5G Handover

Basic handover procedure in 5G. Source: Park 2021, fig. 1.
Basic handover procedure in 5G. Source: Park 2021, fig. 1.

Handover is an essential feature/procedure in 5G (and earlier cellular systems). When a mobile moves from one cell to another in connected mode, it's handover that allows the mobile to stay connected. Signalling within the network ensures that context information is suitably transferred from the source cell to the target cell.

In connected mode, UE makes regular measurements of neighbouring cells. It reports these to the network. It's always the network that decides if the UE should be handed over. There are many types of handover depending on the type of source and target nodes, and the support for specific network interfaces.

A good handover strategy makes efficient use of radio and network resources, while also minimizing service interruption time during handover. Since the early 2020s, ML approaches are being researched for optimizing handovers.

Discussion

  • Could you explain a typical handover procedure?
    Basic intra-AMF, intra-UPF handover. Source: ETSI 2023d, fig. 9.2.3.2.1-1.
    Basic intra-AMF, intra-UPF handover. Source: ETSI 2023d, fig. 9.2.3.2.1-1.

    The figure assumes source and target gNBs are interconnected via the Xn interface. Source gNB decides to handover. Handover can be divided into three phases:

    • Handover Preparation: The target system is prepared to receive the UE. Target gNB does admission control based on slice information. If UE is admitted, target gNB informs source gNB along with an RRC message that's to be sent transparently to the UE.
    • Handover Execution: Source gNB sends handover command to UE. UE disconnects from the source gNB and connects to the target gNB. Any data that source gNB receives during this period is forwarded to the target gNB. Source gNB sends SN Status Transfer message to target gNB. This relates to PDCP SN.
    • Handover Completion: UE sends uplink data on target gNB. For downlink data, target gNB sends Path Switch Request to AMF, which triggers UPFs to update paths to target gNB. Once AMF acknowledges that path switch is done, target gNB informs source gNB to release UE context.

    For DAPS Handover, UE disconnects from source cell much later. Items 7a, 8a and 8b are for DAPS only.

  • What conditions trigger handovers in 5G?
    Handover triggered due to changing radio condition. Source: Haghrah et al. 2023, fig. 3.
    Handover triggered due to changing radio condition. Source: Haghrah et al. 2023, fig. 3.

    Handover can be triggered due to changing radio conditions. UE's connection with the source cell may weaken. Once a certain threshold has been reached, the network decides to handover the UE to another cell that has better radio conditions. The network knows these conditions based on measurement reports sent by the UE. The figure shows A3 event for which the target cell is better than the source cell by a certain threshold for at least TTT duration.

    It might happen that a certain gNB is highly loaded. For load balancing purpose, a handover can be triggered.

    Sometimes a specific service may be unavailable in the source cell or system. If such a service is requested, a handover may be performed. An example is a voice call. Initial 5G deployments may not support VoNR. Hence, request for a voice call triggers a handover to 4G (EPS Fallback). There's also 5G-SRVCC (5G NR to 3G UTRAN handover) but this is triggered due to radio conditions.

  • How is handover different from redirection or cell reselection?
    Illustrating handover, redirection and reselection. Source: Elhady 2023.
    Illustrating handover, redirection and reselection. Source: Elhady 2023.

    When a UE in RRC_CONNECTED state "moves" from one cell to another and continues in connected mode, it's called handover. Active calls in the source cell continue in the target cell after the handover. Instead, if the UE moves into idle mode in the target cell, it's called redirection. This can happen for example if the source cell is being shutdown for power savings; or if N26 interface is not available to support 5GC-to-EPC handover.

    Reselection is an idle mode mobility procedure, specifically when UE is in RRC_IDLE or RRC_INACTIVE states. If the UE finds a cell better than the currently camped one, it triggers reselection. Whereas handover and redirection are initiated by the network in connected mode, reselection is initiated by the UE. If UE is camping for the first time, say after power on, it's called PLMN, SNPN or cell selection.

  • What are the different types of 5G handovers?
    Different types of 5G handovers. Source: TelcoSought 2021.
    Different types of 5G handovers. Source: TelcoSought 2021.

    We note the following types of handovers:

    • Intra-gNB: UE moves from one cell/DU to another while anchored at the same CU.
    • Xn-Based Inter-gNB: Source and target gNBs are connected via the Xn interface. Signalling and data packet forwarding are coordinated within the NG-RAN.
    • N2-Based Inter-gNB: If source and target gNBs lack the Xn interface, then handover is coordinated via the AMF. N2 interface connecting gNB to AMF is used. If source and target gNBs are connected to two different AMFs, then N14 interface connecting those AMFs is also used. The two AMFs can belong to different operators or PLMNs.
    • Inter-System: UE moves from gNB to eNB (5GS-to-EPS handover) using the N26 interface that connects 5GC AMF to EPC MME. The same interface is used for EPS-to-5GS handover.
    • Inter-RAT: UE moves from NR to E-UTRA while staying on 5G Core.
    • NSA/MR-DC: In NSA/MR-DC configurations, handover can happen for master and/or secondary nodes.
    • Non-3GPP Access: UE moves from 5GS to trusted or untrusted non-3GPP access, and vice versa.
  • Could you compare Xn-based and N2-based handovers?
    Call flows of Xn-based and N2-based handovers. Source: Adapted from Chandramouli et al. 2019, sec. 5.4.1.
    Call flows of Xn-based and N2-based handovers. Source: Adapted from Chandramouli et al. 2019, sec. 5.4.1.

    The figure shows both Xn-based and N2-based handovers. In both cases, handover is initiated by the source gNB, which sends HandoverCommand to UE. At the end, if there's been a UPF reallocation, sessions in old UPF are deleted. There's more signalling (and delay) involved with N2-based handover.

    We now point out the main differences:

    • HandoverCommand message is generated at target gNB and is sent to UE via source gNB (Xn) or AMF (N2).
    • If there's UPF reallocation, sessions are created for the new UPFs. Downlink tunnel information is updated once target gNB is ready for handover. With Xn, these happen only after UE has moved to target gNB.
    • Data forwarding is direct for Xn but could be direct or indirect for N2. For indirect data forwarding, downlink data from UPF to source gNB is forwarded to target gNB via the same UPF.
    • After UE handover, target gNB sends Path Switch Request (Xn) or Handover Notify (N2) to AMF. These messages include downlink tunnel information. If there's been UPF reallocation, new UPF becomes the anchor.
  • Could you explain handovers for NSA/MR-DC configurations?
    NSA option 3 handovers. Source: 5GWorldPro 2023.
    NSA option 3 handovers. Source: 5GWorldPro 2023.

    The figure shows a UE moving from left to right across the coverage areas of two LTE eNBs (master nodes) and two 5G NR gNBs (secondary nodes). In 5G Non-Standalone (NSA) deployment scenario, option 3 is used. UE uses eNB as the master, gNB as secondary and EPC as the core network. The fact that the UE is connected to both eNB and gNB implies a Dual Connectivity (DC) configuration. In this example, EN-DC is being used.

    A secondary node can be added, changed or removed while master stays the same. It's also possible to handover the master without changing the secondary nodes; or handover both. We can handover from DC to single connectivity, and vice versa. These combinations increase the complexity of handover decisions.

    NSA/EN-DC helps reduce frequent inter-RAT handovers. Without NSA, inter-RAT handover would happen when a UE moves out of 5G coverage. However, inter-RAT handovers are likely in initial 5G deployments where VoNR is not supported. EPS Fallback, RAT Fallback and 5G-SRVCC are inter-RAT handover procedures.

  • How does gNB disaggregation help handovers?
    gNB CU/DU split optimizes handovers. Source: 5G Networks 2020.
    gNB CU/DU split optimizes handovers. Source: 5G Networks 2020.

    A gNB can be disaggregated into Radio Unit (RU), Distributed Unit (DU) and Central Unit (CU). A DU can control one or more cells. Many DUs can be under the control of a single CU. Virtualized implementations are termed vDU and vCU. This sort of disaggregation avoids frequent inter-gNB handovers. A UE is handed over from one DU to another while anchored at the same CU. PDCP and RRC layers are at the CU. Therefore, there's no additional signalling towards the core network. No data forwarding is needed.

    Intra-gNB-DU handover occurs when UE moves from one cell to another within the same gNB-DU. Inter-gNB-DU handover occurs when UE moves from one gNB-DU to another within the same gNB-CU. For EN-DC operation, MCG SRB or SCG SRB (SRB3) carries the signalling for inter-gNB-DU mobility. With MCG SRB, conditional handover of SCG PSCell is possible.

  • What are the main durations associated with 5G handovers?
    Handover simulation result with NetSim. Source: Tetcos 2021, slide 6.
    Handover simulation result with NetSim. Source: Tetcos 2021, slide 6.

    The figure shows throughput over time during a handover. Throughput is directly related to signal quality. Handover Interruption Time (HIT) is that time period when UE is unable to send or receive any data. It starts when UE receives the handover command. In real networks, this is 30-60ms but it could be hundreds of milliseconds. For inter-beam mobility within the same cell, HIT is shorter.

    Time-to-Trigger (TTT) pertains to handover initiation. Low TTT implies the system quickly reacts to poor signal quality at the UE and triggers a handover. Low TTT might lead to too many handovers (ping-pong behaviour), particularly in ultra-dense networks. High TTT may lead to handover failures. TTT could be adaptive. It can be optimized alongside Hysteresis Margin, A3 offset, and L1 and L3 filter coefficients.

    Timer T304 is started when handover is started. Handover should be completed before this timer expires. If not, we've a handover failure.

  • What are the variations of the basic handover procedure?

    Between the UE measuring a channel and receiving the handover command, target cell channel conditions could've worsened. This is likely for high-mobility scenarios or mmWave channels. Conditional Handover (CHO) solves this by allowing the UE to acquire latest measurements even after receiving the handover command. Handover happens only if conditions are better than a threshold. If not, UE stays on the source cell. If UE loses source connection, it can handover even if the condition is not met.

    To achieve zero HIT, Dual Active Protocol Stack (DAPS) Handover allows UE to maintain two protocol stacks, one for the source cell and one for the target cell. UE sends/receives on both cells. Target cell signals the UE to release source connection. UE's PDCP layer reorders and deduplicates packets.

    In 4G/LTE, there's also Make-Before-Break (MBB) Handover. UE releases source connection only after connecting on the target cell. Unlike DAPS, UE releases source connection on its own. RACH-less Handover is another optimization in 4G/LTE. On the source cell, UE is signalled timing adjustment and pre-allocated uplink grant for the target cell. Without such a grant, UE monitors target cell PDCCH.

Milestones

Dec
2017

3GPP approves the first specifications for 5G, called "early drop" of Release 15. This includes the Non-Standalone (NSA) option. The "main drop" happens in June 2018 with support for Standalone(SA) option. Basic handovers for NSA and SA are supported in Release 15 including inter-gNB-DU (EN-DC), inter-gNB-DU (NR-DC), intra-gNB-DU (NR-DC), Xn-Based SA and N2-Based SA handovers.

Mar
2019

3GPP publishes TS 23.216, v16.0.0 towards Release 16. This document specifies 5G-SRVCC. This is a handover from 5G NR to 3G UTRAN where circuit-switched voice call will happen.

Jul
2020

3GPP releases 5G Release 16. Two handover enhancements in this release are DAPS Handover and Conditional Handover. In both cases, the idea is to minimize the handover interruption time. Also in this release, direct data forwarding is supported for inter-system handovers.

Sep
2020

3GPP publishes TS 28.313, v16.0.0 as part of Release 16. This specifies details of Self-Organizing Network (SON). Under the feature MRO (Mobility Robustness Optimisation), this specifies MRO targets, handover-related performance measurements, blacklisting/whitelisting neighbours, and prohibiting X2/Xn handovers. In practice, root cause analysis and configuration settings are hard to get right. Local optimizations could cause problems elsewhere.

Nov
2020

3GPP proposes AI/ML for NG-RAN. During this decade, Machine Learning (ML)-based algorithms are proposed to optimize handover decisions. Optimization could be about beam tracking, load balancing or UE mobility. ML algorithms can overcome the limitations of SON.

References

  1. 3GPP. 2017a. "TS 38 401: NG-RAN; Architecture description." V15.0.0, December. Accessed 2024-01-25.
  2. 3GPP. 2017b. "TS 23 502: Procedures for the 5G System; Stage 2." V15.0.0, December. Accessed 2024-01-25.
  3. 5G Networks. 2020. "Virtualised and Disaggregated 5G-NR vRAN Architecture." 5G Networks, July 27. Accessed 2024-01-24.
  4. 5GWorldPro. 2023. "5G NSA Mobility Management." Blog, 5GWorldPro, June 19. Accessed 2024-01-25.
  5. Chandramouli, D., R. Liebhart, and J. Pirskanen (eds). 2019. "5G for the Connected World." John Wiley & Sons Ltd. Accessed 2024-01-25.
  6. Dahlman, Erik, S. Parkvall, and J. Skold. 2018. "5G NR: The Next Generation Wireless Access Technology." Academic Press. Accessed 2021-02-23.
  7. Dano, Mike. 2019. "Another set of 5G standards was just released, but no one really cares." LightReading, April 5. Accessed 2024-01-17.
  8. ETSI. 2020a. "TS 128 313: 5G; Self-Organizing Networks (SON) for 5G networks." V16.0.0, October. Accessed 2024-01-25.
  9. ETSI. 2022a. "TS 123 216: Digital cellular telecommunications system (Phase 2+) (GSM); Universal Mobile Telecommunications System (UMTS); LTE; Single Radio Voice Call Continuity (SRVCC); Stage 2." V17.1.0, July. Accessed 2024-01-25.
  10. ETSI. 2023b. "TS 123 502: 5G; Procedures for the 5G System (5GS)." V17.10.0, September. Accessed 2024-01-17.
  11. ETSI. 2023c. "TS 136 300: LTE; Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description; Stage 2." V17.5.0, July. Accessed 2024-01-17.
  12. ETSI. 2023d. "TS 138 300: 5G; NR; NR and NG-RAN Overall description; Stage-2." V17.6.0, October. Accessed 2024-01-17.
  13. ETSI. 2023e. "TS 138 401: 5G; NG-RAN; Architecture description." V17.6.0, October. Accessed 2024-01-17.
  14. ETSI. 2023f. "TS 138 304: 5G; NR; User Equipment (UE) procedures in idle mode and in RRC Inactive state." V17.6.0, October. Accessed 2024-01-24.
  15. Elhady, M. 2023. "Handover vs Redirection vs Cell Selection vs Cell Re-Selection." On LinkedIn, May. Accessed 2024-01-24.
  16. Fletcher, Bevin. 2020. "3GPP completes latest 5G NR spec with Release 16." Fierce Wireless, July 6. Accessed 2024-01-18.
  17. Gimenez, L. C., P. H. Michaelsen, K. I. Pedersen, T. E. Kolding, and H. C. Nguyen. 2017. "Towards Zero Data Interruption Time with Enhanced Synchronous Handover." IEEE 85th Vehicular Technology Conference (VTC Spring), Sydney, NSW, Australia, pp. 1-6. doi: 10.1109/VTCSpring.2017.8108504. Accessed 2024-01-17.
  18. Haghrah, A., M. P. Abdollahi, H. Azarhava, and J. M. Niya. 2023. "A survey on the handover management in 5G-NR cellular networks: aspects, approaches and challenges." J Wireless Com Network, EURASIP, article no. 52. Accessed 2024-01-17.
  19. Joshi, P. 2022. "5G Release 16 Handover Enhancements." Award Solutions, May 6. Accessed 2024-01-17.
  20. Khosravi, S., H. S. Ghadikolaei, J. Zander, and M. Petrova. 2023. "Reinforcement Learning-based Joint Handover and Beam Tracking in Millimeter-wave Networks." arXiv, v1, January 12. Accessed 2024-01-17.
  21. Mitchell, G. 2019. "Delivering 5G Voice Services." Blog, Mpirical, February 25. Updated 2022-08-11. Accessed 2024-01-25.
  22. Ohlsson, O. 2020. "Reducing mobility interruption time in 5G networks." Blog, Ericsson, April 2. Accessed 2024-01-17.
  23. Ondrusova, S. and D. Kim (eds). 2020. "5G Implementation Guidelines: NSA Option 3." GSMA, February. Accessed 2024-01-25.
  24. Park, K. 2021. "Mobility Enhancement in 5G NR." White paper, Ofinno, July. Accessed 2024-01-17.
  25. Prince. 2014. "Cell Reselection, Redirection and Handover (2G/3G/4G)." Blog, Long Term Evolution (4G-LTE), May. Accessed 2024-01-24.
  26. Shi, N. and A. Vesely. 2020. "Indirect vs. direct data forwarding: A look at the debate." Blog, Ericsson, April 9. Accessed 2024-01-25.
  27. Tayyab, M., X. Gelabert, and R. Jäntti. 2019. "A Survey on Handover Management: From LTE to NR." IEEE Access, vol. 7, pp. 118907-118930. doi: 10.1109/ACCESS.2019.2937405. Accessed 2024-01-17.
  28. TelcoSought. 2021. "5G Handovers – Connected Mode Mobility." TelcoSought, October 24. Accessed 2024-01-17.
  29. Tetcos. 2021. "Performance analysis of 5G handovers (SA SA, NSA NSA, NSA SA) using NetSim." Slides, October. Accessed 2024-01-17.
  30. Wang, D., A. Qiu, Q. Zhou, S. Partani, and H. D. Schotten. 2023a. "Investigating the Impact of Variables on Handover Performance in 5G Ultra-Dense Networks." arXiv, v1, July 26. Accessed 2024-01-17.
  31. Wang, Y., L. Xu, and J. Jang. 2023b. "First Light of the AI/ML Empowered RAN in 3GPP." Blog, Samsung Research, August 16. Accessed 2024-01-17.
  32. Wirelessbrew. 2021. "5G NR UE Timers and Constants." Wirelessbrew, December 25. Updated 2022-01-02. Accessed 2024-01-23.
  33. Yajnanarayana, V., H. Ryden, and L. Hevizi. 2020. "5G Handover using Reinforcement Learning." arXiv, v4, August 4. Accessed 2024-01-17.

Further Reading

  1. Haghrah, A., M. P. Abdollahi, H. Azarhava, and J. M. Niya. 2023. "A survey on the handover management in 5G-NR cellular networks: aspects, approaches and challenges." J Wireless Com Network, EURASIP, article no. 52. Accessed 2024-01-17.
  2. Tayyab, M., X. Gelabert, and R. Jäntti. 2019. "A Survey on Handover Management: From LTE to NR." IEEE Access, vol. 7, pp. 118907-118930. doi: 10.1109/ACCESS.2019.2937405. Accessed 2024-01-17.
  3. Ahmad Khan, S., I. Shayea, M. Ergen, and H. Mohamad. 2022. "Handover management over dual connectivity in 5G technology with future ultra-dense mobile heterogeneous networks: A review." Engineering Sci and Tech, An International Journal, vol. 35, article no. 101172, November. Accessed 2024-01-17.
  4. Hassan, A., A. Narayanan, A. Zhang, W. Ye, R. Zhu, S. Jin, J. Carpenter, Z. M. Mao, F. Qian, and Z.-L. Zhang. 2022. "Vivisecting Mobility Management in 5G Cellular Networks." SIGCOMM '22: Proceedings of the ACM SIGCOMM 2022 Conference, pp. 86-100, August. doi: 10.1145/3544216.3544217. Accessed 2024-01-17.
  5. ETSI. 2023b. "TS 123 502: 5G; Procedures for the 5G System (5GS)." V17.10.0, September. Accessed 2024-01-17.
  6. ETSI. 2023d. "TS 138 300: 5G; NR; NR and NG-RAN Overall description; Stage-2." V17.6.0, October. Accessed 2024-01-17.

Article Stats

Author-wise Stats for Article Edits

Author
No. of Edits
No. of Chats
DevCoins
Avatar of user arvindpdmn arvindpdmn
3
1
1894
DevCoins due to articles, chats, their likes and article hits are included.
1999
Words
1
Authors
3
Edits
4
Chats
0
Likes
5450
Hits

Cite As

Devopedia. 2024. "5G Handover." Version 3, January 25. Accessed 2024-06-25. https://devopedia.org/5g-handover
Contributed by
1 author

Last updated on
2024-01-25 16:08:46