LTE handovers

LTE is designed to provide seamless mobility. UE handovers occur with imperceptible delay and an acceptable number of packet losses which can be recovered by the upper layers. In contrast to HSPA there is no centralised controller in LTE [46]. All centralised functions are performed at the eNB. Therefore soft handover is not possible for LTE. Whenever UE performs a handover, buffered data in eNB has to be forwarded to the corresponding cell in eNB. Data protection during the HOs is handled by the PDCP layer. UE mobility can be categorised into intra-LTE mobility and inter-RAT (Radio Access Technologies, such as UMTS and CDMA200) mobility. During this work only intra-LTE mobility is considered, which occurs either between eNBs or between cells. Based on this, handovers within E-UTRAN can be also categorised into two types: inter-eNB handovers which occur between two eNBs and intra-eNB handovers which occurs between cells within the same eNB. In both cases the main data handling is performed by the eNB. For the first case, the buffered data in the source eNB, where the UE is currently located has to be forwarded via the X2 interface to the target eNB where the UE is moving to. In this case data should be protected and a fast transmission between two eNBs should be executed. Not only the data but also the load or interference related information also has to be forwarded to the target eNB. In the latter case, the eNB handles the data and signalling information within its own entity and forwards data to the target cell from the source cell based on the new UE connectivity.

2.4.5.1 Inter-eNB handovers

As mentioned above, inter-eNB handovers occur between two eNBs which are called source eNB and target eNB. Since data is forwarded over the X2 interface during HOs, the X2 protocol architecture has to be considered in detail. The X2 user-plane protocol stack is given in Figure 2-22 according to the 3GPP specification [47].

Figure 2-22: X2 protocols and inter-eNB HOs

The PDCP layer takes care of the data protection and the GTP establishes a tunnel through the transport network between source and target eNBs. The transport network guarantees the QoS aspects and provides the required priority to the forwarded data. To minimise the packet losses, data forwarding is performed on a per bearer basis. Minimising delay for forwarding data is required for the real time applications such as VoIP

whereas a very low loss probability is required for best effort applications such as email and web browsing which mainly use the TCP protocol. In the context of an HO, the “lossless”, means a very low loss probability and is performed by the PDCP layer which can use an ARQ scheme. Further, PDCP layer also provides in-sequence delivery functionalities for forwarded data. Once the UE has established the connection with the target eNB, the MME is notified for path switching procedures by the target eNB. In this case, the Late Path Switching (LPS) or the backward handover is used to minimise the losses and also minimise the interruption during HOs.

Figure 2-23: starting phases of the inter-eNB handover process

As shown in Figure 2-23, first the UE is connected to the source eNB (SeNB) and they communicate with each other. Once UE reaches the tracking area of the target eNB, it initiates an HO by informing the MME via the source eNB. Before the UE disconnects from the source eNB, internal bearers are set up between the target eNB and the source eNB. This is done using inter-eNB signalling over the X2 interface. Then the UE starts a new connection process with the target eNB and meanwhile all data including unacknowledged data is buffered and newly arriving data is forwarded to the target eNB.

The target eNB buffers the forwarded data until the UE finalises the new connection process with the target eNB. The Figure 2-24 shows that the buffered data of the source eNB is forwarded and then incoming data to the source eNB is also forwarded via the X2 interface.

As mentioned above, all data received from the source eNB via the X2 interface is stored in the target eNB PDCP buffer sequentially. Once the UE has completed the connection process with the target eNB, the buffered data in the latter is transmitted immediately to the UE and all incoming data is

transmitted to UE afterwards. For the uplink, during the UE interruption period, the UL data is stored in the UE PDCP layer and after the new connection is completed with the target eNB, the UE immediately starts a communication with the target eNB and sends the data via the S1 to the respective end nodes.

Figure 2-24: Data forwarding during HO over process via X2 interface

For DL path switching, information about the new UE connection to the target eNB is sent to the S-GW via the MME. This is the reason why this method is called late path switching. Once the path is switched to the target eNB, new data arriving from the PDN uses the new path via the S1 interface.

Figure 2-25: Path switched procedures and last stage of HO activities

To indicate path switching functionality to intermediate nodes, an end- marker PDU is used. The last PDU before the path switch is the end marker PDU which indicates the end of the transmission via the source eNB. When eNB receives the end-marker it assumes that the handover is completed and no further data will come along this route. Then the source eNB releases the resources for that UE and passes the end mark transparently to the target

eNB. The complete HO procedure according to the 3GPP specification is given in the flow chart in Figure 2-26. The red arrows show the control plane signalling whereas the blue arrows show the user plane messages and data handling. There are 18 steps which are shown in the flow chart with different signalling messages.

At step 1, the source eNB configures the UE measurement procedures according to the area tracking information whereas in step 2, the UE is triggered to send a Measurement Report. The source eNB makes the decision to handover UE to the target eNB based on the received Measurement Report at step 3. Next, the source eNB issues a Handover Request message to the target eNB which passes necessary information to prepare the handover at the target eNB. At step 5, Admission Control will be performed by the target eNB dependent on the received radio bearer QoS information and the S1 connectivity to increase the likelihood of a successful handover.

At step 6, the target eNB prepares the handover with L1/L2 and sends a

Handover Request Acknowledge message to the source eNB. This message

includes a transparent container which includes the new C-RNTI and the value of the dedicated preamble to be sent to the UE as part of the Handover

Command. The source eNB sends an RRC Handover Command message

towards the UE at step 7. Next the SN Status Transfer message is used to transfer PDCP layer information from the source eNB to the target eNB to ensure UL and DL PDCP SN continuity for every bearer that requires PDCP status preservation. Afterwards, DL data forwarding is started from the source eNB to the target eNB.

After the UE receives the Handover Command which is sent by the source eNB to perform the handover immediately to the target eNB. When the UE has successfully accessed the target cell, it sends the Handover Confirm message to the target eNB to indicate that the handover procedure is completed for the UE and starts sending downlink data forwarded from the source eNB to UE, and UE can begin sending uplink data to the target eNB as well.

Figure 2-26: Inter-eNB handover procedures

At step 12, the target eNB sends a Path Switch Request message to the MME to inform it that the handover has been completed. The new TNL information is then passed to the MME and to the target eNB in step 16. The resources at the S-GW are then released some time after step 16.The MME sends a User

Plane Update Request message to the S-GW. The S-GW switches the

downlink data path to the target eNB and sends a User Plane Update

Response message to the MME to confirm that it has switched the downlink

data path. The MME confirms the Path Switch Request message with the

Path Switch Request Ack message. By sending a Release Resource message,

the target eNB informs the source eNB of the success of the handover and triggers the release of resources.

2.4.5.2 Intra-eNB handovers

Intra-eNB handovers are performed among the cells within the same eNB. In this case, there is no signalling with EPC and the HO procedures are very simple compared to inter-eNB HOs and shown in Figure 2-27.

Figure 2-27: Intra-eNB handover procedures

All HO procedures are completely done within the eNB. The HO procedure from one cell to another in the same eNB is shown Figure 2-27. Based on the measurement report received from the UE, the eNB takes the decision for

handover to the target cell. Admission Control will be performed in the target cell and the resources are granted by setting up the required radio bearers for that UE. During this HO process DL data is buffered at eNB and UL data is buffered at the UE. Once the HO process is successful, the UE sends a

Handover Confirm message to the eNB to indicate that the handover

procedure is completed. Then both the UE and the eNB immediately start transmitting data for UL and DL respectively.

3 HSPA Network Simulator

The implementation of a real world system is a time consuming and costly working process. Still, after implementing such a complex system, it might not provide the expected behaviour or performance at last. This can result in wastage of resources and huge financial downturns for any institution. For this reason, these systems have to be realised with a low probability of risk. Therefore, before the real implementation, there should be ways to estimate the risks of behaviour and performance of such systems. To fill this gap, imitation of such complex systems can be modelled in simulations and also analysed with respect to the focused requirements. Understanding the requirements and the objectives are the key elements of designing a simulation model. A model cannot be more accurate than the understanding of the corresponding real-world system. Many approaches of such as simulation and analytical considerations are introduced to cater for such real world scenarios. The analytical approaches are more theoretical and use mathematical modelling. They are efficient in analysing simplified models. However, today real world systems are very complex having many dependent and independent parameters and cannot be easily described using analytical models. Therefore, in order to analyse the performance and evaluate behavioural aspects, computer based simulation models are introduced. Such computer based simulation models can realise many complex systems and can also evaluate the performance in a wider scope. However depending on the complexity, these systems require longer processing time and longer implementation time than some analytical approaches require.