WO2013055430A2

WO2013055430A2 - A method for handover performance improvement in heterogeneous wireless networks

Info

Publication number: WO2013055430A2
Application number: PCT/US2012/048690
Authority: WO
Inventors: Ismail Guvenc; David LOPEZ-PEREZ
Original assignee: Ntt Docomo, Inc.
Priority date: 2011-10-12
Filing date: 2012-07-27
Publication date: 2013-04-18
Also published as: WO2013055430A3

Abstract

A heterogeneous network is disclosed having macrocells and picocells with co-channel operation for both low-mobility and high-mobility users. The co-channel operation occurs over both coordinated and uncoordinated resources. During normal operation without any handover conditions, the high-mobility user communicates with a macrocell base station in both the coordinated and uncoordinated resources. But if a handover condition to a low-power node (a picocell) arises, the high-mobility user is rescheduled to communicate only in the coordinated resources without allowing a picocell handover despite the handover condition to prevent handover failures (HFs) and to minimize ping-pongs. On the other hand, low-mobility users are allowed to make handover to picocells.

Description

A Method for Handover Performance Improvement in Heterogeneous Wireless Networks

Ismail Guvenc & David Lopez-Perez

RELATED APPLICATION

This application claims the benefit of U.S. Provisional Application No. 61/546,245, filed October 12, 2011.

TECHNICAL FIELD

The present invention relates to wireless communications technology. More specifically, the proposed system can be used to improve handover performance in a heterogeneous wireless network.

BACKGROUND

In homogeneous networks, the mobile users typically use the same set of handover parameters while making a handover to a different cell. However, in a heterogeneous network setting (where there are cells with different coverage areas, such as macrocells, picocells, relays, and femtocells), using the same set of parameters for all cells and for all users may increase the number of handover failures (HFs) and/or ping-pongs (PPs). More specifically, if a user equipment (UE) is in a high-mobility state, a shorter time-to-trigger (TTT) should be utilized as opposed to a relatively-longer TTT for lower-mobility users. The rapid movement of the high-mobility user can significantly degrade the user's signal-to-interference-plus- noise ratio (SINR) in a relatively short amount of time. The shorter TTT for high-mobility users thus reduces the occurrence of HF that would otherwise result from the rapid reduction in SINR. Similarly, when making a handover to a small-cell such as a picocell, a shorter TTT would be preferable to decrease HFs, since the SINR degradation in a smaller cell may happen at a faster rate compared to a larger cell. To address this problem, it is conventional to adjust handover-related parameters using mobility state information, measurement results, or the coverage area of cells where the handover will be performed. However, there are limited works available in the context of heterogeneous networks, especially when range- expansion and interference coordination are considered. Some prior art methods for handover performance enhancement are reviewed further below.

Time to trigger optimization based on mobility state estimation:

Several prior art techniques attempt to improve the handover performance by optimizing the TTT based on mobility state information or measurement report processing. For example, it is known to change the TTT based on mobility state information received from the user equipment (UE) to improve handover performance. For example, UEs with higher mobility can be configured to use shorter TTT values. Similarly, it is known to estimate mobility state using both the handover count and the Doppler spread information, which may then be used to configure handover parameters more efficiently. Both the base station (BS) and the UE can estimate the UE mobility state so that the overall UE mobility state may be estimated by comparing the BS-derived and UE-derived estimations. Then, TTT or some other handover parameters are adapted accordingly. Another prior art approach uses low, medium, and high mobility states based on tracking the number of handovers in a cellular network. Then, a scaling factor is introduced for different mobility states, which is multiplied with the TTT (i.e., a mobility-specific TTT is utilized as opposed to a cell-specific TTT).

Time to trigger optimization based on measurement report processing:

In an alternative approach, handover parameters may also be optimized at the base station by tracking the measurement reports from the UEs. The TTT may then be adapted based on the signal quality from different nodes. By adjusting hysteresis and TTT, handover is only triggered when a UE is beyond a certain distance from its serving cell to avoid ping- pong.

Cell-differentiated handover:

In yet another alternative approach, handover parameters for different cells in heterogeneous network are configured differently. For example, cells may be divided into multiple handover-related classes based on their coverage areas, and each class is then assigned a unique set of handover parameters.

Other handover parameter optimization:

Another prior art approach uses UE measurements on the quality of the connection to the best cell. Then, a mobility-related parameter set (including discontinued reception (DRX) configurations) is selected based on this assessment. Post-handover measurements from the UE (with regard to the previous cell) are used for optimizing handover parameters in the future.

Interference coordination for handover performance enhancement:

The communication resources (time, frequency, coding, and so on) in a heterogeneous network may be coordinated or uncoordinated. An uncoordinated resource is shared equally by the various cells. Tn contrast, a coordinated resource is reserved for a certain cell class. For example, coordinated frequency sub-bands may be allocated to cell-edge vs cell-centered UEs based on quality of pilot measurements. Different sub-bands are allocated to cell-edge as opposed to cell-centered UEs, and cell-edge UEs perform measurements more frequently due to higher possibility of handover. But the use of coordinated resources often inefficiently utilizes the available resources. While modifying handover parameters according to the above-mentioned prior art approaches may provide some gains in handover performance, there are typically some adverse effects (such as increased complexity/overhead, and degradation in some other performance metrics). For example, using shorter TTT values for high mobility users or cells with smaller coverage areas decreases the HF probability. On the other hand, a shorter TTT also implies larger number of ping-pongs, which introduces overhead on the network and may result in losing some packets (which adversely affects the call quality). Accordingly, there is a need in the art for improved handover performance in heterogeneous networks.

SUMMARY

The present disclosure is directed to a heterogeneous network having macrocells and picocells with co-channel operation for both low-mobility and high-mobility users. The co- channel operation occurs over both coordinated and uncoordinated resources. During normal operation without any handover conditions, the high-mobility user communicates with a macrocell base station in both the coordinated and uncoordinated resources. But if a handover condition to a low-power node (a picocell) arises, the high-mobility user is rescheduled to communicate only in the coordinated resources without allowing a picocell handover despite the handover condition to prevent HFs and to minimize ping-pongs. On the other hand, low-speed users are allowed to make a handover to picocells. The handover parameter set is selected more flexibly and effectively by benefiting from interference coordination, which improves the overall handover performance due to combined use of interference coordination and handover parameter optimization. As an improvement to prior- art approaches, mobility states of the users are estimated utilizing the topology of the network, such as the number of low-power nodes that are present in the coverage area of a macrocell. BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 illustrates a heterogeneous network and trajectories of three different users with different mobility states.

Figure 2 illustrates the resource coordination for a conventional heterogeneous network.

Figure 3 illustrates the resource coordination for a heterogeneous network in accordance with an embodiment of the disclosure.

Figure 4 is a flowchart for a mobility-based interference coordination technique for handover optimization.

Figure 5 is a chart of handover performance without mobility-based interference coordination.

Figure 6 is a flowchart for a mobility state estimation technique in accordance with an embodiment of the disclosure.

Figure 7 is a block diagram of a base station and a UE configured to practice mobility-based interference coordination in accordance with an embodiment of the disclosure.

DETAILED DESCRIPTION

In a heterogeneous network that includes both macrocells and picocells, it is advantageous for a macroccll user equipment (UE) to handoff to a picocell when appropriate. Determining when it is appropriate to handoff to a picocell depends upon the resource allocation for the heterogeneous network - in particular, whether the macrocells and picocells are sharing resources (co-channel) or whether they operate entirely on dedicated resources (split-channel). Since co-channel operation is more bandwidth efficient, the following discussion will assume that the macrocells and the picocells share co-channel resources.

These shared resources are denoted herein as uncoordinated resources. In contrast, the macrocell also has some dedicated resources (not shared with the picocell) that are denoted as coordinated resources. During normal operation (no handoff conditions being indicated), the macrocell UE exploits both the coordinated and uncoordinated resources. But as a high- mobility lnacrocell UE encroaches on the coverage area of a picocell, the high-mobility macrocell UE may be rescheduled to use only the coordinated resources such that the high- mobility UE is not handed off to the picocell. Conversely, as a low-mobility macrocell UE encroaches the coverage area of a picocell, the low-mobility macrocell UE may be handed off to the picocell.

Since a rescheduled high-mobility macrocell UE may leave the vicinity of picocells, it would be inefficient to continue operation solely in the coordinated resources for such a UE. Thus, the handover performance method disclosed herein includes embodiments in which the handover state (whether or not a handover is indicated) for the UE is continually monitored. If the monitoring indicates that a rescheduled UE has no handover indications (it is traveling in the macrocell in an area without any picocell coverage), the non-coordinated resources are tested to determine if these resources are satisfactory. If the non-coordinated resources are satisfactory, the UE is scheduled to use both the coordinated and non-coordinated resources regardless of the current mobility state for the UE. As used herein, a UE is deemed to be rescheduled if it is communicating only over the coordinated resources and to be scheduled if it is communicating over the coordinated and uncoordinated resources.

Additional details for the improved handover performance in heterogeneous networks may be better understood with reference to the drawings. Figure 1 shows an example heterogeneous network. There are seven base stations such as evolved node B (eNBs) 101 and 102, each having coverage area 100. In addition, the heterogeneous network includes a plurality of picocell base stations (picocell nodeB's (PNBs)) 104 that are placed inside the coverage areas of the macrocell eNBs 102. If no range expansion is applied, each picocell 104 has a coverage area 106, whereas with range expansion, each picocell has an expanded coverage area 108. Three user equipments UE-1, UE-2, and UE-3 are initially within the macrocell controlled by eNB 101. Both UE-1 and UE-2 are high-mobility UEs whereas UE-3 is a low- mobility UE. In other words, UE-1 and UE-2 are traveling at a relatively high speed whereas UE-3 is traveling at a relatively low speed. UE-2 and UE-3 follow paths 110 and 120, respectively, that take them through the coverage area of a picocell 1 OS. Conversely, UE-1 travels a path 115 that is outside of picocell coverage areas. UE-1 should thus have no handover conditions and continue to operate as an unscheduled user in both the coordinated and non-coordinated macrocell resources. UE-2, however, will experience a handover indication as it travels through picocell 105. But rather than handoff to picocell 10S, UE-2 is rescheduled to communicate with base station 101 only in the non-coordinated resources. Since UE-3 is a low-mobility user, it can be handed off to picocell 1 OS as it enters the picocell coverage area.

In homogeneous networks, all the UEs typically use the same set of handover parameters while making a handover to a different cell. However, in a heterogeneous network setting, using the same set of parameters for all cells and for all users may increase the number of HFs and/or ping-pongs. More specifically, if a UE is in a high-mobility state, a smaller time-to-trigger should be utilized to reduce the probability of a HF as the UE's signal to interference plus noise ratio becomes significantly degraded. Therefore, when making a handover to a small-cell such as a picocell, a smaller value of TTT would be preferable to decrease HFs, since the SINR degradation in a smaller cell may happen at a faster rate compared to a larger cell. As discussed above, various conventional approaches adapt handover parameters using the mobility state information, measurement results, or the coverage area of cells into which the handover will be performed. This adaptation attempts to optimize handover performance.

While this conventional optimization of handover parameters may provide some gains in handover performance, the resulting methods suffer from adverse effects and complexity. For example, using shorter TTT values for high-mobility users or for handoff to smaller cells decreases the HF probabilities. On the other hand, shorter TTT also implies larger number of ping-pongs, which introduces overhead on the network and may result in losing some packets and lower call quality. The inverse relationship between HF probability and ping-pong probability as a function of TTT is discussed further below with regard to Figure 5.

Ideally, high-mobility users should never be handed off to picocells or other types of low-power cells, since this would introduce a large number of handovers and a large overhead. But it is conventional to enforce such a no-handoff rule for high-mobility users in a split-channel environment (macrocell and picocell users having their own dedicated resources), which is inefficient. In contrast, the interference coordination approach to improve handover performance disclosed herein for macrocell users in a heterogeneous network is a co-channel approach, which efficiently uses resources.

In a heterogeneous network setting, the high-velocity users can be seen as a third category of victim users, which, with conventional handover procedures, can observe HF and/or ping-pong problems. Inter-cell interference coordination (in combination with handover parameter optimization) can be used as an effective tool to prevent such problems. An example set of N resources are shown in Figure 2 that are allocated in a conventional fashion. While the following discussion assumes that the N resources are time slot resources, the interference coordination disclosed herein is widely applicable to frequency allocation (e.g., component carriers), code allocation (e.g., codes as in CDMA systems), or spatial domain allocation (e.g., beam directions). The conventional resource allocation shown in Figure 2 has the macrocell leave certain resources 212 dedicated to the picocell. In contrast, the picocell occupies all its available resources 215. Thus, the picocell may schedule its own victim users (e.g., its range-expanded users) in its coordinated resources (in this case, slots R₃ and R₄). Because the macrocell is excluded from resources 212, it only occupies a remaining set of resources 205. In contrast, the resource allocation shown in Figure 3 has its own coordinated resources 210 for the macrocell. Thus, the picocell is excluded from a corresponding set of resources 220 (slots R_N-1 and R_N). This is advantageous because high-mobility macrocell users may also be victim users in a heterogeneous network environment due to handover failures and ping-pongs. In order to protect such high-velocity macrocell users, the interference coordination approach disclosed herein reschedules the macrocell users into just the coordinated resources. Note that in a general setting, there may be more than two sets of resources with different SINR characteristics, which may e.g. be due to use of different coordinated resource patterns in different cells.

Which users to reschedule in coordinated resources and how/when to schedule them back in non-coordinated resources can be better understood with reference to the example flowchart in Figure 4. This interference coordination process begins with an estimation of the mobility state for a UE. While there are a number of conventional ways on how the mobility state of a UE can be estimated (where most approaches rely on handover count of a UE due to its simplicity), the interference coordination approach of the present disclosure may use an improved mobility-state estimation method for heterogeneous network scenarios as will be described further herein. Alternatively, a conventional mobility-state estimation can be performed. For example, it is known to use Doppler spread information to estimate a UE's mobility state. As disclosed herein, the mobility state can be estimated in a step 300 at the UE-side (e.g., for idle mode UEs), network-side (e.g., for active mode UEs), or both. If the mobility state of a UE is estimated at the UE side, it may be communicated to the cNB in the form of a single bit that indicates whether the UE is low-mobility or high-mobility. The UE also sends to the serving eNB the signal measurements that are obtained from candidate cells 310 so that handover decisions can be made either at the eNB or higher up in the network. Note that typically, the UEs are configured to send measurements collected from only one set of resources; e.g., from uncoordinated resources 205 or coordinated resources 210, depending on where it is configured to collect measurements and receive its downlink transmissions.

Once the mobility state information and the measurement reports are obtained at the eNB, the eNB decides whether a handover is necessary for the UE in a step 320, using methods similar to those used in the prior art. If a handover is indicated for the UE, the eNB also checks the mobility state of the UE in a step 330. If the UE is a low-mobility UE, the eNB initiates the legacy handover process to the target node in a step 380. On the other hand, if the UE is a high-mobility UE, it may face with a HF before completing such a handover, or may observe ping-pongs. Therefore, high-mobility UEs are rescheduled in the coordinated resources of macrocell in a step 340 without any handover and report their measurements only from the coordinated resources (e.g., from resources 210). Therefore, even if the macrocell UE travels into the coverage area of a picocell, it does not observe interference from the picocell, preventing any HFs due to high picocell interference. Moreover, since a potentially unnecessary handover is avoided (due to very short time of stay of the UE inside the picocell coverage), ping-pongs are prevented. Note that this rescheduled UE may also cause uplink interference to the picocell base station while inside the coverage of the picocell so that similar interference coordination mechanisms discussed in this disclosure can be applied to prevent such uplink interference.

Referring back to step 320, if no handover is required the eNB first checks whether the UE is rescheduled in a coordinated resource group in a step 360. If the UE is not rescheduled in a coordinated resource group, no handover or resource scheduling of the UE is performed in a step 350 such that the eNB takes no action about this UE until the next set of measurements and mobility state information arrives at the eNB. On the other hand, if the determination in step 360 shows that the UE is rescheduled in a coordinated resource group, that UE should release the coordinated resources because there is no handover condition. The lack of a handover indication shows that UE is no longer a potential victim of a future handover

Whether and how such coordinated-resource UEs should release the coordinated resources depends on two issues: 1) a determination of whether the UE is a low-mobility or high-mobility UE in a step 370, and 2) a measurement report in the non-coordinated resources in steps 365 and 375. If the UE is a low-mobility UE, that would mean that the measurements in step 300 were solely in the coordinated resources (because the UE was shown to have been rescheduled in step 360 to just the coordinated resources).

Measurements in just the coordinated resources in step 360 could thus miss the availability of picocell coverage. Thus, if measurements in the non-coordinated resources are not satisfactory in step 375 (e.g., the UE is inside the coverage area of a picocell such that the SINR in the non-coordinated resources is smaller than a given threshold), the UE can not simply be scheduled in the non-coordinated resources due to a potential HF. It follows that if the determination in step 375 is negative, a legacy handover process should be initiated to the target node in step 380. Due to the low mobility state of the UE, it can enjoy the better SINR by connecting to the low-power target node without observing HFs or ping-pongs. On the other hand, if the measurements in the non-coordinated resources are satisfactory and the UE is a low-mobility UE, it can be scheduled in the non-coordinated resources in a step 390 without triggering any handover.

Should step 370 indicate that the rescheduled UE (as used herein, a rescheduled UE refers to one that is using just the coordinated resources), the eNB also checks the

measurements in non-coordinated resources in step 365. If the measurements are satisfactory (e.g., UE is no longer in the coverage area of a picocell), the UE can be rescheduled in the non-coordinated resources in step 390; otherwise, no action is taken for handover in a step 350. Similarly, step 350 involves keeping the rescheduled UE rescheduled. Note that as mentioned before, the UEs typically report measurements only fiom a single resource group set. For example, if a UEs is rescheduled to just use coordinated resources, such a rescheduled UE does not normally report the measurements from the non- coordinated resources in order to prevent messaging overhead. Hence, in order to evaluate the link quality of non-coordinated resources, in one embodiment, rescheduled UEs may be specifically configured by the eNB to obtain and transmit measurement reports in non- coordinated resources. Such measurements may occur in periodical intervals, or, as the need arises (e.g., when the coordinated resources of the eNB become highly loaded by high- mobility users). Alternatively, to avoid the signaling overhead due to communicating additional measurement reports (of different resource groups) between the UE and the serving node, a UE that is rescheduled in the coordinated resources may continue using these resources until the active call terminates.

Representative Performance Results to Illustrate Related Trade-Offs:

Using the proposed approach summarized in Figure 4, it is possible to flexibly select the handover parameters so as to minimize both the HFs and the number of ping-pongs. In order to more clearly demonstrate the operation mechanism and benefits of the interference coordination technique, computer simulations that investigate the HF rates and ping-pongs in different scenarios are performed. Five different handover parameter sets to be used in the simulations are summarized in the following Table 1, where shorter TTTs are used for larger set numbers.

The A3 offset corresponds to the hysteresis used during the handover process, and the Reference Signal Received Power (RSRP) L3 filter value is used to average out the impact of fast fading in signal measurements. The handover performance results are shown in Figure 5 using the handover parameters from Table 1. There are four randomly deployed picocells per sector, and each picocell uses a 8 dB range expansion bias. The range-expanded picocell UEs are scheduled at the coordinated resources of picocell (where no macrocell transmission occurs). On the other hand, there is no interference coordination for the macrocell UEs. The simulation results in Figure 5 show that when large TTTs such as 160 ms or 480 ms are used, the number of ping-pongs are minimized but the HF rate may be very large for high velocities. On the other hand, for small TTTs, the HF rates are low but the ping-pong rates are high. In this particular scenario, the handover parameter set that provides the best compromise between the HFs and ping-pongs would most likely be set-3. However, while satisfactory handover performance can be obtained for low-mobility UEs, high- mobility UEs may suffer from HFs or ping-pongs.

When mobility-state based interference coordination is applied for the macrocell users, the handover performance can be improved at the cost of releasing some resources at the picocells. Consider, for example, that the threshold that classifies a UE as low-mobility or high-mobility is taken as 60 km/hour. Then, if the true velocity of a UE is larger than 60 km/hour, it is rescheduled in coordinated resources of the macrocell, and no HFs or ping- pongs are observed among the macrocell and the picocell. For users with velocities lower than 60 km hour, if handover parameter set-3 is used in this scenario, maximum HF rate is 5% (decreases for higher velocities), while ping-pong rates are between 5% and 10%. Even better, handover parameter set-2 yields slightly larger HF rates, but yields almost no ping- pongs. Therefore, both ping-pong and handover performance can be improved with the proposed approach compared to using interference coordination only at the picocells as in Figure 5. Moreover, using set-2 would be preferable when ICIC is applied, while set-3 (which yields lower number of HFs at higher velocities) would be preferable with no ICIC. Hence, the use of interference coordination may change the ideal set of handover parameters for the best handover performance.

As mentioned before, the interference coordination and improved handover performance for heterogeneous networks disclosed herein requires the release of certain resources at the picocells, which introduces some degradation in the performance (e.g., capacity, measurement report quality due to less averaging, etc.) of picocell users. However, in typical scenarios, releasing only a small portion of a picocell's resources may yield important gains in the performance of high-velocity macrocell users. An adaptive resource partitioning where the amount of resources released at the picocell varies depending on the number of high-velocity users in the macrocell may also be possible. Based on the measurement reports and the mobility state information of all UEs, the eNB may jointly optimize/configure the handover parameter sets and the interference coordination ratio and pattern. For example, if there are no high-mobility users in the network, the eNB may con igure the picocells not to release any resources (e.g., over the X2 interface in LTE). As mentioned before, an optimum handover parameter set also depends jointly on whether the ICIC is utilized or not at a macrocell.

In one embodiment, the blank resource pattern used by the picocell is designed in a way such that it minimizes the interference to crucial resources of macrocell. For example, an LTE picocell may align its blank subframes with the primary/secondary synchronization channel, physical broadcast channel, and system information block of the macrocell. The blank subframe pattern at picocell may also be designed jointly with one or more macrocells. Improvement of Mobility State Estimation in Heterogeneous Networks:

Performance of the proposed interference coordination approach relies on the accuracy of the mobility state estimation technique (step 300 in Figure 4). There are prior-art mobility state estimation approaches that count the number of handovers for a UE, which can be implemented at the UE or at the network. However, these approaches are designed for homogeneous network scenarios, and will not work accurately in heterogeneous network environments. A mobility state estimation threshold should ideally be modified based on the number of picocells and their range expansion bias values inside a macrocell. In one embodiment, it is proposed to use a look-up table to select the handover count threshold while estimating the mobility state of a UE. An example look-up table is shown in Table 2, which shows the dependence of mobility state threshold on number of picocells per sector and the range expansion bias. The look-up table may also involve parameters other than the number of picocells per sector and the range expansion bias.

The table shows how the handover count threshold (which if exceeded, is used to denote a UE as a high-mobility UE) increases as the number of picocells is increased from zero to four. In addition, the table shows how the handover count threshold is increased in response to picocell range expansion for those macrocells having three or more picocells.

In addition to the selection of the handover count threshold, the duration of the sliding time window within which the total number of handovers are counted may also be adjusted in some embodiments. This would enable quicker mobility state estimates in denser heterogeneous network deployments, since smaller window duration to count the number of handovers would be sufficient to have a reliable mobility state estimate due to the larger number of handovers.

A flowchart that summarizes how a lookup table such as Table 2 can be used for improving the mobility state estimation of UEs is shown in Figure 6. First, if mobility state estimation at the UE is supported in a step 301, the UE receives the relevant cell-specific information such as the number of picocells and their range expansion bias configurations within a given sector in a step 302. Alternatively, the serving eNB may also directly calculate the threshold value and communicate the threshold to the UE. These cell-specific data may be transmitted by the system information broadcast of the serving eNB. Then, an appropriate mobility-state threshold value is selected from the look-up table in a step 304 (along with a selection of an appropriate time window). In a step 306, the UE estimates its mobility state based upon the selected threshold value and notifies the eNB of the estimated mobility in a step 306. In addition to the mobility state threshold, the look-up table may also return the duration for the sliding time window. An example value of a sliding window in a homogeneous network may be 30 seconds, but for heterogeneous network environments, smaller time windows can be used. Finally, the eNB decides on the mobility state of the UE in a step 308. Note that this procedure may be applied both for idle-mode and active-mode UEs. If the UE is in active mode, the network already has the information on the number of handovers for the given UE. Therefore, as an alternative embodiment, the mobility state may be directly estimated at the eNB, which removes the requirement for the UE to transmit its mobility state to the eNB. Alternatively, the mobility state may be estimated both at the UE- side and network-side, to obtain a more reliable mobility state decision at the e B.

Block Diagram for the Proposed Mobility-Based Interference Coordination;

A block diagram for a macrocell base station 500 and a UE 550 that implements the proposed mobility-based interference coordination method is shown in Figure 7. The UE is initially connected to the serving eNB. On the other hand, it also receives signal

measurements from nearby nodes (macrocells, picocells, etc.) 600, 610, and 620 at its antenna 555. The UE receives a downlink signal 640 from the serving eNB as well as downlink signals 605, 615, and 625 from the nearby network nodes. Based on these measurements, the UE checks whether certain conditions are satisfied, e.g., if the link quality of serving node 500 is worse than the link quality of one of the target nodes 600, 610. 620 plus a given threshold. This measurement report processing is performed in a measurement processing engine in the UE in an analogous fashion to prior art report processing and communicated to a handover decision engine 535 in the serving eNB. Note that the logical links between different engines are shown in dashed lines, while the information exchange actually happens through scheduling of the corresponding messages at the UE and the eNB, and transmission/reception through antennas 515 and 555.

UE may estimate its mobility state using a mobility state estimation engine 585, such as through tracking the total number of handovers within a given sliding time window and using the look-up table to adjust its mobility state estimation parameters. The eNB may broadcast mobility state estimation parameters in its system information broadcast, which can be used by the UE for improved mobility state estimation. Final decision for the mobility state of a UE is made at a mobility state decision engine 540 of the eNB. Instead of getting a mobility state estimate from the UE, the eNB may also directly utilize a similar approach for counting the number of handovers and deciding on the mobility state of the UE. Once an estimate for the mobility state of a UE is available, it is used jointly with the received measurement reports for deciding whether a handover is needed for the UE. Hie base station receives upstream communication from an operator network 545 as known in the base station arts.

An interference coordination engine 530 in the eNB can configure the ratio of coordinated resources that would be required for mobility-based interference coordination. Depending on the number of high-mobility UEs, it may configure more or less resources as blank in the picocells (e.g., through the X2 interface in LTE). A scheduler 525 schedules the UEs in coordinated or non-coordinated resources based on the measurement reports and mobility state information, such as discussed with regard to Figure 4. A scheduler 575 in the UE responds accordingly such that the UE is either rescheduled or uses both the coordinated and uncoordinated resources. The base station includes a signal generator 505 and a transceiver 510 for its RF communications with the UE. Similarly, the UE includes both a signal generator 570 and a transceiver 565 for its RF communications with both the macrocell base station and the picocell base station. Once a UE is scheduled in a certain type of resource, the measurements are collected only from those resources in the eNB, before being processed in a measurement processing engine 580 of the UE.

It will be appreciated that the various calculation engines shown in Figure 7 such as engines 530, 535, and 540 in the base station as well as engines 580 and 585 in the UE may all be implemented by suitable processors, each associated with a corresponding memory. In that regard, the embodiments shown in Figure 7 have a separate processor for each engine. But a single processor can be readily programmed to implement the necessary engines in each network node. Embodiments described above illustrate but do not limit the invention. Thus, it should also be understood that numerous modifications and variations are possible in accordance with the principles of the present invention. Accordingly, the scope of the invention is defined only by the following claims.

Claims

CLAIMS We claim:

1. A method of interference coordination for a macrocell user equipment (UE) communicating with coordinated and uncoordinated resources within a macrocell, the macrocell including a picocell that communicates over the uncoordinated resources but is excluded from the coordinated resources, the method comprising:

at a macrocell base station serving the UE, receiving first measurement reports from the UE relating to measurements of downlink signals from the macrocell base station and from a picocell base station in the picocell;

determining that a first handover condition exists to handoff the UE from the macrocell to the picocell based upon the measurement reports;

in the macrocell base station, commanding the UE to reschedule so as to communicate using only the coordinated resources in response to a determination that the UE is a high-mobility UE, the macrocell base station declining to command a handover of the high-mobility UE to the picocell in response to the determination that the UE is a high- mobility UE.

2. The method of claim 1 , wherein the coordinated and uncoordinated resources are time-domain resources.

3. The method of claim 1, wherein the coordinated and uncoordinated resources are frequency-domain resources.

4. The method of claim 1, wherein the coordinated and uncoordinated resources are code-domain resources or spatial-domain resources.

5. The method of claim 1 , further comprising:

at the macrocell base station, receiving second measurement reports from the high- mobility UE, the second measurement reports relating only to the coordinated resources; based upon the second measurement reports, determining that the handover condition no longer exists for the high-mobility UE;

in response to a third measurement report detennining that the non-coordinated resources are satisfactory and in response to the determination that the handover condition no longer exists, commanding the high-mobility UE to communicate in both the coordinated and uncoordinated resources.

6. The method of claim 1, wherein determining that the UE is a high-mobility UE comprises counting a number of handover conditions for the UE over a period of time.

7. The method of claim 6, wherein determining that the UE is a high-mobility UE further comprises comparing the number of handover conditions to a threshold.

8. The method of claim 7, further comprising looking up the threshold in a lookup table as a function of a number of picocells within the macrocell.

9. The method of claim 8, wherein looking up the threshold in the lookup table is also a function of picocell range expansion.

10. A base station for a macrocell, comprising:

a handover decision engine configured to determine whether a user equipment (UE) served by the base station should be handed off to a picocell within the macrocell;

a mobility state decision engine configured to determine a mobility state for the UE; and

an interference coordination engine configured to generate a command to command the UE to reschedule from communicating over coordinated and uncoordinated resources to just coordinated resources in response to the handover decision engine determining that the UE equipment should be handed off to the picocell and in response to the mobility state decision engine determining that the UE is a high-mobility UE.

11. The base station of claim 10, wherein the coordinated and uncoordinated resources are time-domain resources.

12. The base station of claim 10, wherein the coordinated and uncoordinated resources are frequency-domain resources.

13. The base station of claim 10, wherein the coordinated and uncoordinated resources are code-domain resources or spatial-domain resources.

14. The base station of claim 10, wherein the mobility decision engine is further configured to determine the mobility state using a count of handover conditions for the UE that occur over a period of time.

15. The base station of claim 14, wherein the mobility decision engine is further configured to determine the mobility state by comparing the count of handover conditions to a threshold based upon a count of picocell within the macrocell.

16. A method of interference coordination to improve handover performance in a heterogeneous network of a macrocell that includes a picocell, comprising:

at a user equipment (UE) within the macrocell, measuring signal quality over both coordinated resources and uncoordinated resources, wherein the coordinated resources are reserved for the macrocell and the uncoordinated resources are shared with the picocell; transmitting the measured signal quality of both the coordinated and uncoordinated resources from the UE to a macrocell base station, wherein the measured signal quality of the uncoordinated resources compared to the measured signal quality in the coordinated resources is such that a handover condition exists for handing off the UE from the macrocell base station to a picocell base station; and

despite the existence of the handover condition, rescheduling the UE to communicate only over the coordinated resources without handing off to the picocell base station in response to a determination that the UE is a high-mobility UE.

17. The method of claim 16, further comprising:

within the UE, making the determination that the UE is a high-mobility UE based upon a count of handover conditions experienced by the UE over a period of time.

18. The method of claim 16, further comprising:

at the high-mobility UE subsequent to the rescheduling, making additional signal quality measurements in just the coordinated resources;

transmitting the additional signal quality measurements to the macrocell base station, wherein the additional signal quality measurements do not indicate an existence of a handover condition;

in response to the indication that a handover condition does not exist and in response to an indication that signal quality in the uncoordinated resources is satisfactory, scheduling the high-mobility UE to communicate over both the coordinated and

uncoordinated resources.

19. The method of claim 16, wherein the coordinated and uncoordinated resources are time-domain resources.

20. The method of claim 16, wherein the coordinated and uncoordinated resources are frequency-domain resources.