The 5G system architecture · In depth
Every question of this course
All 160 questions on one page, in the order the chapters come. Pick an answer and it tells you at once whether you were right and where the answer lives.
1 Who writes this document, and why
Q1.1 Which of the three stages does TS 23.501 belong to?
The first sentence of the scope says the document defines the Stage 2 system architecture for the 5G System. §1
Q1.2 In a version number x.y.z, what does a first digit of 2 mean?
The Foreword spells the three cases out. A first digit of 1 means presented for information, 2 means presented for approval, and 3 or greater means approved and under change control. §Foreword
Q1.3 Once the document is under change control, what does its first digit track?
The change history shows 16.0.0 approved at SP#83, 17.0.0 at SP#91E and 20.0.0 at SP#110, each one the opening version of a release. §Annex X
Q1.4 Which two documents does the scope name as companions to this one?
The scope names TS 23.502 for the procedures and flows, and TS 23.503 for policy control and charging. All three are stage 2. §1
Q1.5 Many change-history lines share one TDoc number, such as SP-260441. What does that tell you?
The TDoc is the paper submitted to the plenary. Related changes travel together in one pack, so one TDoc number covers many change requests. §Annex X
2 How to read a 3GPP specification
Q2.1 A reference is written without a date or version number. Which version applies?
Clause 2 calls that a non-specific reference. For a 3GPP document it points at the latest version in the same release as the document you are holding, so the target moves as the release does. §2
Q2.2 A term is defined both in clause 3.1 and in TR 21.905. Which definition applies here?
Clause 3.1 says a term defined in the present document takes precedence over the same term in TR 21.905. The same rule is repeated for abbreviations in clause 3.2. §3.1
Q2.3 What does the letter in a clause number like 4.2.5a tell you?
A letter suffix is how a new clause is slotted between two existing ones. 4.2.5a sits between 4.2.5 and 4.2.6, and every cross reference to 4.2.6 elsewhere stays valid. §4.2.5a
Q2.4 A NOTE says a feature is not supported in this release. How binding is that?
A NOTE carries no "shall", but the restriction it records is often stated nowhere else. Clause 5.2.5 is one of them, and an implementation that ignores it is wrong. §5.2.5
Q2.5 A clause title reads only "Void". What happened?
Deleting a clause outright would renumber everything after it and break every cross reference in every other specification. The number stays and the body becomes the single word "Void". §5.8.2.15
Q2.6 You need the exact message sequence for a device registering with the network. Where does clause 1 send you?
The scope names TS 23.502 as the companion holding the stage 2 procedures and flows. TS 23.501 describes what registration is for and which function owns it, never the sequence. §1
3 The shape of the 5G core
Q3.1 How may two control plane functions in the 5G core talk to each other?
The specification says control plane network functions shall only use service-based interfaces for their interactions. The reference points between them are names for those same conversations, not a second way of having them. §4.2.1
Q3.2 What is "Nsmf"?
A name of the form N + lower-case function name is the interface a function exhibits, meaning the set of services it offers. It belongs to one function, not to a pair of them. §4.2.6
Q3.3 Which of these is a real point-to-point reference point, not one realized by service-based interfaces?
N1, N2, N3, N4, N6 and N9 are listed before the sentence that says the rest are realized by service-based interfaces. They are the six that genuinely are their own link. §4.2.7
Q3.4 Where does a network function put unstructured data such as a UE context?
The UDSF holds data whose structure 3GPP does not define. The UDR holds the standardized sets — subscription data, policy data, data for exposure, application data and Ambient IoT data. §4.2.5
Q3.5 What does the specification call the SCP and the SEPP?
The list of network functions is followed by a second list, opened with "also comprises the following network entities", and the SCP and the SEPP are on it. The SCP does not offer services of its own. §4.2.2
Q3.6 Is there a roaming architecture for Energy Efficiency and Energy Saving?
The clause defines a non-roaming architecture only and then says in plain words that the roaming architecture is not supported in this Release. §4.2.18.1
4 What each network function does
Q4.1 The AMF carries session management messages between the phone and the SMF. What does it do with them?
The clause calls the AMF a transparent proxy for routing session management messages. Deciding on a UPF is the SMF's job. §6.2.1
Q4.2 Which function turns a hidden subscription identifier (SUCI) back into the real one (SUPI)?
De-concealment of the privacy-protected identifier is in the UDM's list, together with generating the authentication credentials. §6.2.7
Q4.3 Why can two different UDM instances serve the same subscriber in two different transactions?
The UDM may hold only the application logic and read its data from the UDR, so no instance owns the subscriber. §6.2.7
Q4.4 Which function counts how many devices are registered to a network slice?
The NSSF picks the slice; the NSACF monitors and controls the number of registered devices and of established sessions per slice. §6.2.28
Q4.5 What does the SEPP do to control-plane messages travelling between two operators?
Filtering, policing and topology hiding are its two jobs, and a note says it deliberately does not repeat the producer's authorization. §6.2.17
Q4.6 A function knows only a device's IP address and needs the PCF that handles it. Where does it start?
The BSF holds the registration a PCF made for itself, and a consumer with only an IP address, DNN and slice uses it to get the SUPI. §6.3.3.1
5 Services instead of interfaces
Q5.1 Why does the same piece of behaviour need two different reference point names when it sits between two different pairs of functions?
A reference point is the conceptual point where two functional groups meet, so its name belongs to that pair. A service-based interface belongs to the producer alone, so one name covers every consumer that calls it. §Annex A
Q5.2 What must a consumer put in a subscribe request?
The producer has to be told where to send the notification and how the consumer will recognise which subscription it belongs to. §7.1.2
Q5.3 In communication model D, who finds the producer?
Model D is indirect communication with delegated discovery. The consumer does no discovery and no selection at all; it adds the parameters and the SCP does the rest. §E.1
Q5.4 A network function is told to serve traffic through an SCP. Does that settle how all its messages travel?
Direct and indirect are chosen from the local configuration of the consumer or producer, and the clause says plainly that an NF may not use an SCP for all its communication. §7.1.1
Q5.5 What does the NRF check during service discovery, as opposed to what the producer checks later?
Authorization has two steps at two different granularities. The per-request, per-UE check is embedded in the service logic of the producer, not in the NRF. §7.1.4
Q5.6 How many services does the SCP itself produce?
The SCP has a single service that exposes SCP related information to consumer NFs. Everything else it does is forwarding on behalf of real producers. §7.2.30
6 How one function finds another
Q6.1 What has to happen before a function can be found through the NRF?
Discovery only works for instances that registered, and registration is one message carrying the NF profile. §6.3.1
Q6.2 You get a discovery answer back from the NRF. What may you do with it?
The result belongs to the criteria, not to the person, and the answer carries a validity period saying how long it may be cached. §6.3.1
Q6.3 A producer gives you a binding indication at NF Set level and then stops answering. Where may you go?
The level of the binding is the width of the allowed reselection, and NF Set is the widest of the four. §6.3.1.0
Q6.4 A phone already has a session to one data network and asks for a second one to the same DNN and slice. When is the AMF forced to reuse the same SMF?
Without one of those two subscription flags a different SMF may be picked, which is how load balancing and graceful shutdown work. §6.3.2
Q6.5 Where can the SMF learn what a UPF is capable of?
The N4 node-level exchange is a first-class source of capability information, and asking the NRF is optional on top of it. §6.3.3.1
Q6.6 Must the PCF chosen by the AMF for the device be the same PCF the SMF chooses for a session?
The specification says plainly that the two selections may land on the same instance or on different ones. §6.3.7.1
7 Every name the system uses
Q7.1 Why does a phone almost never send its SUPI to the network?
When the UE has to indicate its SUPI, it provides it concealed — that concealed form is the SUCI. The SUPI itself is used only inside the 3GPP system. §5.9.2
Q7.2 What are the two parts of a 5G-GUTI?
The structure is a GUAMI followed by a 5G-TMSI. The GUAMI identifies one or more AMFs; the 5G-TMSI identifies the UE within them. §5.9.4
Q7.3 Is there one GPSI for every SUPI?
A GPSI is either an MSISDN or an External Identifier, and the note in the clause says plainly that no 1-to-1 relationship is implied. §5.9.8
Q7.4 A device supports no 3GPP access technology at all. What form does its PEI take?
The IMEI or IMEISV format is required only for UEs that support at least one 3GPP access technology. §5.9.3
Q7.5 Moving from 5G to 4G with N26 in use, what does the 5G-TMSI become?
In the mapping, MCC and MNC carry over, AMF Region ID and AMF Set ID become MMEGI and part of MMEC, the AMF Pointer becomes the rest of MMEC, and the 5G-TMSI becomes the M-TMSI. §Annex B
Q7.6 On 3GPP access, what does a UE do with NSSAI in the access stratum by default?
Mode d) is the default for 3GPP access, and the serving AMF may not instruct any other mode unless the home network has said it is allowed. §5.15.9
8 What actually travels on the wire
Q8.1 What sits directly under NGAP on N2, between the access network and the AMF?
The legend of the N2 stack names SCTP, which guarantees delivery of signalling messages between the AMF and the access node. §8.2.1.2
Q8.2 A phone is served by one AMF over 3GPP access and non-3GPP access at the same time. How many N1 NAS signalling connections does it have?
A single N1 connection is used for each access the UE is connected to, even when a single AMF serves both. §8.2.2.1
Q8.3 What does the AMF do with the session-management part of a NAS message?
The SM message is created and processed in the NAS-SM layers of the UE and the SMF. The AMF handles only the NAS-MM part around it. §8.2.2.3
Q8.4 At what granularity does GTP-U encapsulate user traffic on N3 and N9?
The legend says GTP encapsulates all end user PDUs and provides encapsulation on a per PDU Session level. The QoS Flow marking rides inside that same layer. §8.3.1
Q8.5 Which protocol does TS 23.501 name for the service-based interfaces?
Clause 8 gives the service-based interfaces one sentence, and that sentence hands the whole stack to TS 29.500. §8.2.3.1
Q8.6 On untrusted non-3GPP access, what may fragment a large NAS message?
Once the signalling IPsec security association is up, NAS rides on TCP over an inner IP layer, and either of those two may fragment. §8.2.4
9 Registration, reachability and paging
Q9.1 A registered phone is in your pocket overnight, sending nothing. Which pair of states is it in?
The network holds its context and can page it, but there is no AN signalling connection, no N2 connection and no N3 connection. §5.3.3.2.2
Q9.2 What makes a registered device start a Mobility Registration Update?
The TAI list is exactly the area inside which the device may move silently. Leaving it is what breaks the AMF's ability to page. §5.3.2.2.3
Q9.3 A device is CM-CONNECTED with RRC_INACTIVE. Who pages it?
In RRC_INACTIVE the RAN manages reachability and paging, with assistance information from the core. §5.3.3.2.3
Q9.4 A Forbidden Area and a Service Area Restriction both cover where the device is standing. Which one applies?
The specification gives one explicit precedence rule — evaluation of Forbidden Areas takes precedence over Service Area Restrictions. §5.3.4.1.1
Q9.5 What does the network send a device whose service area is limited?
If an Allowed Area is sent, everything outside it is non-allowed; if a Non-Allowed Area is sent, everything outside it is allowed. §5.3.4.1.1
Q9.6 With dual connectivity, how many N2 termination points does one device have?
Two radio nodes provide radio resources, but a single N2 termination point exists between the AMF and the RAN. §5.11.1
10 Radio, Wi-Fi, cable — every way in
Q10.1 A device is in CM-CONNECTED with RRC_INACTIVE and a packet arrives for it. Who pages it?
In CM-IDLE the AMF pages. In RRC_INACTIVE the N2 connection is still up, so the radio network keeps the device and pages it itself. §5.4.3.1
Q10.2 Why does the AMF hold on to the UE Radio Capability information?
The core stores the blob and hands it to the radio network, without understanding its contents, purely to keep it off the air interface. §5.4.4.1
Q10.3 What makes a non-3GPP access network "trusted"?
Both kinds use IPsec to the gateway. What differs is the advertisement and the order of selection. §4.2.8.1
Q10.4 A device is CM-IDLE over Wi-Fi and a downlink packet arrives for a session last routed there. What can the network do?
A device cannot be paged over non-3GPP access at all, so the core uses the other access and marks the request as being about the non-3GPP side. §5.5.3.1
Q10.5 With non-seamless WLAN offload, which IP address does the offloaded traffic use?
The traffic never goes through the 5G core, so the core has no address to preserve and no control over the flow. §5.42
Q10.6 Which value travels in the core-network tunnel header so a gNB can pick a paging policy for a device in RRC_INACTIVE?
The SMF tells the UPF to detect the DSCP and put the matching PPI in the tunnel header, because the gNB pages on its own and never sees the SMF. §5.4.3.2
11 Using two accesses at once
Q11.1 A phone registered on the same network over both radio and Wi-Fi wants a multi-access session. How many establishment requests does it send?
When both registrations are with the same PLMN one request is enough, with Request Type "MA PDU Request" carried in the UL NAS Transport message that conveys it; the AMF tells the SMF the device is on both accesses and both tunnels are set up. Two requests are only needed when the two registrations are with different PLMNs. §5.32.2
Q11.2 Which steering mode sends the same packets over both accesses at once?
Load-Balancing splits traffic by percentage and Priority-based spills onto the second access only when the first is congested. Only Redundant duplicates. §5.32.8
Q11.3 Which steering functionality can carry every kind of traffic but cannot do the Redundant steering mode?
ATSSS-LL applies no protocol of its own and handles TCP, UDP and Ethernet alike, but the specification states it does not support the Redundant Steering Mode. §5.32.6.3.1
Q11.4 How does a round-trip-time measurement get from the phone to the UPF?
Both ends send PMF-Echo Requests and answer each other's, and each derives its own average. There is no reporting of the result from one side to the other. §5.32.5.2
Q11.5 Can a guaranteed-bit-rate flow of a multi-access session be split across the two accesses?
The specification says plainly that traffic splitting is not supported for a GBR QoS Flow; the SMF picks one access, except under the Redundant steering mode where both get the profile. §5.32.4
Q11.6 What does the MPTCP proxy in the UPF use as the source address when it talks to the far-end server?
The two link-specific multipath addresses are used only between phone and proxy and may not be routable on N6. The proxy uses the session's own address towards the final destination. §5.32.6.2.1
12 The connection to a data network
Q12.1 Which attribute of a PDU session may change while the session is alive?
The serving network's slice identifier can change when the UE moves to a different PLMN while keeping the session. DNN, session type, SSC mode and session ID are all fixed for the lifetime of the session. §5.6.1
Q12.2 What happens to the IP address under SSC mode 3?
SSC mode 3 is make-before-break. The new anchor brings a new address or prefix, the old one is kept for a stated time, then released. §5.6.9.2.3
Q12.3 Which session types may be given SSC mode 3?
The specification says the UE shall not request and the network shall not assign SSC mode 3 for an Unstructured or an Ethernet session. §5.6.9.3
Q12.4 An uplink classifier has been inserted into an IPv4 session, so there are now two anchors. How many IPv4 addresses does the device hold?
The device is unaware of the traffic diversion and is not involved in inserting or removing the classifier. Only one address or prefix is given out. §5.6.4.2
Q12.5 A device carrying a local-area data network session walks out of the service area. What does the specification require of the device?
Out of area the device must not activate the user plane, send data or establish and modify the session — but it need not release an existing one unless the network asks. §5.6.5
Q12.6 What does marking a session always-on actually change?
An always-on session is one whose user plane has to come up on every idle-to-connected transition, and the device asks for it even with no pending uplink data. §5.6.13
13 Where the packets actually go
Q13.1 Which N4 rule decides whether a packet is forwarded, dropped or buffered?
The PDR only classifies; the Action attribute is the FAR's, and it is what says forward, duplicate, drop or buffer. §5.8.5.6
Q13.2 Does the SMF send the ARP (Allocation and Retention Priority) to the UPF?
ARP decides which flows are admitted and pre-empted. The clause says plainly that its value is not required to be provided to the UPF. §5.8.2.7
Q13.3 Who allocates the CN Tunnel Info (the TEID and IP address) for an N3 or N9 tunnel?
Allocation and release are performed by the UPF; the SMF only indicates when a new one is needed and when to release it. §5.8.2.3.3
Q13.4 When DHCP is used towards the UE, which function acts as the DHCP server?
The SMF is the DHCP server towards the UE. The anchor UPF has no DHCP function at all; it is only told to forward the packets. §5.8.2.2.1
Q13.5 Buffering of downlink packets for a deactivated session is mandatory where?
Support in the UPF is mandatory and in the SMF optional, and the SMF chooses which of the two does the buffering. §5.8.2.19.1
Q13.6 Annex J works out a worst-case overhead of 142 octets. What link MTU does that leave on a 1500-octet transport?
1500 minus 142 is 1358, and the annex says that value is small enough for most deployments. 1280 is the separate figure that leaves room for a tunnel the UE itself runs. §Annex J
14 What the network promises
Q14.1 What is the finest thing 5G can give a separate quality treatment to?
All traffic marked with the same QFI inside one PDU session gets the same forwarding treatment, so the flow is the unit — not the session and not the packet. §5.7.1.1
Q14.2 What does a 5QI value actually carry?
A 5QI is a scalar used as a reference to the characteristics in the standardised table. Bit rates are separate parameters. §5.7.2.1
Q14.3 In the ARP priority level, which end of the range 1 to 15 is the important one?
The range is 1 to 15 with 1 as the highest priority, and levels 1 to 8 are meant for services the serving network has authorised. §5.7.2.2
Q14.4 Session-AMBR limits the total bit rate across which flows of a PDU session?
Session-AMBR caps the aggregate across all non-GBR QoS Flows of the session and is explicitly not applicable to GBR QoS Flows. §5.7.2.6
Q14.5 A phone has an uplink packet that matches none of its QoS rules. What happens?
For IP and Ethernet sessions the rules are evaluated in precedence order, and if no rule matches, the UE discards the uplink packet. §5.7.1.5
Q14.6 With notification control on, what does the radio network do when it can no longer meet the guaranteed bit rate?
The point of notification control is that the flow survives the bad patch while the application is told to adapt. §5.7.2.4.1a
15 Rules, and paying for them
Q15.1 Clause 5.14 of TS 23.501 is one sentence. What does it say?
The whole clause is a hand-over to TS 23.503. Everything 23.501 keeps about policy lives in other clauses — the PCF description, PCF selection, the services table, and the enforcement clauses under 5.7 and 5.8. §5.14
Q15.2 Session binding is the problem of connecting what to what?
The PCF has to correlate the AF service session over N5 or Rx with the PDU session it handles over N7, which is why a session must be uniquely identifiable from the UE address and DNN as well as from the UE ID. §6.3.7.0
Q15.3 What does the BSF do?
Nbsf_Management allows a PCF to register and deregister itself and to be discoverable. The PCF for the UE and the PCF for the PDU session register separately. §7.2.15
Q15.4 Which function actually counts the bytes for charging a PDU session?
The UPF collects and reports usage data to the SMF over N4; the SMF adds session and subscriber meaning and talks to the CHF. §5.12.1
Q15.5 A PCC rule says "non-blocking" for service data flow handling while credit is requested. What happens to the traffic?
The SMF asks the UPF to report usage for that charging key and gives it a default threshold, so the traffic is not held up while the quota request is outstanding. §5.8.2.6.2
Q15.6 A phone has URSP rules. What does the Network Slice Selection Policy part of them decide?
NSSP rules associate an application with one or more HPLMN S-NSSAIs; the phone uses that to pick the S-NSSAI it indicates at PDU session establishment, or indicates none if it cannot determine one. §5.15.5.3
16 Who you are and who may listen
Q16.1 Where are the authentication methods and key derivations of 5G defined?
Clause 5.10 lists the security features and then says that the detailed security related network functions for 5G are described in TS 33.501. §5.10.1
Q16.2 Which function decides the User Plane Security Enforcement information for a PDU session?
The SMF determines it from the subscribed policy in the UDM, its own local policy per DNN and S-NSSAI, and the maximum integrity rate the UE reported. The NG-RAN only enforces it. §5.10.3
Q16.3 The radio network cannot meet a User Plane Security Enforcement value of "Required". What happens?
Rejecting is the defined behaviour for Required, and the SMF then releases the PDU session. Carrying on and notifying is what happens for Preferred. §5.10.3
Q16.4 Which function de-conceals the SUCI back into a SUPI?
Support of de-concealment of the privacy-protected subscription identifier is listed as a UDM function; the serving network only reads the routing fields that were never concealed. §6.2.7
Q16.5 What has to have happened before slice-specific authentication can run for a device?
NSSAA requires successful primary authentication of the SUPI, and if the SUPI authorization is revoked the slice authorization goes with it. §5.15.10
Q16.6 Which credentials does the remote provisioning of clause 5.39 deliver?
Clause 5.39 covers exactly those two. Provisioning credentials for primary authentication is onboarding, and lives in clause 5.30. §5.39.1
17 One network, many networks
Q17.1 How many S-NSSAIs may the Allowed NSSAI carry in signalling between the device and the network?
At most eight in the Allowed NSSAI and in the Requested NSSAI. The UDM sends the AMF up to sixteen, but only eight fit in the messages. §5.15.2.1
Q17.2 What is the Slice Differentiator, the SD part of an S-NSSAI, for?
The SD is optional extra information that separates several slices of the same SST — for example one per business customer. §5.15.2.1
Q17.3 A device asks for slices, none of them are in its subscription, and it has default subscribed slices. What does the network do?
When no requested S-NSSAI is permitted, the AMF or NSSF builds the Allowed NSSAI from the S-NSSAIs marked as default in the subscription. §5.15.5.2.1
Q17.4 A slice needs its own authentication and it has not finished yet. Where does that S-NSSAI sit meanwhile?
S-NSSAIs under an ongoing slice-specific authentication go in the Pending NSSAI and are removed from the Allowed NSSAI; the device must not ask for them again while they are pending. §5.15.5.2.1
Q17.5 Who keeps the count of PDU sessions in a slice, and who asks it?
The NSACF holds the maximum per S-NSSAI and tracks the current number; the anchor SMF triggers the check at session establishment and release. §5.15.11.2.1
Q17.6 While a slice is replaced by an Alternative S-NSSAI, which value does the device use to match an application to a session?
Application-to-session matching by URSP and session establishment stay on the replaced S-NSSAI; the Alternative one is carried alongside it. §5.15.19
18 Being served by somebody else's network
Q18.1 In Local Break Out, where does the packet leave the mobile network for the internet?
Local Break Out is defined as the case where the PDU Session Anchor and the SMF controlling it are in the serving network. §3.1
Q18.2 In a home-routed session, where does the phone's session-management signalling terminate?
NAS SM terminates in the SMF in the VPLMN, which then forwards session management information to the SMF in the HPLMN. §5.6.3
Q18.3 Who decides whether a given session is home routed or broken out locally?
The HPLMN controls the mode via subscription data, and different simultaneous sessions of one device may use different modes. §5.6.3
Q18.4 What are the two jobs the specification gives the SEPP?
Those are exactly the two bullets in the SEPP clause; everything else about it lives in TS 33.501. §6.2.17
Q18.5 When a visitor uses a non-standard slice value, who maps it?
The NSSF of the VPLMN maps subscribed values to values used in the VPLMN, and need not inform the HPLMN which values it picked. §5.15.6
Q18.6 A visited network cannot meet the QoS the home network asked for. What may the V-SMF do?
Visited-network policies from the agreement may lead the V-SMF to reject the session or some flows, and it may locally downgrade values. §5.7.1.11
19 Living next to 4G
Q19.1 What does the N26 interface join together?
N26 is an interface between two core networks. It carries a subset of what S10 carries between two MMEs, and deploying it is optional. §4.3.1
Q19.2 A device supports both 5G NAS and 4G NAS. Which registration mode must it support?
Single-registration mode is mandatory for any device that supports both 5GC and EPC NAS. Dual-registration mode is the optional extra. §5.17.2.1
Q19.3 The operator does not deploy N26. What keeps a device's IP address across the move?
Without N26 there is no direct core-to-core exchange, so the shared subscriber database holds the anchor's name and hands it to the target core. §5.17.2.3.1
Q19.4 An Ethernet PDU session moves to a 4G core that does not support the Ethernet PDN type. What happens to it?
Ethernet and Unstructured both become "non-IP" in that case, and the device and the SMF keep the real type so it is restored on the way back. §5.17.2.1
Q19.5 In one E-UTRA cell wired to both cores, what tells the base station to route the signalling to an AMF rather than an MME?
The device's NAS layer tells its radio layer that it wants the 5G core, and the radio layer passes that on as an explicit indication. §5.17.1.1
Q19.6 A dual-registration device wants to register in 4G before it moves anything. What must the 4G side support?
Early registration means registering with nothing to carry yet, so the EPC has to allow an attach that establishes no PDN connection at all. §5.17.2.3.3
20 Computing close by, and opening the door
Q20.1 Why does the core pick a user plane function close to the device for edge computing?
The clause names exactly those two gains — reduced end-to-end latency and reduced load on the transport network. §5.13
Q20.2 An application the operator does not trust to talk to core functions directly must reach them how?
Trusted application functions may be allowed to interact directly. Everything else shall use the external exposure framework via the NEF. §6.2.10
Q20.3 Which of these is NOT one of the five categories of external exposure?
The five are monitoring, provisioning, policy/charging, analytics reporting and member UE selection. Nothing in the list touches the radio scheduler. §5.20
Q20.4 What is an AF specific UE Identifier?
It shall not correspond to an MSISDN, it is a GPSI in the form of an External Identifier, and the value handed out is different for different application functions, to protect privacy. §5.20
Q20.5 In user plane direct exposure, who sends the report to the consumer?
The function lets the UPF report network information directly to the consumer based on instructions provided by the SMF, bypassing SMF, PCF and NEF on the reporting path. §5.20d.1
Q20.6 Must an application use the NEF's member UE selection to pick devices for a training round?
The clause says the AF decides whether to use the function, and that without it an AF in either domain can select the members itself using location reporting, user plane information and analytics. §5.46.2
21 Emergency calls, messages, voice and location
Q21.1 A phone with no valid subscription dials the emergency number. What does the specification say the network may do?
A network supporting Emergency Services for phones in limited service state provides them regardless of whether the phone can be authenticated, has roaming or mobility restrictions, or a valid subscription. §5.16.4.1
Q21.2 The network tells a phone "IMS voice over PS session supported". What does that promise?
The indication may also be set when neither NR nor E-UTRA connected to 5GC can carry the call, as long as the radio network can hand over or redirect the phone to EPS when the voice flow is asked for. §5.16.3.2
Q21.3 What does the UPF do with traffic on an emergency PDU session that is not addressed to an emergency function?
The flows of the emergency DNN are dedicated to IMS emergency sessions, and the UPF blocks anything that is not from or to the addresses of functions providing Emergency Services, such as the P-CSCF. §5.16.4.9
Q21.4 A visiting phone makes an emergency call. Which network's SMF serves it?
Emergency SMF selection always derives an SMF in the serving PLMN or SNPN, which is what guarantees the IP address is allocated there too. §5.16.4.5
Q21.5 Can a phone's SMSF be changed while it stays registered in the same network?
Each phone has one SMS Function in the registered PLMN, and reallocation while it is in RM-REGISTERED state in that network is not supported. §4.4.2.1
Q21.6 What does an MPS subscription entitle in the USIM?
The subscription entitles a USIM with a special Access Identity, which is what Unified Access Control uses to let the phone in ahead of ordinary ones when the radio is congested. §5.16.5
22 Clocks, TSN and deterministic delivery
Q22.1 What kind of PDU session must a device have before any of these features work?
The UE has to ask for an always-on session. Home routing is not supported, only SSC mode 1 is, and the session does not survive a move to 4G. §5.27.0
Q22.2 The NW-TT stamped a Sync message on the way in, the DS-TT stamps it on the way out. What does the difference become?
With the grandmaster outside the 5G system, the time carried in the message is never rewritten. The system measures how long it held the message and adds that to the correction field. §5.27.1.2.2
Q22.3 What is one 5GS TSN bridge made of?
The bridge is per UPF, per network instance or DNN and slice, and its bridge ID is bound to the UPF ID. §5.28.1
Q22.4 TSCAI gives the radio a Burst Arrival Time. Expressed against which clock?
The SMF maps the value from the external clock to the 5G clock using the offset and rate ratio the UPF measures and reports. §5.27.2.1
Q22.5 The radio cannot line the requested burst arrival up with a transmission opportunity. What may it send back?
If the application said it can adapt, the radio returns an offset inside the burst arrival window, and may also return an adjusted periodicity. §5.27.2.5.1
Q22.6 Which four parameters configure hold-and-forward buffering in a translator?
Those four give the externally visible behaviour of scheduled traffic with protected windows, so only one transmission gate is ever open. §5.27.4
23 Very fast, very reliable, very picky
Q23.1 A device sets up two redundant PDU sessions. What tells the radio network that those two belong together?
The PDU Session Pair ID identifies the two sessions as a pair; the RSN is what tells them apart, so the radio gives each one different user plane resources. §5.33.2.1
Q23.2 With two N3 tunnels carrying the same QoS Flow, how does the receiver throw away the copy?
The sender replicates the packet and gives both copies the same GTP-U sequence number; the receiver eliminates on that number, and no reordering is required. §5.33.2.2
Q23.3 What must a PCC rule carry before the radio will do PDU Set based QoS handling?
At least a PSIHI, or at least both a PSDB and a PSER, has to be sent. The PDU Set Size is carried per packet in the GTP-U header, not agreed in a rule. §5.7.7.1
Q23.4 When the anchor UPF does the ECN marking for L4S, where does it learn that the radio is congested?
The radio reports the percentage of packets the UPF should mark; if there is no uplink packet to carry the report, the radio may make a dummy one. §5.37.3.3
Q23.5 Why would an application server put media information into a UDP option on the way to the UPF?
The inner datagram is the encrypted packet between server and device; the outer one carries the media information the UPF would otherwise have had to read out of the payload. §Annex W
24 Networks that are not for everybody
Q24.1 What identifies a stand-alone non-public network?
The combination of a PLMN ID and a Network identifier names the SNPN. Neither half is required to be unique on its own. §5.30.2.1
Q24.2 A device registers in an SNPN whose NID came from coordinated assignment, and there is no subscription for it. What does the AMF do?
A coordinated NID is meant to be unique, so the refusal can be trusted forever. With a self-assigned NID the bar is only temporary, because the same NID may belong to a different network elsewhere. §5.30.2.5
Q24.3 A device camped in an SNPN wants a service from a public network. What role does the SNPN play towards that public network?
The device registers a second time with the public network over the SNPN user plane, reaching an N3IWF, exactly as it would over a café Wi-Fi. §5.30.2.7
Q24.4 How many 5G VN groups can one PDU session reach?
A PDU session provides access to one and only one 5G VN group, and a group is tied to a single DNN and S-NSSAI combination. §5.29.3
Q24.5 Which element of a Personal IoT Network carries the traffic to the 5G network?
The PEGC is the gateway and is a UE with PIN subscription data. The PEMC manages the network and is otherwise an ordinary UE. §5.44.1
Q24.6 What stops a device from resuming a connection in a Closed Access Group cell it is not allowed in?
The Allowed CAG list travels to NG-RAN inside the Mobility Restrictions, so the radio network can refuse a resume or a handover without asking the core. §5.30.3.4
25 Small devices, small messages
Q25.1 In Control Plane CIoT 5GS Optimisation, where does the user's data ride?
The point of the optimisation is that no user-plane connection is set up for the PDU session at all; the data travels inside signalling. §5.31.4.1
Q25.2 Serving PLMN Rate Control is expressed in which unit?
A deci hour is a tenth of an hour, six minutes. The limits are separate for uplink and downlink and configured in the (V-)SMF. §5.31.14.2
Q25.3 Which control exists so an operator can sell "a maximum of Y messages per day"?
Small Data Rate Control is the home network's tool. Serving PLMN Rate Control protects the visited network's own AMF and signalling bearers. §5.31.14.1
Q25.4 Which PDU sessions may the Reliable Data Service be used with?
The restriction exists so that one instance of the service never has to straddle the user plane and the control plane at the same time. §5.31.6
Q25.5 How many PDU sessions can have active user-plane resources at once over NB-IoT?
NB-IoT supports a maximum of two data radio bearers, and the mapping from the single QoS Flow of a session to a bearer is one-to-one. §5.31.19
Q25.6 When is the Service Gap timer started?
A connection the network asked for does not cost the UE its next turn, and neither does a bare registration; one the UE started to send something does. §5.31.16
26 Relays, satellites, femtos and shared networks
Q26.1 An IAB node relays traffic for the base station above it. At which layer, and what does that save?
IAB relays at layer 2 and the F1 interface between donor and node is not visible to the 5GC, so no user-plane function sits in the relay. §5.35.1
Q26.2 A MWAB is a gNB on a vehicle. What carries its N2, N3 and Xn traffic back to the core?
The MWAB-UE registers and establishes PDU sessions like any device, and the gNB's interfaces ride over the IP or Ethernet connectivity they give. §5.49.1.1
Q26.3 What does a mobile base station relay send when it establishes its RRC connection?
The mobile IAB-indication makes the donor pick an AMF that can authorise a mobile node; sending the plain IAB-indication as well is forbidden. §5.35A.1
Q26.4 The AMF tells the SMF that the backhaul category is DYNAMIC_LEO. What does the word "dynamic" add?
Dynamic categories mean the capabilities of the backhaul vary, for example because inter-satellite links change; only one category is ever indicated. §5.43.4
Q26.5 What does the NR RedCap indication actually give the core?
The indication travels from the radio to the AMF, which treats it as the RAT type and signals it onward, for example for charging differentiation. §5.41
Q26.6 A PDU session is served by an SMF and an inserted I-SMF. Which of the two talks to the PCF and the charging function?
The I-SMF controls the user-plane functions the SMF cannot reach, but the SMF stays the network function with the PCF and CHF interfaces. §5.34.1
27 Overload, energy, deployment and the rest
Q27.1 What does an overloaded AMF send to make the radio network turn phones away?
N2 overload control is the AMF invoking the overload procedure towards all or a proportion of the radio nodes it has N2 connections with, and it may carry a percentage of requests to restrict. §5.19.5.2
Q27.2 A Mobility Management back-off timer is running in a phone. Which of these is it still allowed to start?
While the timer runs the phone starts no NAS request except deregistration and procedures not subject to congestion control, and after such a deregistration the timer keeps running. §5.19.7.2
Q27.3 Why does the specification tell the core to pick a different back-off value for each phone?
Identical timers would make a large number of phones retry almost simultaneously, which recreates the storm the timer was meant to break. §5.19.7.1
Q27.4 An AMF is being taken out of service and there is no UDSF in the network. What happens to the contexts of the phones it serves?
Without a UDSF the AMF forwards registered UE contexts, grouped by the same GUAMI value, to target AMFs within the same AMF set, and then deregisters itself from the NRF. §5.21.2.2.2
Q27.5 A user switches on 3GPP PS Data Off. What still gets through?
The feature blocks IP, Unstructured and Ethernet traffic via 3GPP access except the 3GPP PS Data Off Exempt Services, and the status of a PDU session does not affect transfer over non-3GPP access. §5.24
Q27.6 Which function works out how much energy one phone, slice or session consumed?
The EIF collects the inputs, calculates consumption at UE, S-NSSAI, PDU session and service data flow granularity, and exposes the result to authorised consumers. §5.51.2.1