FDD-LTE Handover Optimization Part 1

1         Handover Overview

1.1       Necessity for Handover

Wireless communication has an important feature: mobility control. When a UE moves to the edge of a cell, the system needs to hand over the UE to another cell with better signal strength to ensure service continuity. During a handover, the UE and network operate together to complete signaling interaction. Handovers in LTE are hard handovers, meaning that there is a short service interruption when the handover is performed. To ensure that services are not affected, the handover success rate and handover throughput must be guaranteed. If a handover fails, user experience is severely degraded.

1.2       Handover Procedure

Handover instructions are sent from the eNodeB side. UEs actively report their radio conditions to the eNodeB, so that the eNodeB can determine that the UEs are located on the cell edge. There are two types of reporting, event-triggered reporting and periodic reporting. In most cases, event-triggered reporting is used, meaning that the eNodeB sends predefined measurement control rules to UEs. If the measurement result of a UE meets the threshold requirement in the rules, a Measurement Report (MR) is triggered.

1.3              Intra-LTE Handover Measurement Events

The LTE system sends measurement control rules to UEs. For the details of the measurement control rules, refer to 2.2 Measurement Control. Measurement event types are specified by these rules.

Mobility measurement events in the E-UTRAN system include the following:

1.     Event A1: The serving cell quality is higher than an absolute threshold (serving > threshold). It can be used for stopping an ongoing inter-frequency or inter-RAT measurement and deactivating a gap.

2.     Event A2: The serving cell quality is lower than an absolute threshold (serving < threshold). It can be used for starting an inter-frequency or inter-RAT measurement and activating a gap.

3.     Event A3: The quality of a neighbor cell is offset higher than the serving cell (neighbor > serving + offset). It can be used for coverage-based inter-frequency or intra-frequency handover.

4.     Event A4: The neighbor cell quality is higher than an absolute threshold. It can be used for inter-frequency handover and load-based handover.

5.     Event A5: The serving cell quality is lower than threshold 1 (serving < threshold1), and the neighbor cell quality is higher than threshold2 (neighbor > threshold2). It can be used for coverage-based inter-frequency or intra-frequency handover.

The events are described as follows:

1.      Event A1/A2 is often used for starting and stopping inter-frequency or inter-RAT measurement. Inter-frequency and inter-RAT measurements have a large effect on the current service rate, so the measurements should be started only when necessary. Therefore, event A2 is defined. When a UE determines that the serving cell quality is lower than a threshold, it reports an MR. When inter-frequency or inter-RAT measurement is implemented for a UE, and the signal quality of the UE is enhanced, no measurement is required. In this case, the UE needs to notify the eNodeB of the signal quality, so event A1 is defined. When the eNodeB receives the event A1 MR, it determines that the signal quality is good, and inter-frequency or inter-RAT measurement can be stopped.

2.      Event A3 is most commonly used, and is often used in intra/inter-frequency measurement.

Entering condition for event A3:  

Leaving condition for event A3:  

Where

Mn: neighbor cell measurement result, not including any offset

Ofn: frequency-specific offset of the neighbor cell

Ocn: cell-specific offset of the neighbor cell

Hys: lag between entering and leaving of the event

Ms: local cell measurement result, not including any offset

Ofs: frequency-specific offset of the serving frequency point (frequency point of the local cell)

Ocs: cell-specific offset of the local cell

Off: event A3 offset, which needs to be configured on a higher layer

Figure 1 1 Event A3 Measurement Conditions

As shown in the above figure, Mn strength is raising, and Ms strength is declining. Under this obvious trend, Ofn, Ocn, and Hys are considered for neighbor cell signal strength, and Ofs, Ocs, and Off are considered for serving cell signal strength.

6.     Event A4 means that the neighbor cell quality is higher than an absolute threshold. Event A5 means that the serving cell quality is lower than threshold1 (serving < threshold1) and the neighbor cell quality is higher than threshold2.

Entering condition for event A4: Mn+Ofn+Ocn-Hys>Thresh

Leaving condition for event A4: Mn+Ofn+Ocn+Hys<Thresh

Entering condition for event A5:  and 

Leaving condition for event A5:  and 

1.4              Handover Classification

&  Tips: Handover in LTE

Handovers in LTE are hard handovers.

Intra-LTE handovers refer to handovers within the LTE system, including LTE FDD (or TDD) intra-frequency and inter-frequency handovers, and handovers between LTE FDD and LTE TDD.

Inter-LTE handovers refer to handovers between LTE and other radio system modes (such as UMTS, GSM, and CDMA2000).

Intra-LTE handovers can be classified from various aspects.

Depending on whether the system is changed, intra-LTE handovers can be classified into intra-TD-LTE handovers, intra-FDD-LTE handovers, and handovers between TD-LTE and FDD-LTE.

Depending on whether the eNodeB is changed, intra-LTE handovers can be classified into intra-site handovers and inter-site handovers. Intra-site handovers can be classified into S1 handovers and X2 handovers. For intra-EUTRAN handovers, the system determines whether to use the X2 interface or S1 interface. If X2 interfaces exist between two eNodeBs, the system preferably chooses the X2 interfaces for handover.

Depending on whether the frequency is changed, intra-LTE handovers can be classified into intra-frequency handovers and inter-frequency handovers. Different from 2/3G networks, the inter-frequency handover procedure is implemented between two cells with the same central frequency point but different bandwidth.

Depending on whether the UE is within the same MME, intra-LTE handovers can be classified into intra-MME handovers and inter-MME handovers. In many cities, an MME pool is used to establish the network, meaning that each eNodeB is connected to multiple MMEs, so true inter-MME handovers rarely occur, and are not described in this document.

Figure 1 2  LTE Handover Classification

1.4.1                Intra-Site Handover

Intra-site handovers refer to UE handovers between different cells within the same eNodeB. The intra-site handover procedure is slightly different from the inter-site handover procedure. Intra-site handover preparation messages are transmitted between boards of the eNodeB instead of through the S1 or X2 interface.

The following figure shows the procedure.

Figure 1 3 Intra-eNodeB Handover Procedure

As shown in the above figure, when the UE reports an MR, the eNodeB determines that an intra-site handover is required, and sends a handover request to the target cell. If the target cell is prepared, it notifies the source cell of its resource information, and the source cell sends a reconfiguration message to the UE through the Uu interface, notifying the UE to implement handover. After the handover is completed, it is not required to notify the core network, because no S1/X2 link is involved during the handover process.

1.4.2                S1-Based Inter-Site Handover

The interface between the E-UTRAN and the EPC is defined as the S1 interface. The S1 interface is composed of the S1-MME interface on the control plane and S1-U interface on the user plane. The S1-MME interface is used between the eNodeB and the MME, and the S1-U interface is used between the eNodeB and the S-GW. The following figures show the protocol stacks of the S1-MME and S1-U interfaces.

Figure 1 4 S1 Control-Plane Protocol Stack

Figure 1 5 S1 User-Plane Protocol Stack

The following figures show S1-based handover (S1 handover).

Figure 1 6 S1-Based Handover (a): Before Handover

Figure 1 7 S1-Based Handover (b): After Handover

As shown in the above figures, S1-based inter-eNodeB handover (S1 handover) includes the following:

1.     Interaction between the source eNodeB and the MME through the S1 interface

2.     Interaction between the UE and the source eNodeB through the Uu interface

3.     Interaction between the UE and the target eNodeB through the Uu interface

4.     Interaction between the target eNodeB and the MME through the S1 interface

5.     Data forwarding from the source eNodeB to the target eNodeB

During S1 handover, a direct data forwarding path may exist between the source eNodeB and the target eNodeB. Data forwarding in S1 handover can be classified into direct forwarding and indirect forwarding. In S1 handover, the source eNodeB may have data not sent to the UE. After the handover, the data can be directly or indirectly forwarded to the target eNodeB. Direct forwarding means that the data is forwarded through the X2 interface, which is the interface built on the user plane (data transmission channel), and indirect forwarding means that the data is forwarded through the S1 interface.

Upon receiving an MR and determining that handover is required, the source cell where the UE is located determines whether the source cell and target cell are within the same eNodeB. If they are not within the same eNodeB, the source cell needs to determine the handover type. If the source cell determines that no X2 handover can be implemented, it initiates an S1 handover procedure.

The following figure shows the S1 handover procedure.

Figure 1 8 S1-Based Handover Procedure

 Tips: S1-Based Handover Procedure

A normal S1-based handover procedure includes the following phases:

Handover preparation phase: steps 1–5 in the above figure

Handover implementation phase: steps 6–11 in the above figure

Handover completion phase: steps 12–14 in the above figure

The S1 handover procedure is described as follows:

1.     The UE in RRC connection state sends an MR to the eNodeB. If the measurement event is Neighbour + Offset > Serving (Offset is a negative value) in the message, the target cell does not belong to the local eNodeB, no X2 association exists between the source eNodeB and the target eNodeB, and the source and target eNodeBs belong to the same MME, S1 handover is implemented.

2.     The source eNodeB sends a Handover Required message to the MME through the S1 interface. This message contains the eNB UE X2AP ID allocated by the local eNodeB, MME UE S1AP ID allocated by the MME, and handover type, handover cause, UE capability, and UE security information.

3.     Upon receiving the Handover Required message, the MME sends a handover request to the target eNodeB through the S1 interface. Upon receiving the handover request, the target eNodeB creates a new DCI instance, allocates a new GID and a new eNB UE S1AP ID, and stores UE parameters from the source eNodeB.

4.    The target eNodeB sends a handover response containing UE admission information to the MME.

5.     The MME sends a handover command to the source eNodeB, and establishes a forwarding tunnel.

6.     The source eNodeB retrieves RRC connection reconfiguration information from the handover command, and sends it to the UE.

7.     The source eNodeB forwards data and sends the PDCP SNs of uplink and downlink services to the MME through a UE status message.

8.     The MME sends the PDCP SNs of uplink and downlink services to the target eNodeB through an MME status message.

9.     The UE implements a random access to the target eNodeB.

10.   The target eNodeB permits the access of the UE.

11.   The UE notifies the target eNodeB that RRC connection reconfiguration is completed.

12.   The target eNodeB sends a handover notify message to the MME through the S1 interface, and sends uplink data to the core network.

13.   The MME sends a UE Context Release Command message to the source eNodeB, notifying the source eNodeB to release resources for the UE.

14.   The MME responds with a UE Context Release Complete message, and the source eNodeB releases memory resources and deletes the UE instance.

1.4.3                X2-Based Inter-Site Handover

Figure 1 9 X2 Interface Protocol Stack

The X2 interfaces connect two eNodeBs for signaling interaction. Load or interference information and handover information need to be transmitted between two eNodeBs through the X2 interfaces.

The following figures show X2-based handover (X2 handover).

Figure 1 10 X2-Based Inter-eNB Handover (a): Before Handover

Figure 1 11 X2-Based Inter-eNB Handover (b): After Handover

As shown in the above figures, X2 handover includes the following:

1.     Interaction between the source eNodeB and the MME through the S1 interface

2.     Interaction between the UE and the source eNodeB through the Uu interface

3.     Interaction between the UE and the target eNodeB through the Uu interface

4.     Interaction between the target eNodeB and the MME through the S1 interface

5.     Data forwarding from the source eNodeB to the target eNodeB

The entire procedure involves the source eNodeB and target eNodeB, MME/S-GW, and UE.

Upon receiving an MR and determining that handover is required, the source cell where the UE is located determines whether the source cell and target cell are within the same eNodeB. If they are not within the same eNodeB, the source cell needs to determine the handover type (whether based on the X2 interface or S1 interface). If there is an X2 association between the source eNodeB and the target eNodeB, and the two eNodeBs are connected to the same MME, then X2-based inter-eNodeB handover is implemented. If there is no X2 association between them, S1 handover is implemented.

The following figure shows the X2 handover procedure.

Figure 1 12 X2-Based Inter-eNodeB Handover Procedure

&  Tips: X2-Based Handover Procedure

A normal X2-based handover procedure includes the following phases:

Handover preparation phase: steps 1–3 in the above figure

Handover implementation phase: steps 4–8 in the above figure

Handover completion phase: steps 9–11 in the above figure

The X2 handover procedure is described as follows:

1.     The UE in RRC connection state sends an MR to the eNodeB. If the measurement event is Neighbour + Offset > Serving (Offset is a negative value) in the message, the target cell does not belong to the local eNodeB, an X2 association exists between the source eNodeB and the target eNodeB, and the source and target eNodeBs belong to the same MME, X2 handover is implemented.

2.     The source eNodeB sends a handover request message to the target eNodeB through the X2 interface, which contains Old eNB UE X2AP ID allocated by the source eNodeB, MME UE S1AP ID allocated by the MME, UE capability, UE security, and UE history information, list of E-RABs to be established, and destination address on the core network side for each E-RAB.

Upon receiving the handover request message over the X2 interface, the target eNodeB creates a new DCI instance, allocates a new GID and a new eNB UE X2AP ID, stores UE parameters from the source eNodeB, queries the database to obtain admission parameters, admits the UE, and creates a service bearer channel.

3.     The target eNodeB sends a Handover Request Ack message to the source eNodeB through the X2 interface, which contains the New eNB UE X2AP ID, Old eNB UE X2AP ID, D-eNB admission success or failure information, and a handover command to be sent by the source eNodeB to the UE. If non-competitive handover is implemented, an RACH preamble is contained in the handover command.

4.     The source eNodeB receives the Handover Request Ack message over the X2 interface, retrieves the RRC connection reconfiguration data, and sends it to the UE.

5.     The source eNodeB stops downlink data transmission over the Uu interface and uplink transmission data over the S1 interface, collects uplink and downlink PDCP SNs, and prepares for data forwarding. The source eNodeB sends uplink and downlink PDCP SNs to the target eNodeB through an SN Status Transfer message.

6.    Upon receiving the handover command, the UE synchronizes the command to the target eNodeB, and implements a random access to the target eNodeB. Upon receiving the random access request, the dispatcher starts to work. The target eNodeB returns a Random Access Response message to the UE, which contains uplink authorization and TA information.

7.     For details, refer to Step 6.

8.     The UE sends an RRC Connection Reconfiguration Complete message to the target eNodeB.

9.     The target eNodeB sends a Path Switch Request message to the MME, which contains the MME UE S1AP ID on the source side, eNB UE S1AP ID allocated by the target eNodeB, Switch-required E-RAB information, and UE security information.

10.   The MME sends a Path Switch Request Ack message to the target eNodeB, which contains Path Switch success or failure information. Upon receiving the message, the target eNodeB determines whether the destination address on the uplink transport layer on the core network side is changed. If the destination address is changed, a bearer channel needs to be established.

11.   The target eNodeB sends a UE Context Release message to the source eNodeB through the X2 interface, notifying the source eNodeB to release the resources and delete the UE instance.

The largest difference between the X2 and S1 handover procedures lies in the moment of interaction between the radio access network and the core network. In S1 handover, the source eNodeB establishes a connection with the target eNodeB through the core network. In X2 handover, the source eNodeB directly interacts with the target eNodeB, and notifies the core network after handover.

Data configuration is the same for X2 and S1 handovers. The neighbor cell parameter whether to support X2 handover is set to yes by default. Whether X2 handover or S1 handover is implemented depends on the existence of an X2 association. If there is an X2 association, the system preferably implements X2 handover.

1.4.4                Inter-Frequency Handover

In inter-frequency handover, the frequency point configuration of the source cell is different from that of the target cell. It is considered as an inter-frequency handover if any one of the following conditions is met: (1) the two cells have different frequency bands, (2) the two cells have different central frequency points, and (3) the two cells have the same frequency band and central point but different bandwidths.

The process of inter-frequency handover from the handover decision to the handover completion is the same as that of intra-frequency handover. The measurement phases of the two types of handovers are different. In intra-frequency handover, the system requests the UE to keep intra-frequency measurement from the beginning of call establishment. In inter-frequency handover, inter-frequency measurement is started only after the UE reports an MR of event A2. If there is an ongoing inter-frequency measurement, and the UE reports an MR of event A1, the system requests the UE to stop the inter-frequency measurement.

E-UTRAN inter-frequency measurement is implemented as follows:

1.     Inter-frequency measurement setup

(1)       After event A2 is reported

2.     Inter-frequency measurement release

(1)       After event A1 is reported

(2)       The UE state changes to RRC_IDLE (it is not required to notify the UE to release measurement through a message. The eNodeB releases the measurement).

E-UTRAN intra-frequency and inter-frequency measurements and handovers comply with the following principles:

1.     For intra-frequency measurement and inter-frequency measurement, different A3 configurations and measurement IDs are used.

2.     A new measurement configuration can be manually created to use event A4/A5.

3.     Inter-frequency measurement and measurement gap are started by event A2 and stopped by event A1.

4.     The priorities of intra-frequency and inter-frequency handovers depend on the MR threshold configuration.

5.     When inter-frequency neighboring relationships that the UE supports are configured at the back end, or inter-frequency ANR is enabled, the eNodeB delivers measurement configuration data through an RRC connection reconfiguration message for starting inter-frequency measurement triggered by event A2.

6.     Event A2 reporting: inter-frequency measurement is started and a gap is activated. Upon receiving event A2, the eNodeB sends event A1 measurement configuration, so that the UE stops inter-frequency measurement when the cell quality is good enough. After receiving the MR of event A2, the eNodeB sends event A3-based inter-frequency measurement configuration data to the UE and notifies the UE to start inter-frequency measurement. If the UE detects that the event A3 condition is met, the UE reports an MR for inter-frequency handover. This flow ends.

7.     Event A1 reporting: The UE periodically implements inter-frequency measurement within the gap, until the signal quality of the serving cell is good. If the event A1 condition is met, the UE reports event A1, and the eNodeB notifies the UE to stop inter-frequency measurement and deactivate the gap.

For inter-frequency handover measurement, the eNodeB needs to deliver event A1-based and event A2-based measurement configuration data to the UE. The event A2-based configuration is used for the UE to start inter-frequency measurement, and the event A1-based configuration is used for the UE to stop inter-frequency measurement. When the UE meets the event A2 condition, the UE reports an MR message to the eNodeB, and the eNodeB sends new inter-frequency measurement configuration (based on event A3) to notify the UE to periodically search for other eNodeBs. When the signal strength of another eNodeB and the local cell meet event A3 condition, the UE reports an MR message, and the eNodeB makes a handover decision and starts inter-frequency handover.

For example,

If the thresholds for event A2 and event A1 are -115 dBm and -105 dBm respectively, when the UE detects the RSRP of the serving cell is lower than -115 dBm, event A2 is triggered. The UE reports an MR message to the eNodeB, notifying the eNodeB that the signal quality is poor. The eNodeB sends new measurement configuration (based on event A3) to the UE, which contains the target frequency points to be searched for (maximum: 8), search gap, A3 offset (for example, 3 dB), lag (for example, 0), and TTT (for example, 320 ms). The eNodeB also sends event A1 measurement configuration, where the threshold is -105 dBm. When the UE detects a PCI signal on a target frequency point and determines that the signal strength is 3 dB higher than the PCI signal strength of the serving cell, which lasts for 320 ms, the UE resends an MR message to the eNodeB. The eNodeB makes an inter-frequency handover decision.

If the UE fails to detect any strong inter-frequency signal, and the signal quality of the serving cell becomes good when the UE moves (-100 dBm) and is higher than the event A1 threshold -105 dBm, then event A1 is triggered. The UE reports a new MR message to the eNodeB, and the eNodeB sends new measurement configuration to the UE to stop the inter-frequency measurement (based on event A3), because the inter-frequency measurement affects the service rate on the UE.