Evaluation of Mobility Performance and Deployment Scenarios in UMTS Heterogeneous Networks Wang Min, Edgar Ramos, Y.-P. Eric Wang
Namir Lidian, Sairamesh Nammi, Mark Curran
Ericsson Research Laboratorigränd 11, Luleå, Sweden
[email protected]
Ericsson AB Torshamnsgatan 44, Stockholm, Sweden
[email protected]
Abstract—While a heterogeneous network increases the system performance in terms of coverage and capacity, if it is deployed without being well planned and without making use of optimized configurations, it could potentially lead to tough requirements on the mobility performance. The system level simulations shows that in typical scenarios the small separated cell deployment (each low power node is deployed as a separate cell) meets mobility requirements from the regular network operations perspective. However, in challenging mobility scenarios, small separated cell deployment might suffer from the degradation in mobility performance due to the non-optimal handover settings. The proposed combined cell deployment (low power nodes and macro cell use the same cell identity) provides an effective alternative to improve mobility in terms of reduced handover numbers and improved transmission robustness for handover signaling, therefore simplifying the network planning and management. Keywords—heterogeneous networks; handover; separated cell; combined cell
I.
low
power
node;
INTRODUCTION
Deployment of heterogeneous networks is one of the key approaches to meet the increasing traffic and data rate demands [1]. In a heterogeneous network, a number of low power nodes (LPNs) are deployed in the coverage area of a traditional macro cell. Deploying LPNs in a macro cell allows the radio resources in the macro cell to be reused by LPNs. This achieves what is called cell-splitting gain in terms of the capacity per unit area. LPNs could be deployed on the same frequencies of the macro cell, or on separate frequencies. This paper only discusses the case where both LPNs and macro cells are deployed on the same frequency. In the UMTS heterogeneous network, each LPN can be deployed as a separated cell namely small separated cell deployment (also called co-channel deployment). In this deployment each LPN has a different cell identity. Hence each LPN creates a new cell with a unique primary scrambling code (PSC) as illustrated in Fig. 1. Each cell transmits individual primary common pilot channel (P-CPICH), high speed-shared control channel (HS-SCCH) and high speed-physical downlink shared channel (HS-PDSCH) channels. The conventional handover will be triggered when UEs moves between macro cell and LPN.
Fig. 1. Small separated cell deployment. With the separated cell deployment, UE inevitably experiences more handovers due to the deployment of LPN cells. This could impose the challenges on the mobility performances in the highly loaded system. As an alternative, we propose another deployment option named as combined cell deployment. In this deployment, LPNs and the macro cell have the same layer 3 (L3) cell identity (ID) with a unique PSC code as shown in Fig. 2. The macro cell and all associated LPNs are referred to as a combined cell. Since all the nodes are part of the same cell no handover is required when UE moves within a combined cell. Each LPN transmits the same P-CPICH as macro cell. More details are available in section III.
Fig. 2. Combined cell deployment. A LPN typically has a lower transmit power level and a low antenna height than those of a macro node. This results in a higher possibility that the LPN signal may be shadowed by a blocking object, giving rise to challenges in handover scenario. For example, a user may quickly lose the LPN signal as it turns a street corner. Furthermore, deploying LPNs as separated cells in a macro coverage area also leads to a higher number of handovers as the user moves. Hence, it is anticipated to be challenging to meet mobility requirements in heterogeneous networks, especially when the UE is moving at high speed. The UMTS mobility procedures includes the serving cell change (SCC) in releases before 3GPP REL-8 [2], and enhanced serving cell change (ESCC) as defined in 3GPP REL-8 [3].
978-1-4799-4482-8/14/$31.00 ©2014 IEEE
There are many papers concerning the mobility performance in UMTS networks with macro node deployment. In [4], the performance of the SCC procedure for delay sensitive services like Voice over IP (VoIP) is evaluated. The simulation results show that the SCC procedure, with proper parameter settings, can be fast enough to accommodate VoIP users traveling at the speed of 120 km/h with an acceptable conversation. In [5], it was shown that while a relatively low signaling radio bearer (SRB)/handover error rate is experienced in a macro cell scenario, in a more challenging Manhattan scenario the error rate is higher. The ESCC procedure results in similar HO failure rates in the Manhattan scenario as the SCC procedure in the macro cell scenario, when the static power allocation for HS-SCCH is in use.
from the target cell is typically more robust and faster than RBR message which is transmitted from the source cell.
There are few studies focusing on the mobility performance in UMTS heterogeneous networks. This paper studies the mobility performance in a UMTS heterogeneous network in both the small separated cell deployment and the combined cell deployment. This paper is organized as follows. In Section II, UMTS handover procedures are described. In Section III, the concept of combined cell is described in details. Section IV describes the simulation models and the performance metrics used in our study. Section V presents the detailed simulation results in the two different scenarios. The final conclusions are summarized in Section VI. II.
UMTS HANDOVER PROCEDURES
In the SCC procedure, the serving cell change message is received only from the source cell. The UE performs the measurements of the neighbor cells periodically and sends report(s) when the measurements satisfy the triggering criteria (hysteresis and time to trigger (TTT)) for different events. The typical handover measurement events include events 1A, 1B, 1C and 1D, which are defined for the triggering of the operations including Active Set (AS) addition, AS deletion, AS replacement and the serving cell change respectively. Upon the reception of the measurement reports, the network typically responds with a command message (the radio bearer reconfiguration (RBR) message or active set update (ASU) message), instructing the UE to perform the operations accordingly. The ESCC procedure was standardized in REL-8. The detailed signaling exchange flow is illustrated in Fig. 3. The UE performs the pre-configuration of the target cell by receiving the target cell information in ASU message (normally AS addition message), which is prior to the reception of the RBR message for the serving cell change. Therefore, the UE can listen to one downlink control channel, namely high speedshared control channel (HS-SCCH) in the target cell which carries the indication of the serving cell change order. As soon as the UE triggers event 1D report, it will prepare for the reception of RBR message from the source cell and L1 HSSCCH order from the target cell. The HS-SCCH order indicates the readiness of the target cell for the serving cell change. The UE may receive only the HS-SCCH order depending on the configuration. Thus, the ESCC would be more robust and faster than SCC since the layer one signaling
Fig. 3. Signaling procedure for ESCC.
III.
PROPOSED CONCEPT OF COMBINED CELL
The combined cell is an alternative deployment option for LPNs in heterogeneous networks. One of its advantages is that it simplifies the cell planning and management since there are fewer cells defined in the combined cell deployment compared to the small separated cell deployment. In a combined cell deployment, the uplink (UL) transmitted signal is received by all the nodes. Hence the received signals can be combined at the antenna level. The higher uplink capacity due to macro diversity can be expected than that of Macro only and small separated cell deployments. This is also anticipated to improve the transmission robustness for measurement report message in UL. In downlink (DL), there are mainly two types of transmission modes, Single Frequency Network (SFN) mode and Node Selection with Spatial Reuse (SR) mode [6]. The selection of the transmission mode and (or) transmission node is controlled by the central scheduler. Hence in one transmission time interval (TTI) the UE can be in SFN mode, while in the next TTI the UE can be SR mode. Note that no higher layer signaling is needed to indicate the transmission mode change. In the SFN mode, all the nodes in the same combined cell transmit exactly the same information on the pilot channel, downlink control channel and downlink data channel, using the same carrier frequency, spreading and scrambling codes in the downlink radio channels. In the SR mode, the central node needs to identify which node is most suitable for a particular UE for downlink transmission. For this purpose, a new probing pilot called Fractional CPICH (F-CPICH) for node selection and a new demodulation pilot called demodulation CPICH (D-CPICH) are introduced. Each LPN has its separate F-CPICH and D-CPICH, which are different from those used by the macro base station. The FDPCHs are transmitted from different LPNs in a time multiplexing way. D-CPICH is transmitted only when the data transmission is launched. The deployment of both new pilot
channels is illustrated in Figure 4. A UE periodically sends measurement reports of the F-CPICHs from the neighboring LPNs which are used as inputs for the central scheduler entity to select the most suitable transmission node. The selected transmission node transmits D-CPICH and data on its own HSPDSCH channel to the UE (HS-SCCH is also transmitted on its own channel). The UE can use the received D-CPICH as a reference and perform the HS-PDSCH reception accordingly. With the SR mode, the system resources are utilized more efficiently than with the SFN mode. Hence it is beneficial for achieving a good system capacity.
deliver handover signaling. For the sake of simplicity, we have only investigated the mobility performance when UE is in the SFN mode since it is expected to show the best mobility performance. B. Evaluation metrics A handover failure is defined as below. For SCC procedure: if a UE fails to receive the RBR message after event 1D is triggered or the ASU message after event 1A or 1C is triggered For ESCC procedure: if a UE fails to receive the RBR from the source cell and fails to receive the HS-SCCH order from the target cell after event 1D is triggered, or the UE fails to receive the ASU message after event 1A or 1C is triggered. Handover failure ratio is defined as the number of handover failures divided by the number of handovers including both successful handovers and failed handovers. V.
Fig. 4. F-CPICH central node determination in combined cell.
The UE does not perform handover when it moves within a combined cell. The handover from the source combined cell to the target combined cell will be triggered only when the measurements based on the combined P-CPICH from the target combined cell satisfy the 1D event criteria. In locations with bad coverage, using the SFN mode, i.e., transmitting the handover signaling over multiple nodes in parallel will guarantee a robust reception. Otherwise, the SR mode can be chosen for handover signaling based on F-CPICH measurements as described above. The selection of the transmission nodes and transmission modes can be switched on a TTI basis. Hence, the transmission node or transmission modes could even be different for different RLC transmission attempts of the same HO signaling message. For example, the first transmission attempt could use spatial reuse in a TTI, and the next retransmission attempt might use the SFN mode involving multiple nodes for transmission. An example for the mode and node selection is illustrated in Fig. 5.
MOBILITY PERFORMANCE EVALUATIONS
The two different scenarios are evaluated. In the first scenario, the system is medium loaded (50% of the system resources, e.g., power, are utilized) by properly selecting the number of FTP UEs in each cell. We refer to this as a typical network scenario. The performance for small separated cells in the same frequency deployment is first investigated in this scenario as a reference case. In the second scenario, the system is highly loaded. This scenario is modeled in a simple way that the system resources are almost all the time fully utilized. Every node always transmits at a full power even in the TTIs when there is no available data for transmission. The power will be transmitted as an overhead. We refer to this scenario as a challenging mobility scenario. A. Simulation Setup TABLE I SIMULATION PARAMETERS Variables Cell Layout
21 cell hexagonal , ISD 500m
LPNs per Macro cell
1, 4 Minimum distance between LPN and macro cell: 75m Minimum distance between LPNs: 40m LPNs are randomly and uniformly distributed within a macro cell satisfying the distance requirement. The minimum distance between UE and macro cell is 35m; The minimum distance between UE and LPN is 10m UE randomly and uniformly distributed within a macro cell
LPN deployment
UE deployment
Fig. 5. Switch transmission node or transmission modes on TTI basis.
IV.
SIMULATION MODELS AND METRICS
A. Simulation models As discussed in section III, the central scheduler in the macro node can apply either the SR mode or SFN mode to
Value
Scenario definition
Speed
Typical scenario (medium loaded): 126 ftp users, 267 KB file size, 10s mean reading time Challenging mobility scenario (full load): 126 ftp users, 267 KB file size, 10s mean reading time, the full transmission power in the antennas when no HS transmission 10, 30, 60, 90 and 120 km/h
Channel Propagation model Node B Max Tx Power SRB transmission Cell individual offset (CIO)
SHO available, max active set size 3 R1a (reporting range constant) = 4.5dB R1b (reporting range constant) = 4.5dB TTT: 1A:320ms, 1b:640ms, 1c:320ms, 1d:640ms Hysteresis: 1a:0dB, 1b:0dB, 1c:1dB, 1d:1dB PA 3GPP case 1, no LOS component with shadow fading Macro Node: 43dBmm, LPN: 37 dBm SRB over HS 0dB for Macro, 3 dB for LPN
The main parameters and assumptions for the performed evaluations are summarized in TABLE I, which are mainly according to mobility simulation assumptions for UMTS heterogeneous networks (see Table 72) [7]. B. Performance in typical scnearios with the deployment of small separated cells (medium load) Typically, a handover failure ratio of 4% is defined as an acceptable level by network operators in a real network environment. According to that, SCC cannot guarantee a robust handover at higher speeds in heterogeneous networks when more LPNs are deployed, as shown by Fig. 6, while ESCC achieves a better handover performance than SCC. The improvements are mainly due to the reduced delay in the handover procedure by the pre-configuration of the target cell and the reception of the L1 HS-SCCH order from the target cell when the 1D event is triggered. HO failure ratio Ratio of failed HOs [%]
8
6
SCC Macro only SCC,1 LPN SCC,4 LPN ESCC,4 LPN
4
2
0 0
50 100 UE moving speed [km/h]
Fig. 6. Handover failure ratio in separate cell deployment. (medium load)
As shown in Fig. 7, with the SCC handover procedure most of the handover failures are due to the transmission failure of the serving cell change command, i.e., RBR signalling message. There are more failed handovers due to active set (AS) addition command than AS replacement command. When the AS addition command is triggered, there are typically fewer UL radio links included in AS than the case when AS replacement command is triggered. Consequently, the probability that the network does not receive the UL RLC acknowledgement in time for the DL ASU message will decrease in case of AS replacement command transmission.
With ESCC, most of the handover failures due to the unsuccessful transmission of the serving cell change command, RBR message are avoided, since in the majority of the cases the handover command signalled by a HS-SCCH order from the target cell increases the probability of a successful reception. 25 Failed HOs per UE and per hour
Soft Handover Parameters
20
SCC,4 LPN ESCC,4 LPN
serving cell change fail
15 10 AS addition fail
5 AS replacement fail
0 0
1
handover complete fail
2 3 4 5 HO failure reasons [60 km/h]
6
Fig. 7. Handover failure reasons in separate cell deployment. (medium load)
C. Performance in challenging mobility scenario (full load) Only the ESCC handover procedure is considered in this section. Based on the evaluation results for typical scenarios, the SCC performance is not expected to match the ESCC performance in this scenario. In Fig. 8, it is observed that UEs in the combined cell deployment experience fewer handovers than separate small cells deployment because handovers between LPNs and the Macro cell are removed. The combined cell case has also fewer handovers compared to the macro only case. It has been observed from our simulations that the shadow fading is the main reason. In the combined cell deployment, the handover measurements in the UE are executed based on the combined P-CPICH signal received from all nodes in the combined cell. This provides a better node diversity for the reception of the PCPICH signal than the macro only scenario since it is not often that P-CPICH signals from all nodes are attenuated due to the shadow fading at the same location. Hence, the impact of the shadowing fading on handovers is mitigated, which results in fewer handovers. In Fig. 9, it can be observed that the separate small cell cases show high handover failure ratio with 4 LPNs deployment. The increase in handover failure compared to that of the typical scenario is mainly due to the high load in the system. When the system has higher load and with fast moving UEs, the deployment of the small separated cells need to tune handover settings in order to meet the mobility requirements from the typical network operation aspect. Our simulations are based on suboptimal handover settings and random placements of LPNs in the macro cell. Hence, the simulation results indicate the necessity for good cell planning and management for small separated cell deployment in challenging mobility scenarios.
degrade fast, especially when the UE moves from LPN at a high speed. However, L2M and L2L handovers are removed in combined cell deployment. Hence, a low handover failure ratio is observed in combined cell deployment, as seen in Fig. 11. The low handover failure ratio in combined cell also thanks to the lower handover numbers and the better reception of the signalling due to the antenna diversity in SFN mode.
Macro only 1 LPN,separate cell 4 LPN,separate cell 1 LPN,combined cell,SFN 4 LPN,combined cell,SFN
1200 1000 800 600
Failed HOs per UE and per hour
Total HO number per UE per hour
1400
400 200 0 0
50 100 UE moving speed [km/h]
150
Fig.8. Handover number in separate cell and combined cell deployments (full load)
Ratio of failed HOs [%]
25 20
Macro only 1 LPN, separate cell 4 LPN, separate cell 1 LPN, combined cell, SFN 4 LPN, combined cell, SFN
1 LPN, separate cell 4 LPN, separate cell
6
4
2
0
M2M
VI.
10 5
50 100 150 UE moving speed [km/h] Fig. 9. Handover failure ratio in separate cell and combined cell deployments in challenging scenario. (full load)
25 1 LPN, separate cell 4 LPN, seperate cell
20
[1] AS replacement fail
[2] serving cell change fail measurement report fail
L2L
CONCLUSIONS
REFERENCES
10 5
L2M
This paper evaluated the mobility performance in heterogeneous network with different deployment options. In the typical network scenarios, the simulation results indicate the small separated cell deployment option can provide a satisfactory mobility performance, e.g., less than 4% of handover failure ratio. However, in challenging mobility scenarios with the small separated cell deployment, the handover settings would require to be well-tuned, and the placement of LPNs needs to be well-planned to meet the mobility requirements. The proposed concept of combined cell offers an improved handover performance because of the experienced gains in terms of both the reduced handover number due to a better node diversity and improved transmission robustness for handover signaling messages.
AS addition fail
15
M2L
Fig. 11. Handover failure type in separate cell deployment (full load)
0 0
Failed HOs per UE and per hour
8
HO failure types [60 km/h]
15
0
10
handover complete fail
1 2 3 4 5 HO failure reasons [60 km/h]
Fig. 10. Handover failure reasons in separate cell deployment. (full load)
In Fig. 10, it is observed that there are more handovers failed due to the transmission failure of ASU messages than that in typical scenarios (see Fig. 7). In challenging mobility scenarios, handover failures occur more often when UEs move from LPN to Macro (L2M) and from LPN to LPN (L2L). It is mainly due to that the radio connection to the LPN may
[3] [4]
[5]
[6] [7]
S. Landström, A. Furuskär, K. Johansson, L. Falconetti, and F. Kronestedt, “Heterogeneous networks increasing cellular capacity,” Ericsson Review, No. 1, Feb. 2011. 3GPP TS 25.308 version 7.12.0, “High speed downlink packet access (HSDPA), overall description, stage 2” 3GPP TS 25.308 version 8.11.1, “High speed downlink packet access (HSDPA), overall description, stage 2” S. Wager and K. Sandlund, “Performance evaluation of HSDPA mobility for Voice over IP,” in Proc. IEEE Veh. Technol. Conf., Spring 2007, pp. 809-813. T.Nihitlä, K. Aho, and I. Repo, “On serving cell change reliability in HSDPA network,” in Proc. IEEE Wireless Commun. and Networking Conf., Apr. 2009, pp. 2068-2072. R1-132603, “Overview of Spatial Reuse Mode in Combined Cell Deployment for Heterogeneous Networks”, 3GPP document, 2013. 3GPP TR 25.800, v2.0.0; “Technical Specification Group Radio Access Network; Study on UMTS heterogeneous networks (Release 12),” August 2013.
