providing coverage for each UE in a measurement report, can convert each neighboring PCI reported in the measurement report in the list of cells into a set of one or more neighboring cells, and the creation of the three - part graph is performed using the set of one or more neighboring cells. 5>

[0126]

[0160] In some embodiments, this process also includes offloading a set of UEs having connections to a serving cell to one or more other nearby active cells (processing block 3105), depending on the identification of cell coverage for the UE. In some embodiments, the offloading occurs depending on the identification of one or more nearby active cells that overlap with the serving cell in radio coverage. Offloading can be based on a list of cells, which is an option for use when providing coverage for each UE in the measurement report.

[0127]

[0161] After offloading the set of UEs to one or more other nearby active cells, the processing logic puts the serving cell into a low-power state (processing block 3106).

[0128]

[0162] Figure 32 is a dataflow diagram of several embodiments of a process for performing neighbor cell disambiguation. This process can be performed by processing logic that may include hardware (e.g., circuitry, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0129]

[0163] Referring to Figure 32, this process includes the processing logic receiving a user equipment (UE) measurement report (processing block 3201).

[0130]

[0164] In some embodiments, the processing logic constructs a geometric wireless coverage area for each cell in a combination of neighboring cells (processing block 3202). In some embodiments, the processing logic constructing a geometric wireless coverage area for each cell in a combination of neighboring cells includes constructing one or more convex polygons, where the combination of one or more convex polygons represents the geometric coverage area for each cell in the combination of neighboring cells. In some embodiments, the processing logic constructs one or more convex polygons representing the geometric coverage area for each cell in a combination of neighboring cells by using sector list information and / or one or more path loss propagation models.

[0131]

[0165] In some embodiments, the processing logic identifies a common area of ​​wireless coverage between the UE's serving cell and a combination of neighboring cells (processing block 3203). In some embodiments, the common area of ​​wireless coverage is the wireless coverage area in which the UE was located when it collected one of its measurement reports. In some embodiments, the processing logic makes this identification based on when the serving cell overlaps in coverage with one or more cells in a combination of neighboring cells.

[0132]

[0166] In some embodiments, the process also includes, depending on which combinations of neighboring cells would have resulted in a set of physical cell IDs (PCIs) in the UE MR, the processing logic representing the geometric coverage areas of cells as combinations of coverage areas represented by complex polygons, and identifying that if there is one intersection of coverage areas represented by complex polygons for each cell, then there exists a common coverage area among the sets of cells that would have resulted in a set of physical cell IDs (PCIs), and that the intersection is also a convex polygon represented by linear constraints. Such identification can include identifying whether an intersection is empty or not based on the feasibility of a linear program by applying the examination of all possible combinations of coverage areas, and identifying that each such combination is represented by a complex polygon, and identifying that such an intersection exists if at least one intersection of such combinations is not empty.

[0133]

[0167] After identifying the common region of radio coverage between the UE's serving cell and the combination of neighboring cells, the processing logic identifies which combination of neighboring cells would have resulted in the set of physical cell IDs (PCIs) in the UE MR by identifying the cells in each combination that have a common region of geometric radio coverage overlap with the UE's serving cell (processing block 3204). In some embodiments, this process includes the processing logic resolving ambiguity of the set of PCIs associated with the UE measurement report.

[0134]

[0168] In some embodiments, the process further includes putting one or more serving cells into a low-power consumption mode (processing block 3205) depending on which combination of neighboring cells would have resulted in a set of physical cell IDs (PCI) in the UE MR, and offloading UEs connected to the serving cell to one or more other neighboring cells (processing block 3206) depending on the identification information of one or more other neighboring cells that overlap with the serving cell in radio coverage.

[0135]

[0169] Figure 33 is a dataflow diagram of several embodiments of a process for identifying cell coverage for a UE. This process can be performed by processing logic that may include hardware (e.g., circuitry, dedicated logic, memory, etc.), software (e.g., running on a general-purpose computer system or dedicated machine), firmware (e.g., software programmed in read-only memory), or a combination thereof. In some embodiments, this process is performed by a network controller in a mobile network communication system.

[0136]

[0170] Referring to Figure 33, this process includes the processing logic receiving the MR that was transmitted wirelessly by the UE (processing block 3301) and obtaining the set of PCI from the MR (processing block 3302).

[0137]

[0171] This process also includes the processing logic identifying cells in each combination that have a common region of geometric wireless coverage overlap with the UE's serving cells (processing block 3303). As part of processing geometric wireless coverage overlap, this process may include generating data representing pairs of serving cells and their respective neighbor cells, and generating a set of identifiers for neighbor cells that are likely to overlap in wireless coverage based on the data representing pairs of serving cells and their respective neighbor cells. In some embodiments, the processing logic identifies cells in each combination that have a common region of geometric wireless coverage overlap with the UE's serving cells by identifying sector overlaps based on geometric regions. In some embodiments, the coverage overlap lies between two or more geometric sectors and leverages the structure of the geometric coverage region.

[0138]

[0172] In some other embodiments, coverage overlap is the coverage overlap of at least two cells, and is identified by testing all tuples of convex coverage regions of at least two cells, determining whether there is coverage overlap with at least one tuple of convex coverage regions of at least two cells, and providing an indication that coverage overlap has been detected when coverage overlap with at least one tuple of convex coverage regions of at least two cells is identified. In some embodiments, when identifying coverage overlap, the processing logic identifies connections between convex regions, each convex region being described by a set of linear constraints. In some embodiments, the convex regions are polygons. In some embodiments, the linear constraints represent regions within the shape of a sector. In some embodiments, the processing logic identifies connections between convex regions by mapping each convex polygon to a set of linear constraints, and by using a linear program with an objective function and linear constraints describing the polygons to determine whether some convex polygons from all the polygons are executable simultaneously.

[0139]

[0173] Based on which cells in each combination have a common region of geometrical wireless coverage overlap with the UE's serving cells, the processing logic identifies which combinations of neighboring cells would have resulted in a set of PCIs in the UE MR (processing block 3304).

[0140]

[0174] In some embodiments, the process further includes, based on the identification of neighboring cells in each combination having a common region of geometric coverage overlap with the serving cell, converting the set of PCIs reported by the UE in the UE measurement report to one or more likely combinations of neighboring cells that have a common region of geometric coverage overlap with the serving cell (processing block 3305).

[0141]

[0175] In some embodiments, the process further includes putting one or more serving cells into a low-power consumption mode (processing block 3306) depending on which combination of neighboring cells would have resulted in a set of physical cell IDs (PCI) in the UE MR, and offloading UEs connected to the serving cell to one or more other neighboring cells (processing block 3307) depending on the identification information of one or more other neighboring cells that overlap with the serving cell in radio coverage.

[0142] Example of a base station

[0176] Figure 30 is a block diagram of several embodiments of a base station. Referring to Figure 30, in one embodiment, the base station 3000 provides service to one or more cells, N t Book Antenna 3035 a ~3035 t The base station 3000 includes a transmit processor 3015, which receives data for one or more UEs from the data source 3010, processes the data for each UE, and transmits the data for each UE. In one embodiment, the processor 3015 also receives and processes information from the controller processor 3070 and provides control symbols. In one embodiment, the processor 3015 also generates reference symbols for one or more reference signals. The transmit (TX) MIMO processor 3020 performs precoding for data symbols, control symbols, and / or reference symbols for each UE based on one or more precoding vectors identified for that UE. In one embodiment, the processor 3020 generates (up to) N tN output streams are provided to each modulator (MOD) in modules 3030a to 3030t, one each. Each modulator 3030 processes its own stream (e.g., for OFDM) to obtain an output sample stream. Each modulator 3030 further processes the output sample stream (e.g., converts to analog, amplifies, filters, upconverts, etc.) to obtain a downlink signal. Up to N outputs are provided from modulators 3030a to 3030t. t Each of the output streams is N t It is transmitted via antennas 3035a to 3035t.

[0143]

[0177] The uplink signal from the UE is received by antenna 3035, processed by demodulator 3030, detected by MIMO detector 3040, and further processed by receiving processor 3045 to obtain the decoded data and control information sent by the UE. Processor 3045 provides the decoded data to data sink 3050 and the decoded control information to controller / processor 3070.

[0144]

[0178] The channel processor 3080 in the base station 3000 estimates channel responses from UE200 and other target UEs and provides a channel matrix for each UE. In one embodiment, processors 3070 and / or 3080 identify channel information based on the channel matrix for each target UE. According to one embodiment, processor 3080 stores the identified channel matrix in memory module 3060 for subsequent use. The base station local admission policy 3090 for base station 3000 handles UE admission, while the resource partitioning block 3075 handles resource partitioning for base station 3000. The calibration processor 3085 performs and controls calibration operations.

[0145]

[0179] In one embodiment, scheduler 3016 schedules the UE for data transmission on the downlink and / or uplink. The scheduler 3016 in base station 3000 and / or other processors and modules are capable of performing processes related to the techniques described herein. These include scheduling the transmission of control information in the uplink by the UE.

[0146]

[0180] Controller / processor 3070 directs the operations in base station 3000. Memory 3060 is capable of storing data and program code related to base station 3000. These operations include the operations described above.

[0147]

[0181] There are several exemplary embodiments described herein.

[0148]

[0182] Example 1 includes receiving a measurement report wirelessly transmitted from a user equipment (UE), and creating a three - part graph including a first, a second, and a third set of nodes in response to the measurement report, wherein the first set of nodes represents a plurality of cells, the third set of nodes represents a plurality of UEs receiving coverage from the plurality of cells, each node in the second set of nodes represents one or a group of one or more of the plurality of cells, in the three - part graph, one or more cells in the group are connected to one or more UEs, and cell coverage for one or more UEs can be obtained from at least one cell in the group if one or more cells in the group are shown as on in the three - part graph, and identifying cell coverage for a plurality of UEs based on the three - part graph using a set of active cells from the plurality of cells.

[0149]

[0183] Example 2 is a method of Example 1 which optionally includes the steps of: offloading a set of UEs having connections to a serving cell to one or more other nearby active cells, where one or more other nearby active cells are identified in accordance with the identification of one or more nearby active cells that overlap with the serving cell in radio coverage; and putting the serving cell into a low-power state after offloading the UEs to one or more other nearby active cells.

[0150]

[0184] Example 3 is the method of Example 1, which optionally includes a step of outputting a recommended state for the serving cell of a measurement report, wherein the recommended state includes an active state or a sleep state.

[0151]

[0185] Example 4 is the same as the method in Example 1, which optionally includes the steps of identifying a list of cells that are optional for use when providing coverage for each UE in the measurement report, and converting each neighbor PCI reported in the measurement report in the list of cells into one or more sets of neighbor cells, wherein the step of creating a ternary graph is performed using one or more sets of neighbor cells.

[0152]

[0186] Example 5 is the method of Example 1, which optionally includes the step of identifying cell coverage for multiple UEs using a set of active cells from multiple cells, based on a tripartite graph, by running a coverage algorithm on the tripartite graph.

[0153]

[0187] Example 6 is the method of Example 5, which optionally includes the steps of: the coverage algorithm incrementally turns on candidate nodes that are in the off state one at a time; calculates metrics for candidate nodes when they are on to determine the impact of turning on the cell groups associated with the candidate nodes; compares the metrics for all candidate nodes; turns on one or more candidate nodes based on the comparison of metrics; and runs another iteration of the coverage algorithm on the ternary graph using candidate nodes that have good metrics when they are on.

[0154]

[0188] Example 7 is an optional method of Example 6 in which the metric may include a criterion indicating whether an additional UE node is covered if the cell group associated with the candidate node is turned on.

[0155]

[0189] Example 8 is an optional method of Example 6 in which the metric may include a criterion indicating whether additional UE nodes are covered, which is calculated by dividing the cell group associated with the candidate node by the number of extra cells that are on.

[0156]

[0190] Example 9 is the method of Example 6, which optionally includes a step of turning on one or more candidate nodes based on a comparison of metrics, which includes turning on one candidate node that has a superior metric compared to the metrics of the other candidate nodes.

[0157]

[0191] Example 10 is the method of Example 6, in which the step of turning on one or more candidate nodes based on a comparison of metrics includes turning on a set of candidate cell groups, which optionally includes one or more of the following: cell groups having a better metric than the metrics of other candidate nodes; cell groups having a metric that is within a predetermined mathematical relationship with the better metric; a predetermined percentage of cells in the cell group; and a predetermined number of cells in the cell group.

[0158]

[0192] Example 11 is the method of Example 1, which optionally includes that the three-part graph is a graph representing a combination of the first and second two-part graphs, and the first graph has a second set of nodes and a third set of nodes, and the first node from the third set of nodes representing UEs is connected to the first node by an edge, and is turned on and only if the second node from the second set of nodes representing a cell group is turned on and only if the second node from the second set of nodes representing a cell group is turned on and only if the second node from the second set of nodes representing a cell group is turned on and only if the third node from the first set of nodes representing cells is connected to the fourth node by an edge.

[0159]

[0193] Example 12 is a method of Example 11 in which the step of determining cell coverage for multiple UEs based on a ternary graph using a set of active cells from multiple cells includes running multiple coverage algorithms on a ternary graph, the coverage algorithms optionally include identifying candidate nodes in a second set of nodes that cover cells in a second graph and UEs on a first graph, and using the total number of covered nodes to calculate a metric for the candidate nodes when they are on to determine the impact of turning on the cell groups associated with the candidate nodes.

[0160]

[0194] Example 13 is a network controller capable of receiving measurement reports (MRs) from serviced user equipment (UEs) and one or more processors coupled to the transceiver, the one or more processors being capable of receiving measurement reports wirelessly transmitted from the UEs and creating a ternary graph in response to the measurement reports, the ternary graph being composed of first, second, and third sets of nodes, where the first set of nodes represents a plurality of cells, the third set of nodes represents a plurality of UEs that receive coverage from the plurality of cells, and each node in the second set of nodes represents a group of one or more cells from the plurality of cells, and in the ternary graph, one or more cells in the group are connected to one or more UEs, and cell coverage for one or more UEs can be obtained from at least one cell in the group if one or more cells in the group are shown as ON in the ternary graph, and identifying cell coverage for multiple UEs based on the ternary graph using a set of active cells from the plurality of cells.

[0161]

[0195] Example 14 is a network controller of Example 13, which optionally includes further operation to perform the offloading, and after offloading the UEs to one or more other nearby active cells, where one or more processors, in response to identifying cell coverage for multiple UEs, offload a set of UEs having connections to a serving cell to one or more other nearby active cells, where one or more other nearby active cells are identified in response to identifying the identification information of one or more nearby active cells that overlap with the serving cell in radio coverage.

[0162]

[0196] Example 15 is a network controller of Example 13, which optionally includes one or more processors that output a recommended state for a serving cell of measurement reports, the recommended state including an active state or a sleep state.

[0163]

[0197] Example 16 is a network controller of Example 13, which can optionally include one or more processors identifying a list of cells that are options for use when providing coverage for each UE in the measurement report, and converting each neighbor PCI reported in the measurement report in the list of cells into one or more sets of neighbor cells, and that creating a ternary graph is performed using one or more sets of neighbor cells.

[0164]

[0198] Example 17 is a network controller of Example 13, which optionally includes the requirement to run a coverage algorithm on a tripartite graph to determine cell coverage for multiple UEs using a set of active cells from multiple cells, based on a tripartite graph.

[0165]

[0199] Example 18 is a network controller of Example 17, which optionally includes the following: the coverage algorithm incrementally turns on candidate nodes that are in the off state one at a time; calculates metrics for candidate nodes when they are on to determine the impact of turning on the cell groups associated with the candidate nodes; compares the metrics for all candidate nodes; turns on one or more candidate nodes based on the comparison of metrics; and runs another iteration of the coverage algorithm on the ternary graph using candidate nodes that have good metrics when they are on.

[0166]

[0200] Example 19 is a network controller of Example 18, which may optionally include one of a group of metrics consisting of a criterion indicating whether additional UE nodes are covered if the cell group associated with the candidate node is on, and a criterion indicating whether additional UE nodes are covered if the cell group associated with the candidate node is on, divided by the number of extra cells that are on.

[0167]

[0201] Example 20 is a network controller of Example 18, which optionally includes enabling one or more candidate nodes based on a comparison of metrics, which means enabling one candidate node that has a superior metric compared to the metrics of the other candidate nodes.

[0168]

[0202] Embodiments of this technology may be described herein in relation to flowcharts of methods and systems according to embodiments of this technology, and / or procedures, algorithms, steps, operations, formulas, or other computational descriptions that may also be implemented as computer program products. In this regard, each block or step of a flowchart, and combinations of blocks (and / or steps) in a flowchart, and any procedure, algorithm, step, operation, formula, or computational description can be implemented by various means, such as hardware, firmware, and / or software containing one or more computer program instructions embodied in computer-readable program code. It will be understood that any such computer program instruction can be executed by one or more computer processors, including but not limited to general-purpose computers or dedicated computers, or other programmable processing units, to produce a machine, thereby generating means for performing the specified (one or more) functions of those computer program instructions executed on the (one or more) computer processors or other programmable processing units.

[0169]

[0203] Accordingly, the flowchart blocks, and procedures, algorithms, steps, operations, expressions, or calculation descriptions described herein support combinations of means for performing a specified (one or more) function, combinations of steps for performing a specified (one or more) function, and computer program instructions (such as those embodied in computer-readable program code logic) for performing a specified (one or more) function. It will also be understood that each block of the flowchart diagrams described herein, as well as any procedure, algorithm, step, operation, expression, or calculation description, and any combination thereof, can be implemented by a dedicated hardware-based computer system, or a combination of dedicated hardware and computer-readable program code, for performing a specified (one or more) function or (one or more) steps.

[0170]

[0204] Furthermore, these computer program instructions (such as those embodied in computer-readable program code) can also be stored in one or more computer-readable memories or memory devices, which can instruct a computer processor or other programmable processing unit to function in a particular manner, thereby producing a product in which the instructions stored in those computer-readable memories or memory devices produce a product in which instruction means perform functions specified in one or more blocks of a flowchart. These computer program instructions can also be executed by a computer processor or other programmable processing unit to produce a computer execution process in which a series of operation steps are performed on that computer processor or other programmable processing unit, thereby providing steps for performing functions specified in one or more blocks of a flowchart, one or more procedures, one or more algorithms, one or more steps, one or more operations, one or more expressions, or one or more calculation diagrams.

[0171]

[0205] It will be further understood that the terms “programming” or “programmable” as used herein refer to one or more instructions that can be executed by one or more computer processors to perform one or more functions described herein. These instructions can be embodied in software, firmware, or a combination of software and firmware. These instructions can be stored locally on a device in non-temporary media, or remotely, such as on a server, or all or part of these instructions can be stored locally and remotely. Instructions stored remotely can be downloaded (pushed) to a device automatically by user initiation or based on one or more factors.

[0172]

[0206] It will be further understood that, as used herein, the terms processor, hardware processor, computer processor, central processing unit (CPU), and computer are used synonymously to describe a device capable of executing instructions and communicating with input / output interfaces and / or peripheral devices, and that the terms processor, hardware processor, computer processor, CPU, and computer are intended to encompass single or multiple devices, single-core and multi-core devices, and variations thereof.

[0173]

[0207] In the claims, references to elements in the singular are not intended to mean “only” (unless explicitly stated otherwise), but rather “one or more.” All structural and functional equivalents to elements of disclosed embodiments known to a person of ordinary skill in the art are expressly incorporated herein by reference and are intended to be encompassed by the claims of the present invention. Furthermore, elements, components, or method steps in this disclosure are not intended to be made available to the public, whether or not such elements, components, or method steps are expressly listed in the claims. Elements of the claims herein should not be construed as “means plus function” elements unless they are expressly listed using the phrase “means for.” Elements of the claims herein should not be construed as “step plus function” elements unless they are expressly listed using the phrase “step for.”

[0174]

[0208] In addition to all other claims, the applicants / inventors(s) claim every embodiment of the technology described herein, every aspect, component, or element of any embodiment described herein, and every combination of aspects, components, or elements of any embodiment described herein.

[0175]

[0209] For anyone with ordinary skill in the art, after reading the above description, many changes and modifications to the present invention will undoubtedly become apparent. However, it should be understood that none of the specific embodiments shown and described as examples are intended to be limiting. Therefore, references to the details of various embodiments are not intended to limit the scope of the claims, and the claims themselves list only the features that are considered essential to the present invention.

Claims

1. A method for resolving ambiguity between neighboring cells, The steps include receiving user equipment (UE) measurement reports, The steps include identifying which combination of neighboring cells would have resulted in the set of physical cell IDs (PCIs) in the UE measurement report by identifying the cells in each combination that have a common region of geometrical wireless coverage overlap with the serving cell of the UE, A method that includes this.

2. Steps include constructing a geometric wireless coverage area for each cell in the aforementioned combination of neighboring cells, The method according to claim 1, further comprising:

3. The step of constructing a geometric wireless coverage area for each cell in the aforementioned combination of neighboring cells includes constructing one or more convex polygons, The method according to claim 2, wherein the combination of one or more convex polygons represents the geometric coverage area for each cell in the combination of neighboring cells.

4. The method according to claim 3, wherein constructing one or more convex polygons representing the geometric coverage area for each cell in the aforementioned combination of neighboring cells is performed by using sector list information and / or one or more path loss propagation models.

5. The aforementioned method, Steps include identifying a common area of ​​wireless coverage between the aforementioned combination of neighboring cells and the UE's serving cell, The method according to claim 1, further comprising the above, wherein the common area of ​​wireless coverage is the wireless coverage area in which the UE was located when the UE collected one of the UE measurement reports.

6. A step of identifying when the serving cell overlaps in coverage with one or more cells in the combination of neighboring cells. The method according to claim 5, further comprising:

7. A step to resolve the ambiguity of the set of PCIs associated with the aforementioned UE measurement report, The method according to claim 1, further comprising:

8. In response to what was identified in the aforementioned identification step, a step of putting one or more serving cells into a low power consumption mode, The method according to claim 1, further comprising:

9. Steps to offload a UE having a connection to the serving cell to one or more nearby cells, in response to the identification in the aforementioned identification step, and in response to the identification of one or more other nearby cells that overlap with the serving cell in wireless coverage, The method according to claim 1, further comprising:

10. The aforementioned method, In response to what was identified in the aforementioned identification step, The step of representing the geometric coverage area of ​​a cell as a combination of coverage areas represented by complex polygons, Step 1: If there is an intersection of coverage areas, each represented by a complex polygon, one for each cell, then there is a common coverage area among the set of cells that would have given rise to the set of physical cell IDs (PCIs), wherein the intersection is also a convex polygon represented by a linear constraint. It further includes, The step of identifying the existence of the aforementioned common coverage area is: By applying the examination of all possible combinations of coverage areas, it is determined, based on the feasibility of the linear program, whether the intersection is empty or not, and each such combination is identified as being represented by a complex polygon. Identifying that such an intersection exists if at least one such intersection of combinations is not empty, The method according to claim 1, including the method described in claim 1.

11. A transceiver for collecting measurement reports (MR) from user equipment (UE) being serviced, One or more processors coupled to the transceiver, Equipped with, The aforementioned one or more processors are Receiving user equipment (UE) measurement reports, By identifying the cells in each combination that have a common region of geometrical wireless coverage overlap with the serving cell of the UE, it is possible to determine which combination of neighboring cells would have resulted in the set of physical cell IDs (PCIs) in the UE measurement report. A network controller capable of performing the following actions.

12. The network controller according to claim 11, wherein one or more processors are operable to construct a geometric wireless coverage area for each cell in the combination of neighboring cells.

13. The network controller according to claim 12, wherein the one or more processors construct a geometric wireless coverage area for each cell in the combination of neighboring cells by constructing one or more convex polygons that collectively represent the geometric coverage area for each cell in the combination of neighboring cells.

14. The network controller according to claim 13, wherein one or more processors are operable to use sector list information and / or one or more path loss propagation models to construct one or more convex polygons that collectively represent the geometric coverage areas of each cell in the combination of neighboring cells.

15. The one or more processors are capable of operating to identify a common area of ​​wireless coverage between the combination of neighboring cells and the serving cell of the UE. The network controller according to claim 11, wherein the common area of ​​wireless coverage is the wireless coverage area in which the UE was located when the UE collected one of the UE measurement reports.

16. The network controller according to claim 15, wherein the one or more processors are operable to determine when the serving cell overlaps in coverage with one or more cells in the combination of neighboring cells.

17. The network controller according to claim 11, wherein one or more processors are operable to resolve ambiguity in the set of PCIs associated with the UE measurement report.

18. The network controller according to claim 11, wherein one or more processors are operable to put one or more serving cells into a low-power consumption mode in response to something that has been identified.

19. The aforementioned one or more processors are The network controller according to claim 11, which is operable to offload a UE having a connection to the serving cell to the one or more nearby cells in response to the identification made in the aforementioned identification, and in response to the identification of one or more other nearby cells that overlap with the serving cell in wireless coverage.

20. The one or more processors, in response to the identification, Representing the geometric coverage area of ​​a cell as a combination of coverage areas represented by complex polygons, If there is an intersection between coverage areas, each represented by a complex polygon, then it is necessary to identify that there is a common coverage area among the sets of cells that would have produced the set of physical cell IDs (PCIs), and that the intersection is also a convex polygon represented by a linear constraint. It is capable of performing the following actions: The aforementioned one or more processors are By applying the examination of all possible combinations of coverage areas, it is determined, based on the feasibility of the linear program, whether the intersection is empty or not, and each such combination is identified as being represented by a complex polygon. Identifying that such an intersection exists if at least one such intersection of combinations is not empty, The network controller according to claim 11, which performs the following:

Images