>
HomeLTE 4GFoundationsIdentifiers
◍ FoundationsBeginner

LTE Identifiers (RNTI, GUTI, ECGI, PCI, TAI…)

The identity zoo — RNTIs, IMSI, GUTI/S-TMSI, ECGI, PCI, TAI and the EPS bearer identifiers.

📚 3GPP-basedTS 36.300TS 23.003

An LTE network has to answer four different questions at the same instant — who is this subscriber, which radio message on the control channel is meant for whom, where in the network is the device, and what data flow does this packet belong to. Each question has its own family of identifiers, so the notorious "alphabet soup" of LTE IDs is really four small, well-organised vocabularies that map into one another. This page is grounded in TS 36.300 (E-UTRAN overall), TS 36.321 (MAC / RNTIs) and TS 23.003 (numbering, addressing and identification).

Introduction

An identifier in LTE is any label the network uses to pick out one thing — one subscriber, one control message, one cell, one data flow — from all the others. There are dozens of them, and beginners meet them as an intimidating "alphabet soup." The organising insight is that they fall into exactly four domains, each answering one question: who (subscriber/UE identities), which message for whom (RNTIs on the PDCCH), where (cell and location identities), and what flow (bearer and tunnel identities).

These identifiers appear at every moment of the UE lifecycle. During cell search the UE derives the PCI from sync signals and reads the ECGI and TAI from broadcast system information. During Attach and authentication the permanent IMSI is used once and then replaced by a temporary GUTI. Every scheduled grant in RRC_CONNECTED is addressed by a 16-bit RNTI. Every data pipe is simultaneously an EPS Bearer, an E-RAB, a DRB and a pair of TEIDs. Knowing which domain an identifier serves tells you where it lives, who assigns it, and how long it lasts.

Why it matters: identifiers are the thread that stitches every LTE procedure together, and their design is driven by a set of conflicting requirements — secret versus advertised, tiny versus globally routable, permanent versus disposable. Once you see that each family optimises for one property and the network translates between them, "why yet another ID?" stops being a mystery and becomes predictable.

Why LTE needs so many identifiers

💡

In plain words: think of how you are identified in daily life. Your passport number is permanent and precious, so you lock it in a drawer and almost never show it. Your seat number on today's flight is tiny and disposable — it means nothing tomorrow. Your home address has to be public so post can reach you. And a courier tracking number names one specific parcel in transit. You would never use one label for all four jobs, because their requirements conflict. LTE splits identities for exactly the same reason.

The concrete purpose is that no single identifier can satisfy the opposing constraints. A permanent subscriber key (the IMSI) must be globally unique and secret, so it must almost never travel over the air. A radio label (the C-RNTI) rides on every single PDCCH grant, so it must be tiny (16 bits) and can be local and disposable. A location label (the TAI, ECGI) must be broadcast so idle phones can find a cell unaided. A bearer/tunnel handle (the TEID) must route packets to the right pipe at the gateway. Because secret conflicts with advertised, and tiny conflicts with globally routable, LTE keeps the families separate and translates between them — handing out temporary identities that shield, shorten, or localise the permanent ones.

What

Four identifier families, each optimised for one property: subscriber/UE (who), RNTIs (which message for whom), cell/location (where), and bearer/tunnel (what flow).

Why

The jobs have opposite needs — permanent-and-secret vs tiny-and-disposable vs broadcast-and-findable — so one label cannot do all four.

How

The network assigns temporary identities (GUTI, C-RNTI), rotates them to defeat tracking, and nests them so a smaller ID (S-TMSI) is a slice of a larger one (GUTI).

Four Domains, One Connection

No single identifier can do every job. A permanent subscriber key is precious and must almost never travel over the air; a radio identity has to be tiny because it rides on every single PDCCH grant, and it must be disposable; a location identity has to be broadcast so that idle phones can find it without help. Because these jobs actively conflict — secret versus advertised, tiny versus globally routable, permanent versus throwaway — LTE keeps the identifiers separate and translates between them as procedures demand. The whole zoo falls cleanly into four domains, and almost every "why do we need yet another ID?" question is answered by noticing which domain it serves.

What

Four identifier families: subscriber / UE (IMSI, GUTI, IMEI — who you are), RNTIs (which message on PDCCH and for whom), cell / location (PCI, ECGI, TAI — where you are), and bearer / tunnel (EPS Bearer ID, DRB ID, TEID — which data flow).

Why

Different jobs need opposite properties. A permanent, globally unique key cannot also be small and privacy-safe on the air; a 16-bit radio label cannot be globally routable. So each family optimises for one property and the network stitches them together.

How

The network hands out temporary identities (GUTI, C-RNTI) to shield the permanent ones (IMSI, IMEI), re-allocates them often, and nests them — the S-TMSI is literally a slice of the GUTI so paging can address you with far fewer bits.

🎯

Golden rule: permanent identities (IMSI, IMEI) stay hidden; the network assigns temporary identities and rotates them frequently, so an eavesdropper cannot track a subscriber over time. Every temporary ID exists to protect, shorten, or localise a permanent one.

RNTIs — Radio Network Temporary Identifiers

An RNTI is a 16-bit value the eNB uses to address something on the PDCCH. It is never sent as a visible field. Instead, the eNB computes the 16-bit CRC of a Downlink Control Information (DCI) message and then masks (bit-wise XORs) that CRC with the target RNTI before transmission. A UE that is monitoring a given RNTI XORs the same value back out and checks whether the CRC now passes. If it does, the DCI was addressed to it — or, for the broadcast RNTIs, to everyone. This is why an RNTI is simultaneously an address and an implicit acknowledgement of "this control message decoded correctly and is for me."

Because the mask is only 16 bits, every RNTI lives in the range 0x00000xFFFF. The top of that range is carved out for fixed, spec-defined values that every UE knows in advance, while the lower range is assigned dynamically per UE by the MAC layer. Which RNTI is in play also implies where the UE looks: broadcast and common RNTIs (SI-RNTI, P-RNTI, RA-RNTI) are decoded in the common search space of the PDCCH, while a UE's own C-RNTI grants live in its UE-specific search space. The DCI they scramble points at data on the PDSCH (downlink) or grants resources on the PUSCH (uplink).

RNTIValue / sizeScrambles the CRC of…Usage
C-RNTI16-bit, assigned (0001–FFF3)The UE's dynamic DCI (DL assignments, UL grants)The UE's unique identity in a cell while RRC_CONNECTED; the workhorse RNTI on every scheduled grant.
Temporary C-RNTI16-bit, assigned in RARThe Msg3 grant and Msg4 DCI during random accessHanded out in the Random Access Response (Msg2) before the UE is confirmed; promoted to a full C-RNTI once contention resolution (Msg4) succeeds.
RA-RNTI16-bit, derived (0001–003C)The DCI scheduling the RAR (Msg2) on PDSCHNot assigned but computed from the PRACH time/frequency: RA-RNTI = 1 + t_id + 10 × f_id. Tells the UE which PDCCH carries its preamble's response.
P-RNTIFixed = FFFEThe DCI pointing to the paging message on PDSCHPaging. A single constant every idle UE monitors during its paging occasions.
SI-RNTIFixed = FFFFThe DCI scheduling System Information (SIB) delivery on PDSCHSystem information. A constant known to every UE, since SIBs are broadcast to all.
SPS-C-RNTI16-bit, assignedThe DCI that activates / reactivates / releases the SPS grantSemi-Persistent Scheduling. Sets up a periodic allocation (e.g. VoLTE) that recurs without a fresh grant each interval; a separate RNTI so the UE can tell an SPS action from a dynamic one.
TPC-PUCCH-RNTI16-bit, assignedGroup power-control DCI (format 3/3A) for PUCCHTransmit Power Control for the uplink control channel. One DCI adjusts many UEs' PUCCH power at once; each UE reads its own field by index.
TPC-PUSCH-RNTI16-bit, assignedGroup power-control DCI (format 3/3A) for PUSCHTransmit Power Control for the uplink shared channel — the same group mechanism applied to data-channel power.
M-RNTIFixed = FFFDThe DCI carrying the MCCH change-notificationMBMS. Tells UEs receiving broadcast/multicast services that the MCCH (MBMS control) content has changed and must be re-read.
G-RNTI16-bit, assigned per groupThe DCI scheduling group MBMS data (SC-MTCH) on PDSCHGroup RNTI for Single-Cell Point-to-Multipoint (SC-PTM): all members of an MBMS group share one G-RNTI so a single transmission reaches them together.
How an RNTI addresses a UE on PDCCH DCI payload (grant / assignment) 16-bit CRC RNTI (16 bits) DCI + (CRC ⊕ RNTI) transmitted on PDCCH UE un-masks with its own RNTI → CRC passes = for me
Figure 1. An RNTI is never sent as a field. The eNB XORs it into the DCI's CRC; only a UE monitoring that RNTI can un-mask and get a passing CRC.
🔑

Fixed vs assigned: SI-RNTI (FFFF), P-RNTI (FFFE) and M-RNTI (FFFD) are constants baked into the spec because their traffic is broadcast to all UEs. C-RNTI, Temporary C-RNTI, SPS-C-RNTI, the two TPC-RNTIs and G-RNTI are handed out by the eNB. RA-RNTI is neither — it is a deterministic function of when and where the preamble was sent, so both sides compute the same value without any signalling.

Q&A Quick Q&A

Q. How does an RNTI "address" a UE if it is never transmitted?

A. The eNB masks the 16-bit CRC of the DCI by XORing it with the target RNTI. The UE XORs its own RNTI back out and checks the CRC; a pass means the message was for it. So the address is carried implicitly inside the error-check, which saves overhead on every grant.

Q. Why is a Temporary C-RNTI needed at all — why not assign the C-RNTI directly?

A. During random access several UEs may have picked the same preamble, so the eNB cannot yet trust who it is talking to. It gives out a Temporary C-RNTI in Msg2 and only promotes it to a permanent C-RNTI after contention resolution (Msg4) proves exactly one UE won. Losers drop the temporary value and retry.

Subscriber and UE Identities

These answer who the user and the device are. They live mostly in the core network (the EPC) and are deliberately shielded on the radio. It helps to separate two ideas that people conflate: the subscription (tied to the SIM/USIM and the operator's HSS) and the equipment (the physical handset). The IMSI identifies the former, the IMEI the latter, and they are checked by completely different network functions.

IdentifierStructureWhat it is / does
IMSIMCC (3) + MNC (2–3) + MSIN, ≤15 digitsInternational Mobile Subscriber Identity — the permanent subscription key stored on the USIM and in the HSS. Globally unique; kept off the air whenever a temporary identity can be used instead.
IMEITAC (8) + SNR (6) + check digit (1) = 15 digitsInternational Mobile Equipment Identity — the permanent identity of the hardware. Checked against the EIR to block stolen or non-type-approved devices; independent of the subscription.
IMEISVTAC (8) + SNR (6) + SVN (2) = 16 digitsIMEI Software Version — the same hardware identity but replacing the check digit with a 2-digit software version number, so the network can also reason about firmware level.
GUTIGUMMEI + M-TMSIGlobally Unique Temporary Identity — the MME-assigned stand-in for the IMSI. It both names the serving MME (via the GUMMEI) and gives the UE a temporary identity that is re-allocated to defeat tracking.
S-TMSIMMEC (8 bits) + M-TMSI (32 bits)SAE-Temporary Mobile Subscriber Identity — the short local form of the GUTI. Because it is compact (40 bits) yet unique within an MME pool, it is what actually addresses the UE for paging and in the Service Request.
M-TMSI32 bitsMME Temporary Mobile Subscriber Identity — the innermost temporary number, unique only within one MME. Everything larger (S-TMSI, GUTI) is built by prefixing MME/PLMN context onto it.

Where does the radio-layer C-RNTI fit against all of this? It does not belong to this family at all — and that distinction is a favourite interview trap. The C-RNTI is a throwaway 16-bit label meaningful only inside one cell, for as long as the UE stays connected there; it changes on every handover and is forgotten the moment the UE goes idle. The GUTI, S-TMSI and M-TMSI, by contrast, are core-network identities that persist across cells and across idle/connected transitions, surviving until the MME chooses to reallocate them. In short: the C-RNTI answers "which radio pipe in this cell," while the temporary MSIs answer "which subscriber the core is tracking."

💡

Nesting to remember: the M-TMSI is the innermost number; prefix the MMEC and you get the S-TMSI; prefix the whole GUMMEI (which itself carries the PLMN ID) and you get the full GUTI. Each layer adds exactly enough context to widen the scope from "one MME" to "the whole world."

Network and Cell Identities

These answer where — which cell on the air, which cell in the operator's global map, which tracking area for paging, and which operator altogether. Two of them are constantly confused, so pin the difference early: the PCI is a physical-layer label the UE detects blindly and instantly during cell search, while the ECGI is a globally unique administrative name that the UE only learns after decoding system information.

The PCI (Physical Cell Identity) has just 504 possible values and is reconstructed from the two synchronisation signals. It is formed as PCI = 3 × NID(1) + NID(2), where NID(1) (0–167, the cell-ID group) comes from the SSS and NID(2) (0–2, the identity within the group) comes from the PSS. Because only 504 exist, operators must plan them so that no two neighbouring cells reuse the same value — otherwise PCI collision (adjacent cells sharing a value) or PCI confusion (two neighbours of one cell sharing a value) breaks synchronisation and handover.

IdentifierStructureMeaning
PCI1 of 504Physical Cell Identity = 3 × NID(1)(SSS) + NID(2)(PSS). Detected during cell search; also seeds the reference-signal scrambling.
ECGIPLMN ID + ECIE-UTRAN Cell Global Identifier — the globally unique name of one cell, read from SIB1.
ECI28 bits = eNB ID (20) + Cell ID (8)E-UTRAN Cell Identity — unique within a PLMN; the cell part that the PLMN ID is prefixed to in order to make the ECGI.
eNB IDMacro: 20 bits (Home eNB: 28 bits)Identity of the eNB within a PLMN; embedded in the ECI. A macro eNB gives 20 bits to the node and 8 to its cells; a Home eNB uses all 28 bits itself.
TAIPLMN ID + TACTracking Area Identity — the globally unique name of a tracking area.
TAC16 bitsTracking Area Code — the granularity at which idle UEs are paged and perform Tracking Area Updates. A UE registers a TAI list, not a single cell.
PLMN IDMCC (3) + MNC (2–3)Public Land Mobile Network ID — identifies the operator. A building block inside GUTI, GUMMEI, ECGI and TAI.
GUMMEIPLMN ID + MMEGI + MMECGlobally Unique MME Identifier — names exactly one MME worldwide; the front part of the GUTI.
MMEGI16 bitsMME Group Identifier — identifies a pool (group) of MMEs within a PLMN.
MMEC8 bitsMME Code — identifies one MME within its group; also the prefix that turns an M-TMSI into an S-TMSI.
ECGI = PLMN ID + ECI  where  ECI = eNB ID (20) + Cell ID (8) = 28 bits

The location identities form a clean hierarchy. A PLMN ID (MCC+MNC) names the operator; append a TAC and you have a TAI naming one tracking area; the UE is registered to a whole TAI list so it can roam across several tracking areas without a TAU at every boundary, and is paged across all cells of that list. The cell hierarchy runs in parallel: the 20-bit macro eNB ID names the base station, appending the 8-bit Cell ID gives the 28-bit ECI, and prefixing the PLMN ID gives the globally unique ECGI. Both the serving ECGI and the TAI/TAI-list are broadcast in SIB1, which is how a UE knows exactly which cell, base station, tracking area and operator it is camped on the moment it decodes system information.

🎯

PCI vs ECGI: the PCI is a local, reusable radio label (only 504 of them) found blindly from PSS/SSS; the ECGI is a globally unique administrative name read from SIB1. The UE detects the PCI first, then reads the ECGI to know exactly which cell — and which eNB, via the embedded eNB ID — it is camped on.

Bearer and Tunnel Identities

These answer what data flow a packet belongs to. A single connection is sliced into layered segments from the EPC down to the air interface, and each layer keeps its own handle for the very same logical pipe — they are not five different pipes but five names for one QoS flow, plus a couple of connection-level handles the control plane uses to keep the two ends in step.

IdentifierScope / sizeWhat it names
EPS Bearer ID (EBI)End-to-end (UE–PGW), 4 bitsThe end-to-end EPS bearer — the QoS pipe the core sets up. The default bearer is created at attach; dedicated bearers add guaranteed or specific QoS. Values 0–4 are reserved, so usable EBIs are 5–15.
Linked EPS Bearer Identity (LBI)End-to-endThe EBI of the default bearer that a dedicated bearer is associated with. It ties every dedicated bearer back to the default bearer in the same PDN connection, so releasing the default tears down the whole set.
E-RAB IDS1 + radio leg (UE–SGW)E-UTRAN Radio Access Bearer — the concatenation of the S1 bearer and the radio bearer. It is the S1AP-level view of an EPS bearer and numerically tracks the EBI.
DRB IDRadio-only (UE–eNB), 1–32Data Radio Bearer identity — the PDCP/RLC leg over the air, the radio half of the E-RAB.
TEIDPer tunnel end, 32 bitsGTP Tunnel Endpoint Identifier — names one end of a GTP-U user-plane tunnel on S1-U / S5-S8, so packets are routed to the correct bearer at the SGW/PGW. Each direction and each hop has its own TEID.
eNB UE S1AP IDPer UE per S1 link, 24 bitsIdentifies the UE association over S1-MME as the eNB sees it. Allocated by the eNB when signalling for that UE begins.
MME UE S1AP IDPer UE per S1 link, 32 bitsIdentifies the same UE association as the MME sees it. The pair (eNB UE S1AP ID, MME UE S1AP ID) uniquely names one UE's control-plane context on the S1 interface.

Note the two S1AP IDs are not bearer identities at all — they identify the UE's signalling association on the S1-MME interface, one number chosen by each end. They are how the eNB and MME refer to the same UE in every S1AP message without re-sending the GUTI or IMSI, and they share the "per-connection plumbing" mindset with the bearer and tunnel handles: transient, interface-scoped, and invisible to the subscriber.

💡

Same pipe, many names: one QoS flow is an EPS Bearer ID to the core, an E-RAB ID across S1, a DRB ID on the radio, and a pair of TEIDs on the GTP-U tunnels — while the LBI ties dedicated bearers back to their default, and the two S1AP IDs keep the eNB and MME pointing at the same UE context. Each layer simply relabels the flow it inherited from the layer above.

Anatomy of the GUTI

The GUTI is the keystone of the temporary-identity system, so it is worth dismantling piece by piece. Its design goal is to address a specific subscriber and identify which MME holds that subscriber's context, without ever exposing the IMSI on the air. Reading it from the outside in: the GUMMEI answers "which MME in the world," and the M-TMSI answers "which UE inside that MME." Everything the UE needs for paging and Service Request is a sub-slice of this one structure.

GUTI = GUMMEI + M-TMSI GUTI GUMMEI (which MME in the world) M-TMSI (32 bits) PLMN ID (MCC+MNC) MMEGI (16 bits) MMEC (8) S-TMSI = MMEC + M-TMSI (used for paging) GUMMEI names the MME • M-TMSI names the UE inside it • S-TMSI is the short slice
Figure 2. The GUTI nests the operator's PLMN, the MME identity (MMEGI + MMEC) and a temporary number; the short S-TMSI is just the MMEC + M-TMSI slice used for paging and Service Request.

The elegance is that paging never needs the full GUTI. Within the tracking area, the pool of MMEs is already known, so the MMEC plus the M-TMSI — the 40-bit S-TMSI — is enough to uniquely identify the UE while staying small enough to fit comfortably in a paging record. The PLMN ID and MMEGI only matter when the identity is interpreted outside its home context — for example when a UE presents an old GUTI to a new MME that must work out which old MME to fetch the context from.

⚠ Common pitfalls / gotchas

  • Treating the C-RNTI as a subscriber identity. It is a cell-local, 16-bit radio label that changes on every handover and vanishes at idle — the core tracks the subscriber with the GUTI/S-TMSI, not the C-RNTI.
  • Confusing PCI with ECGI. The PCI (1 of 504) is reused all over the network and is only a physical-layer scrambling seed; the ECGI is the globally unique name. Neighbour planning bugs (PCI collision/confusion) come from forgetting the PCI is reusable.
  • Assuming the TAC in a TAI is the same TAC as in an IMEI. They are unrelated: Tracking Area Code (location) versus Type Allocation Code (device model). Same three letters, different worlds.
  • Forgetting that usable EPS Bearer ID values start at 5 — EBIs 0–4 are reserved, so a "bearer 5" is the first real bearer, not the sixth.
  • Reading the eNB ID width wrong: a macro eNB ID is 20 bits (8 bits left for the cell), but a Home eNB uses all 28 bits of the ECI for itself. Parsing an ECGI with the wrong split misidentifies both node and cell.

Master Reference

Pulling the four families together, the useful axes are the domain, who assigns the identifier, and what it is for. Scan this when you need a one-line reminder of where any LTE ID sits.

IdentifierDomainAssigned byPurpose
IMSISubscriberHome operator (on USIM)Permanent subscription key
IMEI / IMEISVUE hardwareDevice manufacturerEquipment identity / theft & software-version check
GUTISubscriberMMETemporary identity hiding the IMSI
S-TMSI / M-TMSISubscriberMMEShort IDs for paging / Service Request
C-RNTIRadioeNB (MAC)Address a connected UE on PDCCH
RA-RNTIRadioDerived from PRACH slotAddress the RAR (Msg2)
P-RNTI / SI-RNTI / M-RNTIRadioFixed by spec (FFFE/FFFF/FFFD)Paging / system info / MBMS notification
SPS-C-RNTI / TPC-RNTIs / G-RNTIRadioeNBSPS, group power control, MBMS group data
PCICell (radio)Operator RF planningPhysical-layer cell label (PSS/SSS)
ECGI / ECI / eNB IDCell (global)OperatorGlobally unique cell / eNB name
TAI / TACLocationOperatorPaging area / tracking-area update
PLMN IDNetworkRegulator + operatorIdentify the operator (MCC+MNC)
GUMMEI / MMEGI / MMECNetwork (core)Operator / MME planningIdentify an MME and its pool
EPS Bearer ID / LBISessionMMEEnd-to-end QoS pipe / link to default bearer
E-RAB ID / DRB IDSession (radio)MME / eNBS1+radio bearer / radio-only bearer
TEIDSession (core)SGW / PGWGTP-U tunnel endpoint routing
eNB / MME UE S1AP IDConnectioneNB / MMEIdentify the UE's S1 signalling association
🔀

LTE ↔ NR: the four-domain structure carries straight into 5G, with renamed members. The GUTI becomes the 5G-GUTI and the S-TMSI becomes the 5G-S-TMSI; the GUMMEI/MMEC become the GUAMI/AMF Set ID+AMF Pointer since the AMF replaces the MME. The permanent IMSI becomes the SUPI, and — a genuine new idea — it is never sent even once in the clear: it travels as the public-key-encrypted SUCI. On the radio, the C-RNTI and friends survive almost unchanged; the cell's ECGI becomes the NCGI (with a longer gNB ID), PCI grows to 1008 values, and bearers give way to QoS flows identified by the QFI, mapped into DRBs by the new SDAP layer while TEIDs still route the GTP-U tunnels.

Q&A Quick Q&A

Q. Why does paging use the S-TMSI instead of the full GUTI or the IMSI?

A. The IMSI is permanent and must stay off the air for privacy; the full GUTI is unnecessarily large. The S-TMSI (MMEC + M-TMSI, 40 bits) is compact yet still unique within the MME pool serving the tracking area, so it is the efficient, privacy-safe choice for paging and Service Request.

Q. What is the difference between an eNB ID, an ECI and an ECGI?

A. The eNB ID (20 bits, macro) names the base station within a PLMN. Append an 8-bit Cell ID and you get the 28-bit ECI, which names one cell within the PLMN. Prefix the PLMN ID and you get the ECGI, the globally unique cell name. They nest: eNB → cell in PLMN → cell in the world.

Q. How is a RA-RNTI different from the other RNTIs in how it is obtained?

A. It is not signalled or pre-assigned; it is computed identically by UE and eNB from the PRACH resource used: RA-RNTI = 1 + t_id + 10 × f_id, where t_id/f_id are the subframe and frequency index of the preamble. That lets the UE know which PDCCH carries its Random Access Response without any prior handshake.

Q. A UE hands an old GUTI to a brand-new MME. What in the GUTI lets the new MME recover the UE's context?

A. The GUMMEI part — specifically the PLMN ID + MMEGI + MMEC — identifies exactly which old MME holds the context, so the new MME can contact it and fetch the UE's state. The M-TMSI then identifies the UE within that old MME. This is why the GUTI carries MME routing information and not just a random number.

Summary

The LTE "alphabet soup" is really four small vocabularies, one per question. Who: the permanent IMSI (subscription) and IMEI (hardware) stay hidden, shielded by the MME-assigned GUTI and its slices S-TMSI and M-TMSI. Which message for whom: 16-bit RNTIs address the PDCCH by masking the DCI's CRC — the assigned C-RNTI for a connected UE, the derived RA-RNTI, and the fixed broadcast constants P-RNTI/SI-RNTI/M-RNTI. Where: the reusable physical-layer PCI found from PSS/SSS, versus the globally unique ECGI (with its embedded eNB ID) and the TAI/TAI-list read from SIB1. What flow: one QoS pipe wearing four names — EPS Bearer, E-RAB, DRB and a pair of TEIDs — plus the S1AP ID pair that keeps the eNB and MME pointing at the same UE.

The through-line is that every temporary identity exists to protect, shorten, or localise a permanent one, and the identifiers nest so a smaller label is a slice of a larger — M-TMSI inside S-TMSI inside GUTI, eNB ID inside ECI inside ECGI. Learn which domain an ID serves and its assigner, scope and lifetime follow automatically — and every one of these ideas re-appears, renamed, in the 5G identifier set.

Where these identifiers come from and go

The radio and location IDs are learned during cell acquisition, the subscriber IDs during attach and authentication, and the bearer IDs when data pipes are established — so identifiers are the thread running through every LTE procedure. Follow any one of them into its home procedure next:

Cell Search — how PSS/SSS give the PCIAttach & Authentication — how the IMSI becomes a GUTIEPS Bearers — how EPS Bearer / E-RAB / DRB / TEID are created