Devaki Chandramouli1, Subramanya Chandrashekar2, Jarmo Makinen3, Mikko Säily3 and Sung Hwan Won4
1Nokia, Irving, TX, USA
2Nokia, Bangalore, India
3Nokia, Espoo, Finland
4Nokia, Seoul, South Korea
The fifth generation (5G) access control and mobility management (MM) framework is intended to be adaptive and flexible, to cater for the disparate 5G mobility requirements coming from very diverse use cases.
Some main drivers for this MM framework are the following:
With the interest to increase convergence between 3GPP and non‐3GPP access, features that are common to both access were considered “mandatory” and other access specific features were considered optional for the network to support. So, with the new system, registration procedure is part of the essential features for all access while features such as paging, reachability, etc. are 3GPP access specific thus considered optional rather it is necessary to support only for 3GPP access. Furthermore, need to support “mobility on demand” has also caused 3GPP to revisit the use of the term “mobility management” for MM framework in 5G system. The network function that supports registration, connection management (referred to as MM earlier) is called as AMF that stands for “access control and mobility management function.”
In a broader sense “mobility management” includes all procedures used to support registration and mobility of a user equipment (UE), such as:
Furthermore, certain security related tasks are usually performed as part of the mobility management procedures: authentication, confidentiality protection of the subscriber's identity and confidentiality and/or integrity protection of signaling messages and user data. Detailed security aspects are specified in Chapter 7.
This chapter provides a detailed overview of access control and mobility management framework in 5G Systems. It also provides a detailed overview of radio resource control protocol (RRC) states, non‐access stratum (NAS) states, handover and mobility procedure in various states, also handover procedures for Non‐Standalone (NSA) mode and Standalone (SA) mode. In addition, it provides some insights on challenges due to high mobility and requirements to support ultra‐reliable low latency with mobility. Furthermore, it outlines the procedures for inter‐system mobility. Finally, it provides an outlook for supporting enhanced Mobility features in subsequent 3GPP releases.
The mobility requirements for 5G services are significantly more stringent compared to 3G or 4G. The user plane and control plane latency, reliability and robustness to radio link failures, mobility and mobility interruption time requirements are defined to ensure the introduction of mobile systems which support the new capabilities of systems beyond IMT‐2000 and IMT‐Advanced. IMT‐2020 [1] aims at more flexible, reliable and secure mobile system providing diverse services in the enhanced mobile broadband (eMBB), ultra‐reliable and low‐latency communications (URLLC), and massive machine type communications (mMTC). Further, 5G System is expected to operate in a wide range of frequencies (1–100 GHz) meaning that beamforming techniques are needed to compensate for the higher propagation loss at high frequencies. In this chapter, the 5G System mobility solutions are introduced which can address the 5G mobility requirements.
One of the key target agreements made so far in 3GPP is that handover interruption should be close to 0 ms with UE implemented with a single transceiver capability. While this is challenging and depends on the exact definition of interruption‐less mobility, this can be considered as a major improvement over 4G, which by default always involves a data interruption during handover due to legacy baseline handover procedure. Without going into more detailed solution, there are two basic principles on how the interruption‐less handover can be achieved. With a single TX/RX solution the UE and network can deploy a normal handover with Make‐Before‐Break handover (connectivity to source cell is maintained until connectivity to target cell is established), which won't be interruption‐less but can achieve close to 0 ms interruption with reasonable implementation cost. With more than one TX/RX solution at UE the connection to source cell can be maintained with uninterrupted user plane data while the connection to target cell is established, also known as multi‐connectivity in 5G.
Performance‐wise, the mobility solutions and supporting architectural decisions are having a paramount impact on the resulting capability of system performance. The user plane latency is targeted to be in the order of 4 ms for eMBB, while the URLLC services need far lower latency in the order of 1 ms. The 5G control plane needs to support both extreme power saving and low latency, e.g. transition time from battery efficient state to ready to data transfer. The minimum requirement for control plane latency is 20 ms and 3GPP has decided to target for 10 ms latency.
Mobility supported in 5G for 3GPP access could be classified into several different categories. One classification is based on the RRC state of the UE:
Further, in each of these states, depending on the type of mobility event, there is a further categorization:
Then there is mobility in SA connectivity and NSA connectivity modes. This is applicable to both intra NR and inter‐RAT use cases.
There are more flavors available based on type of deployments, namely classical gNB deployment and cloud gNB deployment. Cloud gNB is a gNB where the constituents of a gNB (CU‐CP(Central Unit Control Plane), CU‐UP(Central Unit User Plane), DU) are separated over F1 (see TS 38.473 [2]) and E1 interfaces (see TS 38.463 [3]). A classical gNB is a gNB where the F1 and E1 interfaces are internal. A cloud gNB may have more than one gNB‐DU under its control, while a classical gNB has only one gNB‐DU under its control.
For more details on RAN architecture, please refer to Chapter 4.
While the architecture of a classical gNB is realized by juxtaposing the logical entities of a cloud gNB, the 3GPP defined interfaces between those logical entities (F1 and E1) will simply become internal. Therefore, we describe the mobility events for a cloud gNB and explain how they are realized for a classical gNB.
We describe in this chapter the various kinds of mobility for different RRC states (e.g. Intra gNB, Inter gNB. Intra DU, Inter DU, and Inter RAT). RRC Connected, RRC Inactive and RRC_IDLE have their own mobility procedures.
The NAS layer can be categorized into two main sublayers (See TS 24.501 [4]): MM sublayer and SM sublayer. The main function of the MM sublayer includes registration management (RM), and the main function of the SM sublayer is to manage the protocol data unit (PDU) session context.
The NAS state machine is applicable for both 3GPP and non‐3GPP access. In evolved packet system (EPS), 3GPP NAS SM protocol for non‐3GPP access was emulated to support wireless local area network control plane protocol (WLCP) between UE and enhanced packet data gateway (ePDG) (see [5]). However, the NAS specifications and the actual protocol were still different and evolved separately for 3GPP and non‐3GPP access. With 5G architecture, target was to specify a single NAS protocol that evolves together for both 3GPP and non‐3GPP access. Although not all information elements within NAS are expected to be common and applicable for both 3GPP and non‐3GPP access, it is expected that the overall mobility, authentication, and session management framework are applicable for both thus enabling support for converged core network. This also enables seamless mobility between 3GPP and non‐3GPP access without the need to re‐authenticate in all cases.
The simplified form of the state machine managed in the MM sublayer for RM is depicted in Figure 5.1. The state machine is managed per access type (i.e. 3GPP access or non‐3GPP access).
In the RM‐NULL state, 5GS services are disabled in the UE. No MM function is performed in this state. The UE in the RM‐DEREGISTERED state can enter to this state by disabling the N1 mode, a mode of a UE allowing access to the 5GC via the 5G‐AN. The UE in the RM‐NULL state can transition to the RM‐DEREGISTERED state if the N1 mode is enabled.
In the state RM‐DEREGISTERED, no MM context has been established and the location of the UE is unknown to the 5GC and hence it is unreachable by the 5GC. To establish a MM context, the UE shall start the initial registration procedure. The UE in the RM‐REGISTERED state can enter to this state by (locally) deregistering from the 5GS.
In the state RM‐REGISTERED, a MM context has been established. Additionally, one or more PDU session context(s) may be activated at the UE. The UE may initiate the non‐initial registration procedure (including the mobility registration update and periodic registration update).
Furthermore, in the state RM‐REGISTERED, UE can either be in CM‐CONNECTED state or CM‐IDLE state (Figure 5.2). When the UE is in CM‐IDLE state, the network must page the UE to be able to reach the UE in the case of the 3GPP access. Similarly, the UE needs to send an initial NAS message to transition to CM‐CONNECTED state to transmit either signaling or user plane packets. When the UE is in CM‐CONNECTED state, UE and network can communicate at the NAS layer level.
The simplified form of the state machine managed in the SM sublayer is depicted in Figure 5.3. The state machine is managed per PDU session.
In the INACTIVE state, no PDU session context exists. The successful PDU session establishment allows transition to the ACTIVE state.
In the ACTIVE state, PDU session context is active in the UE. The state of a PDU session context with the ACTIVE state changes to the INACTIVE state if the PDU session context is released by the PDU session release procedure. The PDU session context in this state can be modified by the PDU session modification procedure.
With 5G, there is a need to address the “smart” problem with “always on” applications:
Hence, to tackle this challenge, the 5G systems was expected to adopt connectivity with flexibility in its configuration to system access. In 5G, the system is expected to offer a solution to support “always connected” applications taking “always on” to the next level. Thus, the access to the network and state transition to connected state should be instantaneous from the application perspective, in the order of 10 ms [6].
An overview of existing RRC states and state transitions specified by 3GPP for LTE is shown in Ref. [7], which also illustrates the mobility support between evolved universal mobile telecommunications system terrestrial RAN (E‐UTRAN), universal terrestrial radio access network (3G RAN) (UTRAN) and GSM/EDGE RAN (GERAN). Details of mobility management for Evolved Packet System is described in TS 23.401[8] and TS 24.301[9]. The RRC states in LTE, namely RRC_IDLE and RRC_CONNECTED, try to minimize the UE power consumption, network resource usage and memory consumption while the RRC_CONNECTED was introduced for high UE activity and network controlled mobility respectively. The state transition from CONNECTED to IDLE and vice‐versa requires a considerable amount of signaling to setup the UE's AS context in RAN and introduces delay and extra e2e system signaling. Smartphone applications and machine type communications (MTC) devices have resulted in the reconsideration of state handling efficiency in LTE systems. It is costly to keep all the UEs in RRC_CONNECTED state, as high user activity increases power consumption and mobility related signaling. On the other hand, the state transition from RRC_IDLE to RRC_CONNECTED, in the order of 50 ms, cannot meet the 10 ms latency requirements of 5G. This resulted in the need to reconsider RRC state machine for 5G System.
Although the UE is in CM CONNECTED state from the core network (CN) perspective, the RAN can suspend the RRC state of the UE during inactivity periods. From the RAN perspective, the UE can be in RRC connected, RRC Inactive or RRC Idle state. When inactivity is detected, the UE may request based on a configured timer that the network can suspend the RRC connection. Alternatively, the RAN may suspend the RRC connection from RRC_CONNECTED to RRC_INACTIVE after the data buffers are empty or if there is a temporary inactivity detected. The RRC_IDLE state may be rarely used, for example, as a recovery state when RRC resume fails. This resulted in the introduction of RRC_INACTIVE state for 5G and the RRC state machine comprises of RRC_IDLE, RRC_CONNECTED and RRC_INACTIVE [10]. Figure 5.5 illustrates the RRC State machine and expected transitions (for details see Ref. [11]):
Figure 5.6 shows how RRC state machine fits within the NAS states:
The UE mobility is controlled by the network during RRC connected state. However, during RRC Inactive state, the mobility of the UE is controlled by the UE with the assistance of the network. In RRC_IDLE state, i.e. in a kind of recovery state, the UE mobility using cell reselections is autonomously controlled by UE. Lightweight procedures called RRC suspend and RRC resume are respectively used to resume and suspend RRC connection. The RRC suspend message may contain service tailored configuration to address UEs with diverse service requirements.
The UE state transition happens when either the UE or the next generation RAN (NG‐RAN) node detects a low activity condition and UE has no ongoing data traffic in the user‐plane.
When the network commands the UE to Inactive state, the last serving NG‐RAN node (gNB or ng‐eNB) sends an RRC Connection Suspend message to the UE. The message that contains (at least) Resume Identification (ID, in this case the Last NG‐RAN node (gNB or ng‐eNB) ID), Inactive state related timing Information (e.g. Registration period), up‐to‐date list of TAs (Tracking Areas) in which the UE can move without TA update, and Security Information for UE identification while re‐connecting to the network.
Active connection in RRC Connected state is needed again when an application needs to send data. The UE is already connected so it will reconnect to the network via its current NG‐RAN cell and sends the RRC Connection Resume Request message to the NG‐RAN node including (at least) UE ID, Resume ID, Inactive state related timing Information (e.g. time spent in inactive state), and Security Information to verify the UE context. The NG‐RAN node responds to the UE with the RRC Connection Resume Complete message and UE is back to CONNECTED state.
A connection failure during the RRC Inactive can happen for example due to failed cell reselection or if the cell update to RAN was not acknowledged back to UE. Also RAN can detect and assume a connection failure or UE may have been powered off if RAN has not received any location update information from UE within the maximum reporting time period.
In legacy networks, when a UE in RRC_IDLE state measures and selects to camp in a better cell in a new RAT, it enters RRC_IDLE in the new RAT. Similarly, it is agreed in Release 15 that UE in the RRC_INACTIVE state can perform selection to another RAT and the UE enters the RRC_IDLE state in that RAT. The tight interworking between NR and LTE evolution during inactive state could be exploited more effectively for NG‐RAN mobility. An NG‐RAN node can either provide NR or E‐UTRA control plane with RRC protocol terminations toward the UE. Therefore, it could be assumed that eLTE will support also RRC_INACTIVE and allow cell re‐selection between the systems within the NG‐RAN context. The RRC_INACTIVE state could have a common configuration with RAT specific details and a common RAN Notification area configuration shared within NG‐RAN. The RRC States in the UE and the anchor eNB remain the same, even if the UE selects a cell that belong to another RAT and its RAN area. The procedure can be referred as NG‐RAN RRC reactivation procedure to activate the RRC connection between the UE and the current anchor eNB, possibly located in another RAT. In addition to activating the RRC connection between the UE and the anchor NB, the anchor eNB may send the necessary UE context information to the new eNB.
As an example, let's consider a scenario where eLTE (E‐UTRA) and NR are connected to 5GC (Figure 5.7). The UE is under the NR coverage and network suspends the RRC connection to RRC_INACTIVE. The UE AS context is stored in the anchor gNB and the UE. The stored UE context may include, e.g. data radio bearer (DRB) configuration, AS security context, etc. The UE is provided with a NG‐RAN notification area that consists of cells that belong to eLTE and NR. If the UE reselects to eLTE, it does not need to update its location nor change RRC state to RRC_IDLE. When the UE has the data available for transmission, it initiates an NG‐RAN RRC activation procedure to the current cell, which fetches the UE Context from the anchor gNB and resumes the RRC connection with the state transition to RRC_CONNECTED.
The new state model and proposed new RRC state together optimize the power consumption of mobile devices during the low activity periods while minimizing the latency for the first packet transmission from the UEs to the network. The identified characteristics of the RRC state model are shown in the Table 5.1. The mobility and system access procedures of the new state model are configurable based on different aspects of use cases, device capability, access latency, power saving, security requirements and privacy.
Table 5.1 Characteristics of RRC states in 5G.
5G state | Mobility | Paging | UE location information | Storage of UE context |
RRC_CONNECTED | NW based cell selection & reselection | NR‐RAN and 5GC | Cell level | UE & NG‐RAN |
RRC_INACTIVE | UE based cell selection & reselection | NR‐RAN and 5GC | RAN Tracking Area | UE & NG‐RAN |
RRC_IDLE | UE based cell selection & reselection | 5GC | CN Tracking Area | 5GC |
The mobility state design for 5G has taken the learnings from the pros and cons of state handling design in the existing technologies and considered the 5G use cases and their requirements. Some benefits of the new 5G RRC state model using RRC_INACTIVE are that it:
UE(s) need to register to the network to receive services, e.g. enabling mobility and reachability of mobile devices. The registration procedure is used when the UE performs initial registration to the 5G system, a mobility registration update upon changing to a new Tracking area (TA) outside the UE's registration area in both CM_CONNECTED and CM_IDLE mode or when a UE performs a periodic registration update (due to a predefined time of inactivity), and additionally when the UE needs to update its capabilities or protocol parameters that are negotiated in the Registration procedure.
Initial registration is performed after the UE power‐on, when the UE is not yet registered nor connected to the network. UE starts connection to the network using access control procedures which start from network selection. UE needs to determine to which public land mobile network (PLMN) to attempt registration and which access network to select. Initial access consists of functions where the cell search and acquisition are done at first allowing a UE to synchronize to the target cell. The 5G enables the synchronization channels to have different periodicity (5, 10, 20, 40, 80 or 160 ms) depending on, e.g. traffic load and power saving requirements, which is quite different from LTE which uses periodicity of 5 ms for synchronization signals. The reference signals (RS) are necessary for signal reception and are transmitted in 5G in the same sub‐frame using the same bandwidth and beam as the corresponding user data.
The system information (SI) is distributed over each cell and this enables the UE to acquire the information needed to access the network. The SI includes the cell bandwidths and various configurations for downlink and uplink and is broadcasted using the Master Information Block (MIB) and the Secondary Information Blocks (SIBs). The MIB contains the information on how to read the SIB information. The SIB includes configurations for the UE, such as cell selection and reselection, random access (RA) parameterization and permissions such as access barring and configuration of broadcast services such as warning messages. In 5G the SI is divided into minimum SI and “other SI.” Minimum SI is periodically broadcasted and comprises of basic information required for initial access, while the other SI is provisioned on‐demand basis.
During the initial access over the radio interface, the UE attempts the connection to RAN. Initial access to RAN starts from Random Access procedure. There is a time‐frequency location and identification, when the UE can access random access channel (RACH) but in principle the procedure happens in a random fashion. When the RAN has responded to RA, the connection setup over the radio interface triggers UE's sending an initial NAS message (the REGISTRATION REQUEST message in case of the initial registration), which is used to establish an NAS signaling connection between UE and AMF. After the security setup, the RRC connection can be enabled (i.e. the UE transitions to RRC_CONNECTED state) and a NAS signaling connection can be established.
The following figures depict interactions among UE and network functions in the 5GC upon initiation of the REGISTRATION REQUEST message (i.e. the initial registration procedure). Figure 5.8 shows the initial registration procedure where the AMF does not fetch UE context from any other AMF, and Figure 5.8 exhibits the initial registration procedure in which the AMF (new AMF) receives UE context from a different AMF (old AMF) that had served the UE previously: the new AMF attempts to receive UE context from the old AMF if the REGISTRATION REQUEST message contains the 5G globally unique temporary identifier (5G‐GUTI) and the new AMF can contact the old AMF identified by the globally unique AMF identifier (GUAMI) derived from the 5G‐GUTI. In both figures and corresponding step‐wise descriptions, the procedure is simplified compared to the actual initial registration procedure. See 3GPP TS 23.502 [12] for further details.
Figure 5.9 shows registration procedure with the AMF being able to retrieve UE context from old AMF:
Connected mode mobility procedure differs depending on the type of deployment: NSA or SA.
Here we consider mobility events related to MR‐DC (LTE‐NR Dual Connectivity). MR‐DC (MultiRAT Dual Connectivity) is a generic term used to represent all the three variants of LTE‐NR Dual connectivity, namely,
The Master Node (MN) and Secondary Node (SN) roles could be taken by any node according to the configurations listed above. The following are the mobility events that could occur while dealing with a UE in any of the MR‐DC configurations. Here we describe the mobility events with a common MN and SN terminology:
We describe mobility for the CU‐DU split option, where the CU can be implemented, e.g. in a cloud, since it has use cases like inter‐DU mobility to support that are not applicable for a gNB where CU and DU functions are collocated.
The intra SgNB mobility can be carried out over the SRB1 and SRB2 (eNB assisted) or over SRB3 (autonomous for gNB). Some topics that were considered:
However, it should also be noted that whenever SRB3 is used for RRC re‐configurations or mobility events, there is a certain optimization in terms of latency of the executed procedure.
This scenario is applicable only for a cloud RAN gNB, since multiple DUs are not possible in a classical gNB.
This procedure is used for the case the UE moves from one gNB‐DU to another gNB‐DU within the same gNB‐CU during NR operation. Figure 5.10 shows the inter‐gNB‐DU mobility procedure for intra‐NR.
This procedure is used for the case the UE moves from one gNB‐DU to another gNB‐DU within the same gNB‐CU when only master cell group (MCG) SRB is available during EN‐DC operation (see [16]). Figure 5.11 shows the inter‐gNB‐DU mobility procedure using MCG SRB in EN‐DC.
As mentioned earlier, using MCG SRBs for mobility causes additional latency for the handover event. This is because of the additional X2 messages in DL and UL. Hence it is recommended to utilize SRB3 whenever supported by UE and available at gNB.
However, there could be several reasons why a gNB must turn to MCG SRBs despite having configured SRB3. Combined PCell and PSCell change in MN and SN, reliability of a handover event are some example use‐cases.
It should also to be noted here that the UE will respond to the network in the same SRB route that it received the DL message.
The procedure remains the same in NGEN‐DC case, except for the fact that the interface between MN and MN is Xn.
The change of Secondary Node procedure is initiated either by MN or SN and used to transfer a UE context from a source SN to a target SN and to change the SCG configuration in UE from one SN to another.
The Change of Secondary Node procedure always involves signaling over MCG SRB towards the UE and it cannot be autonomously performed by SN over SRB3.
Figure 5.12 shows the signaling flow for the MN initiated Secondary Node Change:
Figure 5.13 shows the signaling flow for the Secondary Node Change initiated by the SN:
This is described in a generic manner as it is applicable for both NEDC and NGEN‐DC scenarios.
The MN initiated SN change procedure is used to transfer a UE context from the source SN to a target SN and to change the SCG configuration in UE from one SN to another.
The Change of Secondary Node procedure always involves signaling over MCG SRB towards the UE.
Figure 5.14 shows the signaling flow for the SN Change initiated by the MN:
The SN initiated SN change procedure is used to transfer a UE context from the source SN to a target SN and to change the SCG configuration in UE from one SN to another.
Figure 5.15 shows the signaling flow for the SN Change initiated by the SN:
Inter‐Master Node handover with or without MN initiated Secondary Node change is used to transfer context data from a source MN to a target MN while the context at the SN is kept or moved to another SN. During an Inter‐Master Node handover, the target MN decides whether to keep or change the SN (or release the SN).
Figure 5.16 shows the signaling flow for inter‐Master Node handover with or without MN initiated Secondary Node change. For an inter‐Master Node handover without Secondary Node change, the source SN and the target SN shown in Figure 5.16 are the same node.
Inter‐MN handover with/without MN initiated SN change is used to transfer UE context data from a source MN to a target MN while the UE context at the SN is kept or moved to another SN. During an Inter‐Master Node handover, the target MN decides whether to keep or change the SN (or release the SN).
Figure 5.17 shows the signaling flow for inter‐MN handover with or without MN initiated SN change. Note that for an inter‐Master Node handover without Secondary Node change, the source SN and the target SN shown in Figure 5.17 are identical.
The Master Node to eNB/gNB Change procedure is used to transfer context data from a source MN/SN to a target eNB/gNB. Basically, this is about releasing the dual connectivity and getting back to single connectivity. This could happen either due to UE not requiring DC anymore or due to deteriorating NR coverage and UE cannot see any other gNB cells.
Here the Master Node to eNB change procedure is explained in the context of EN‐DC. This is the case when eNB is acting as the Master Node and gNB is the Secondary Node.
Figure 5.18 shows the signaling flow for the Master Node to eNB Change procedure:
Here the Master Node to target eNB/gNB is explained as applicable to MR‐DC in general. This is applicable for both cases:
The change procedure is used to transfer UE context data from a source MN/SN to a target ng‐eNB/gNB. Essentially, this is also about releasing the dual connectivity and getting back to single connectivity.
Figure 5.19 shows the signaling flow for the MN to ng‐eNB/gNB Change procedure:
To support session continuity for UEs making inter‐NG‐RAN node mobility within the 5G system, the 5G system provides support for seamless handover procedures. If there exists a direct interface between NG‐RAN nodes, i.e. the Xn interface, and the use of it is not prohibited, the Xn handover procedure is performed. In the Xn handover procedure, the signaling for handover preparation and execution phases is exchanged between NG‐RAN nodes and then the core network is involved. Otherwise, the N2 handover is performed. In the N2 handover procedure, the core network takes part from the handover preparation phase.
Relocation of the AMF is not supported by the Xn handover. If the AMF must be relocated, the N2 handover needs to be performed. On the other hand, both handover procedures support the intermediate UPF relocation.
The following call flow (Figure 5.20) shows the Xn based inter NG‐RAN handover.
The following call flow (Figure 5.21) shows the N2 based inter NG‐RAN handover.
During connected mode mobility without dual/multi‐connectivity, the UE is connected to one cell or more cells at the same time. The UE measures the neighboring cells to guarantee that the given factors, e.g. link power or quality, cell load, UE capability, access class, are above the configured thresholds to meet the QoS requirements. The UE is configured to report the measurement results back to network. The measurement reports can be triggered based on a specific mobility event configured by network, such as the neighbor cell becomes better than the serving cell by a configured quantity and this condition must be valid for a given period before the measurement report is sent to the network. Typically, the network sends the handover command to UE soon after receiving the measurement report and UE executes the handover procedure immediately to access the specified target cell. However, radio link failures may happen during the handover procedure due to UE missing the handover command or network not receiving the measurement report. The UE may also fail to access the target cell. The Conditional Handover (Figure 5.22) described in [18] is a mobility procedure where the UE receives the prepared handover command a bit earlier and is configured to evaluate and access the target cell a bit later to avoid the risk of the above‐mentioned failures, for example. Thereafter, the Conditional Handover is a promising technique for improving the handover robustness. The following signaling flow indicates the difference to the baseline Xn handover:
The need for an IDLE mode/state has been questioned in 5G as “Always‐on Applications” used in the smart phones need to send and receive small packets frequently to keep IP connectivity open. Essentially the frequent connection requests are related to “push mail/content” services where the UE is frequently checking, if there are any new data available in the application server. Typically, this is a UE initiated event and happens in uncontrolled way from the radio network perspective. Another problem from network perspective are the “heart beat” or “keep alive” messaging events that may occur once per minute, or once every few minutes, and the amount of data is very small (≪1 KB). These messages are used to keep the device towards the network in RRC Connected state.
Despite the enhanced features and functionality, the IDLE state in 5G mobile systems is needed. One major reason is that the needed fault recovery mechanism will add complexity to the new RRC Inactive state. It is an essential requirement for the UE to be able to revert or fallback to a recovery state in case of sudden connectivity fault or network failure. The IDLE state in 5G can support the bootstrap procedures, initial PLMN selection, UE controlled mobility, contention based uplink transmission and core network based location tracking. The IDLE state allows also to reduce energy consumption in the UE, thus allows for longer re‐charge cycles.
Mobility of a UE in IDLE mode is managed by the AMF with a granularity of a registration area, which is a list of tracking area identities. Therefore, if a UE moves out of the registration area assigned during the most recent registration, it performs a new registration, which is called the mobility registration update procedure. On the other hand, for the network to make sure that the UE located in the allocated registration area has not deregistered locally, i.e. deregistered without informing the network, the network controls the UE so that it performs a new registration periodically, which is called the periodic registration update procedure. If the UE does not perform a registration procedure more than a certain period (by default, four minutes longer than the periodicity of the periodic registration update procedure), the network starts a process to deregister the UE implicitly. Figure 5.23 shows the registration procedure due to mobility or periodic update. See Ref. [12] for further details.
In case user plane traffic (one or more downlink PDUs) or control plane signaling message(s) need to be delivered to the UE in IDLE mode, the AMF pages the UE. Since the AMF is not aware of the exact cell on which the UE is camping, the AMF may need to page the UE for the whole registration area in the worst case.
Figure 5.24 shows the paging procedure. See Ref. [12] for the further details.
This section on RRC Inactive state mobility covers the following topics:
To achieve a seamless UE state transition in the 5G system, a connectivity solution where the UE is kept “always‐on” from Core Network (CN) perspective (control and user plane) was defined. Once the UE is registered to the network, the connection to the CN is kept alive. However, the RAN can suspend the RRC connection when there is inactivity. The RAN also has the opportunity to configure differently the behavior of UEs with different service requirements during times of inactivity.
The new RRC Inactive state has many features of the existing LTE_IDLE state, such as low activity towards network and UE based mobility using the cell reselection procedure. The need for configurability of the RRC Inactive state is motivated by 5G use cases which have highly diverse, and sometimes contradictory requirements in terms of reliability, mobility, latency, bandwidth, security and privacy, battery life etc. For example, the E2E latency requirement varies from <1 ms for use cases with ultra‐low latency requirement such as industrial automation, to latencies from seconds to hours for use cases like massive (low‐cost, long‐range and low‐power) MTC. The battery life requirement is irrelevant in some use cases like autonomous driving, where the device can get unlimited energy from the car, whereas the battery life requirement for battery operated devices ranges from three days for smartphones up to 15 years for a low‐cost MTC device.
Allowing a device to use a specific RRC Inactive state configuration enables flexibility to the state handling mechanism. The requirements in Ref. [19] for state transition optimization for ultra‐low complexity, power constrained, and low data‐rate Internet of Things devices can be taken as an example of network slice specific state handling configuration. However, the relevant solutions may not necessarily be applicable for all use cases, e.g. autonomous driving of vehicle.
If parts of the RAN context are available in the network and UE, some of the potential RRC Inactive state configuration options include:
During low activity periods the UE will evaluate the network conditions and select the suitable cell for camping based on UE measurements and according to rules given by the network. With the NR RRC state machine, the cell reselection is applicable in both RRC_IDLE state and RRC_INACTIVE state, while the cell selection is applicable only in RRC_IDLE state. The NAS controls the RAT selection and is looking for the selected PLMN, which is associated with the current RAT. The UE shall select a suitable cell based on RRC_IDLE state measurements and cell selection criteria. UE regularly evaluates the measurements against cell reselection criteria and searches for a better cell for camping. If a better cell is found, that cell is selected. During the normal service provisioning, the UE will decode the control channels to receive the system information of PLMN and receive the registration area information, as well as other AS and NAS Information. Identification of the cell where the UE should be camping is based on the reselection criteria, where the intra‐frequency reselection is based on ranking of cells and inter‐frequency reselection is based on absolute priorities. The cell reselection can be also UE speed dependent where the fast moving UEs are either preferred to stay at macro cell layer and/or perform the cell reselection faster.
Beam‐mobility and multi‐beam operation will also need support during low activity periods. For cell reselection in multi‐beam deployment, the number of beams and the measurement quantity for cell reselection is computed within the cell beams of the same synchronization signal (SS)/physical broadcast channel (PBCH) block per cell. The linear average of the power values from one beam up to the maximum number of highest beam measurement quantity is derived with values above the threshold, thus indicating the cell reselection criteria.
In 5G, the UE can be paged from 5GC or NG‐RAN. For paging occasions, the UE monitors the Paging Channel (PCH) only during the periods which are not configured with use of discontinuous reception (DRX) to reduce power consumption. When the UE is in RRC_INACTIVE state, the paging can be initiated by both 5GC and NG‐RAN and for this purpose the UE needs to monitor both RAN‐initiated and CN‐initiated paging information from paging control channel (PCCH). To minimize complexity of paging from two different sources, the paging occasions of NG‐RAN and 5GC should overlap and therefore the same paging mechanism can be used. As described in Figure 5.23, the RRC_INACTIVE state is configurable with service specific requirements thus allowing paging DRX cycle to be configurable either for power saving (long cycle) or low latency (short cycle). As the UE monitors both 5GC and NG‐RAN initiated paging, the DRX configuration will be provisioned to UE with the paging initiator specific method. The 5GC‐initiated paging and related DRX cycle is configurable via system information, while the NG‐RAN can configure a UE specific DRX cycle to be used for RAN‐initiated paging. During the paging occasions the UE periodically wakes up and monitors the physical downlink control channel (PDCCH) channel to check for the presence of a paging message. When the PDCCH indicates paging RNTI (P‐RNTI) in the paging message, then the UE demodulates the PCH to see if the paging message was directed to it or someone else.
The mobility procedure during the RRC_INACTIVE is optimized for the RAN based location tracking, which means that the UE in the RRC_INACTIVE state can be configured with a RAN Notification Area (RNA). The RNA is the area where the UE can move in RRC_INACTIVE state without notifying the network about the cell reselections. RNA is equivalent to RAN based paging, e.g. if the paging initiator is NG‐RAN, then the UE is paged only in the configured RNA. This approach provides significant savings in terms of control plane signaling compared to paging always from core network over the whole tracking area, perhaps consisting of hundreds of cells. The RNA is flexible supporting stationary UEs and moderately fast moving UEs in low activity periods. Therefore, RNA can cover a single cell or multiple cells and the area falls within the registration area of 5GC. There are several ways to provision the UE with RNA. Typically, a stationary or slowly moving UE will get a configuration via list of cells, e.g. at least one cell. There is a trade‐off between small and larger RNAs and for above pedestrian speeds the UE could be configured with a list of RNAs or a RAN area ID, where a RAN area ID is a subset of a CN Tracking Area or equal to a CN Tracking Area. If the UE is not configured with a list of cells, then it needs to read the system information and use the broadcasted a RAN area ID for RNA updates.
RRC Inactivation will follow the expiration of inactivity timer, which indicates to network that the UE will benefit from configuration to RRC_INACTIVE low activity period and power saving. After the RRC inactivation, UE remains in CM‐CONNECTED state and gets a configuration for RAN Notification Area where the UE can move without notifying NG‐RAN network. The UE Context is stored in the last serving gNB with the serving AMF and UPF keeping their respective UE‐associated N2 and N3 tunnels available. The NG‐RAN node may configure the UE with a periodic RNA Update timer value at transition to RRC_INACTIVE.
The RRC inactive state of a UE can be configured based on the characteristics of the service(s) provided to the UE if such information is available at the network. According to Figure 5.25, RRC Inactive state is included in the Suspend message (see Ref. [20]). If the UE has multiple services or purposes, e.g. a device with multiple concurrent services, then the configuration might be done based on the service with the most stringent requirement state is included in the Suspend message. If the UE has multiple services, e.g. a device with multiple concurrent services, then the configuration might be done based on the service with the most stringent requirements.
The RRC State transition from RRC_INACTIVE state to RRC_CONNECTED state can be triggered by UE or the network, e.g. when the UE low activity period is over and there is data available for uplink transmission or if the network must page the UE for incoming downlink data. The RRC activation described in Ref. [21] specifies the UE and network triggered transition from RRC_INACTIVE to RRC_CONNECTED.
In general, when the UE is Inactivated, the UE is provided by the last serving gNB with the I‐RNTI identifier which is used to identify during the time of activation. This enables the current serving gNB to resolve the anchor gNB and request for the UE Context data. After the RRC activation the path switch is requested from the AMF and the resources are released from the last serving gNB.
Network initiated state transition from RRC_INACTIVE to RRC_CONNECTED involves the RAN paging procedure with the relevant trigger, e.g. there is incoming DL user plane data, DL specific signaling from 5GC, update of critical system information, etc. The RAN based paging is distributed on the need basis in RAN so that the UE can be paged only from the cells under the last serving gNB footprint, or by distributing the paging over Xn interface to other gNBs, which belong to the same configured RNA.
When the UE has been successfully paged from RAN, the U attempts to activate the RRC connection by sending the RRC Connection Resume Request to the network.
The millimeter wave (mmWave) frequencies identified for 5G enable significant bandwidths for next‐generation cellular mobile terminals, various new services in vertical applications, and industry deployments. However, radio links using mmWave frequencies are prone to quick channel variations and radio link failures due to deep free‐space pathloss and absorption. These challenges can be addressed using beam forming directional antennas at the UE and gNB to improve the link budget for better cell coverage. The application of beam forming requires beam management framework to align the transmitter and the receiver beams, which calls for beam formed initial access, beam selection, beam tracking and fast beam failure recovery to identify and maintain the optimal selection of beams to connect the UE and gNB during active data transmission. The beam forming approach chosen for NR allows beam level mobility with low‐cost analog antenna array beam‐sweeping.
Beam level mobility, e.g. intra‐cell mobility in NR, refers to mobility procedures within the same NR cell and can be done without higher level signaling procedures. That is, Beam level mobility does not require explicit L3 RRC signaling to be triggered and RRC is not required to know which beam is being used at a given point in time.
Multi‐beam operation in NR uses synchronization signal block, SS block (SSB), which comprises a primary synchronization signal (PSS), a secondary synchronization signal (SSS), and a PBCH. The UEs are receiving these signals to operate the beam management functions and a given SSB is repeated within an SS burst set to define the gNB beam‐sweeping transmission. Beam sweeping provisions the cell area with a set of beams in spatial domain. The transmitted beam interval and directions are specified and parameters are provided to UEs using system information and dedicated signaling. For example, during the initial cell selection, the UE assumes a default SS burst set periodicity of 20 ms which is repeated for gNB beam‐sweeping transmissions. In general, an SSB is a group of four orthogonal frequency‐division multiplexing (OFDM) symbols in time and 240 subcarriers in frequency where the Demodulation Reference Signal (DMRS) are associated with the PBCH to measure the SSB. The SSBs are grouped into the first 5 ms of an SS burst. Beam measurement will express the reception of SSB in beams in terms of Reference Signal Received Power (RSRP) of the received power with synchronization signals, Reference Signal Received Quality (RSRQ). Beam refinement is the procedure when the optimal set of beams is selected and configured for beamformed communication. Beam reporting sends the measurement information to network with the power and quality of the beamformed signals and the selection of beams is done during the beam refinement phase.
Narrow beam Channel State Information Reference Signal (CSI‐RS) can be used for mobility management purposes in RRC_CONNECTED state. CSI‐RS are configured to UE to achieve better measurement accuracy and better cell range when the same transmitted energy is applied to a narrower CSI‐RS beam compared to wide beam SSB beam. The CSI‐RS beam properties of serving and neighboring cells should include at least NR cell ID, slot configuration for CSI‐RS and the periodicity e.g. 5, 10, 20, 40 ms, as well as configurable numerologies and association between CSI‐RS and SSB for radio resource management (RRM) measurement. For beam management, it is possible to configure multiple CSI‐RS to the same SS burst so the UE can first obtain cell synchronization using the SS bursts, and then use that as a reference to search and refine for CSI‐RS resources.
Considering directional communications with beams, the best beams need to be periodically identified with radio link monitoring and beam management is used to maintain the alignment between the communicating nodes. For this purpose, SS‐ and CSI‐based measurement results can be jointly used to measure if the best beam is selected for communication. The UE measurement configuration can include both SSB(s) and CSI‐RS(s) for the reported cell(s) if both measurement types are available. Beam management is controlled by the Medium Access Control (MAC) protocol (configured by RRC) where the beam failure recovery procedure indicates to the serving gNB of a new SSB or CSI‐RS when beam failure is detected. Beam failure is detected by counting beam failure instance indication from the lower layers to the MAC entity.
In RRC_CONNECTED, when the UE measures the beams of a cell, in practice, it measures a subset of the available beams because the beams not pointing to the UE are typically not measurable or not countable to the measurement result. Measurement results are filtered at first on physical layer for beam level quality and then at L3 for cell quality, potentially from multiple beams. Cell quality from beam measurements is derived in the same way for the serving cell and for the non‐serving cells. Measurement reports may contain the measurement result of more than of the beams with the best value for measurement quantity, although using just the best beam typically represents the cell quality well when beam management is already done below RRC level. Network configures the beam measurements to be included in measurement reports. The possible reported values may contain only the beam identifier, the beam identifier with the measurement result, or no beam reporting.
For RRC_INACTIVE and RRC_IDLE users the beam management means selection of beamformed initial access allowing the UE to establish a physical link to access the network.
Just like in LTE, NR Release 15 specifies network controlled mobility to UEs in RRC_CONENCTED state, with the difference that the mobility is classified into cell level mobility and beam management, e.g. beam level mobility. Cell level mobility with beams refers to inter‐cell mobility with handovers requiring L3 RRC signaling for inter‐gNB handovers. To perform a cell level handover between two gNBs, the four‐step procedure [21] is needed where the source gNB initiates handover and issues a Handover Request over the Xn interface. The target gNB performs admission control and provides the target RRC configuration as part of the Handover Acknowledgement. The source gNB provides the RRC configuration to the UE in the Handover Command as part of the connection reconfiguration. The prepared RRC Reconfiguration message contains the information required to access the target cell, which includes the target cell ID, the new C‐RNTI, the target gNB security algorithm identifier. For beam mobility the information can include a set of dedicated RACH resources, the association between RACH resources and SSBs and the association between RACH resources and UE‐specific CSI‐RS configuration. Intention is to access the target without reading system information, so that the common RACH resources, target cell SIBs, etc. can also be included. When the Handover Command is executed, the RRC connection moves to the target gNB and the target gNB gets Handover Complete message from UE.
5G targets to address many different use cases. One of the more challenging ones is the case, where the user is moving at high velocity. Examples of this kind of situation are high speed trains and cars, where the passengers wish to keep seamless contact with the network and the critical vehicle communications should be maintained as well.
The 3GPP document TR 38.913 [6] specifies a scenario, where a train is moving at 500 km h−1 with up to 300 passengers connected to each macro cell. This leads to frequent handovers for each user and a very high total number of handovers, even if we assume that only some of the users would be active. Furthermore, each handover should be executed within the short time the UE is within the coverage of both the source cell and the target cell. In this kind of high‐speed scenario, the physical layer of 5G needs to be able to tolerate a very high Doppler shift as well.
Another viable scenario would be a car running along a street, which is covered by both <6 GHz macro cells and > 6 GHz small cells. Since the small cells can use frequencies up to several tens of GHz, their coverage areas will be small and their edges sharp. High frequency radio signals don't experience much diffraction and thus an obstacle between the transmitter and the receiver can cut the signal very abruptly. Thus, for these frequencies, a user velocity of 50 km h−1 would already be high speed.
Figure 5.26 illustrates how the received radio signal can fade away, when a UE in a slowly (30 km h−1) driving car passes a street corner. The figure is based on simulation results presented in Ref. [22]. It shows for example that at 28 GHz the signal level can drop 12 dB in 30 ms, which can be sufficient to cut the connection in some situations.
In order, to facilitate a fast handover, where the UE can attach to the target cell before the connection to the source cell is lost, 5G System supports the following features:
3GPP TS 22.261 [23] defines Ultra Reliable Low Latency Communications (URLLC) as a class of latency critical reliable communication service for e.g. the following use cases, which have 10 ms or less latency requirement:
Low latency and high reliability are basically conflicting requirements. High reliability of data delivery is usually achieved by retransmitting the lost data. However, re‐transmissions take time and increase latency. This effect also impacts the data packets following the lost ones, since RAN is expected to rearrange the received packets and send them to the higher layers in the right order. Thus, the latency requirement should always be accompanied by a reliability requirement, which states the probability of a data packet being delivered within the stated latency.
3GPP TR 38.913 [6] specifies RAN latency as the time it takes to successfully deliver an application layer packet/message from the radio protocol layer 2/3 service data unit (SDU) ingress point to the radio protocol layer 2/3 SDU egress point via the radio interface in both uplink and downlink directions, where neither device nor Base Station reception is restricted by DRX. For URLLC, the target for average user plane latency should be 0.5 ms for UL, and 0.5 ms for DL. The target for maximum latency of a 32‐byte packet is 1 ms with 99.999% reliability.
3GPP TR 38.804 [24] specifies that the latency target of 1 ms with 99 999% reliability specified in TR 38.913 [6] should be met in case no cell change (e.g. handover) is needed. If a cell change is required, e.g. because of mobility, some packets may be delayed more. There is no maximum latency specified for this case, but a general statement that the abovementioned latency should be the target.
In the steady state (no cell change) the main sources of user data latency are:
The additional sources adding latency due to mobility:
As described in the previous paragraph, the only way to avoid additional user plane latency at the air interface during mobility events (or any event, which includes a change in a serving cell) is to deploy make‐before‐break handover (e.g. soft handover). In a make‐before‐break handover, the UE is simultaneously connected to both the source cell and the target cell. Consequently, it is basically a form of temporary multi‐connectivity. A UE carrying URLLC traffic is expected to deploy multi‐connectivity also, when stationary, and thus it is natural that make‐before‐break handover should be based on multi‐connectivity.
As described above, multi‐connectivity can avoid interruptions in the air interface, but it may have a challenge, when relocating PDCP using legacy single connectivity principles. The two basic options for PDCP relocation process are (not yet discussed in 3GPP specifications):
Most often mobility events take place between cells on the same frequency. Consequently, a capability to communicate with two or more cells on the same frequency virtually at the same time is essential, to avoid transmission collisions and latency due to consequent retransmissions. This would benefit from MAC layer coordination to avoid interference problems between the links. However, this coordination doesn't need to be as strict as e.g. in CoMP [25]. It is sufficient that the MAC schedulers in each cell are given guidelines that prevent simultaneous UL and DL allocations on the same frequency band. In FDD and static TDD systems this is automatically the case. However, in a TDD system with dynamic DL/UL split, coordination would be beneficial. Secondly, simultaneous UL transmissions on the same radio frequency (RF) hardware would split the available transmit power and reduce maximum range. The MAC coordination function should avoid this as well.
Mobility restrictions are a set of functionalities exploited to restrict mobility handling or service access of a UE in the 5G system. They are supported by the UE, the NG‐RAN and the 5G core network and consist of service area restrictions, forbidden area management, RAT restriction, and core network restriction.
The service area restrictions consist of either an allowed area, or a non‐allowed area. The allowed area can contain up to 16 tracking areas or include all tracking areas in a PLMN. The non‐allowed area can contain up to 16 tracking areas. The UE can initiate any MM and SM procedures while camped on a cell within a tracking area belonging to the allowed area or not belonging to the non‐allowed area. On the other hand, while camped on a cell within a tracking area belonging to the non‐allowed area or not belonging to the allowed area, the UE is prohibited from performing the mobility and periodic registration update procedure or the service request procedure involving user plane resource reactivation except for emergency services or for responding to core network paging. In addition, the SM procedures are also prohibited.
For the forbidden area management, the UE manages a list of “5GS forbidden tracking areas for regional provision of service,” as well as a list of “5GS forbidden tracking areas for roaming.” The UE updates one of the lists whenever a REGISTRATION REJECT, SERVICE REJECT or DEREGISTRATION REQUEST message is received with the cause value “tracking area not allowed” or “roaming not allowed in this tracking area.” In a tracking area whose identity is included in one of the lists, the UE is not permitted to initiate any communication with the network.
The RAT restriction defines the RAT which a UE is not allowed to access in a PLMN. In a restricted RAT, a UE is not permitted to initiate any communication for the PLMN. During the handover, the NG‐RAN should determine the target RAT and target PLMN considering the RAT restrictions for PLMNs, if any.
Core network type restriction defines whether a UE can connect to 5G core network for a PLMN. During the registration, deregistration, or service request procedure, the UE disables capability to connect to 5G core network if the message includes the cause value “N1 mode not allowed.”
A UE may indicate preference for mobile initiated communication only (MICO) mode during the registration procedure. Based on the UE preference, UE subscription information, and network policies, or any combination of them, network determines whether MICO mode is applied for the UE and indicates it to the UE during registration procedure. The UE and core network can re‐initiate or exit the MICO mode at subsequent registration procedure. The MICO mode needs to be applied explicitly in every registration procedure.
When the AMF applies MICO mode for a UE, the AMF considers the UE always unreachable while in idle mode. The UE in MICO mode is reachable for mobile terminated data or signaling only when the UE is in connected mode and for the PDU sessions that has been reactivated. The network rejects any request for downlink data delivery for an idle UE in MICO mode. The network also does not allow downlink transport over NAS for short message service (SMS), location services, etc.
A UE in MICO mode does not listen to paging while in the idle mode. An idle UE in MICO mode may stop any access stratum procedures, until the UE transitions from idle to connected mode due to one of the following triggers:
5G system provides a means to interwork with EPS, i.e. to support session continuity during the inter‐system change (see Chapter 4, Section 4.11 for an overview on EPS interworking, migration from EPS). This section describes the inter‐system change procedure for each of the interworking modes specified in 3GPP technical specifications [3,12]:
The pre‐condition for this scenario is that the UE is single registered in EPS or 5GS, maintains only NAS state at any time (Figure 5.29).
The pre‐condition for this scenario is that the UE is dual registered in EPS and 5GS (Figure 5.30). The UE can maintain NAS states independently in EPS and 5GS. The network may or may not support N26. It is up to UE to decide which system to use for which session but it is up to the network to determine which session is admitted in the system based on regular SM procedures in the respective systems. It is up to the network and operator policy to determine for which PDU Session and/or packet data network (PDN) connection, service continuity is offered for inter system mobility. This information is expected to be provided to the UE as part of registration procedure and corresponding session establishment procedures in the respective system. It is also up to the UE to decide which session to move from one system to another.
It has been acknowledged that the evolution of LTE should be integrated to the 5G to possibly benefit from the widely deployed coverage of E‐UTRA (LTE) in the 2020 timeframe. Tight integration of LTE (E‐UTRA) and new 5G (New Radio/NR) air interfaces should not introduce additional core network signaling complexity, no extra RRC state transitions due to inter‐RAT mobility and no extra load to RRC signaling. A moving UE RRC connection may be suspended or inactivated and UE is using the RRC Connected Inactive state during the low activity period. With tight integration, the connection resumption or activation back to RRC Connected could be done based on the system, which can provide better coverage or capacity. That is, a connection inactivation in 5G may be followed by resumption in LTE when 5G coverage is not available in the same geographical area.
In future releases, we expect support for mobility management in 5G System to be enhanced to support additional features for improved user experience and support new and emerging use cases in the market place. Therefore, 5G will raise new requirements to support new use cases and verticals, e.g.:
Following are some of the foreseen solutions that are under consideration:
As and when new mobility management features are introduced, these new features may be applicable and used only in scenarios where necessary. This is to enable user centric flexible support of new features when and where essential, and at the same time not mandating support for such enhancements for use cases (e.g. stationary, low mobile, low cost device) where they are unnecessary. This approach goes well with the overall theme of the 5G System to be flexible, provide deployment tools for various use cases, and not mandate deployment and use of features but allow the actual deployment to decide when a certain feature is useful.
All 3GPP specifications can be found under http://www.3gpp.org/ftp/Specs/latest. The acronym “TS” stands for Technical Specification, “TR” for Technical Report.