4G/LTE - Basic Procedures

 

 

 

Initialization Sequence : From Power-On to PRACH

 

Following procedure is to list each steps from Power-On to Initial PRACH. Even though it looks simple, each of these steps can be one of the most complicated procedures happening within the UE. Even more difficult is that a lot of these steps are not specified clearly in 3GPP spec and large portions are up to UE implementation. In addition, if you are a validation/verification engineer of mobile device .. you would have a lot of difficulties since not so many detailed log for this procedure is avaiable, even though it is available it would take long time and practice until you get familiar with those logging.

The Steps

The list below runs from a powered-off UE to its first RACH preamble. It holds one line per step, and the link on each line leads to the page that covers that step in full. What the list carries is the order. The two notes underneath it say where that order can loop.

  •   i) UE is Off
  •   ii) Power On UE
  •   iii) < Frequency Search >
  •   iv) Time and Frame Synchronization: In this process,this process can be broken down to multiple steps as below.     a) detect PSS and get the timing sync for subframe and estimate frequency offset.     b) with using the timing sync, you may construct the resource grid for 6RBs at the center and detect SSS
  •   v) Based on PSS and SSS, you can calculate PCI (Physical Cell ID)
  •   vi) MIB decoding : UE can figure out System Bandwidth in this process. (As you see in Downlink Framestructure, MIB/PBCH is located at the 6 RBs around the center frequency. So the success of MIB decoding does not guarantee that signal quality across the whole band is good)
  •   vii) With the PCI and Bandwidth, detect CSR (Cell Specific Reference Signal) and perform Channel Estimation and Equalization. In this process, UE will detect/measure reference signal across the whole system bandwidth. So RSRP/RSRQ measured at this step can be a good indicator for overall signal quality.
  •   viii) Decode PDCCH and extract DCI information for SIB. PDCCH is spread across the whole bandwidth, so the signal quality across the whole bandwidth should be good enough for this step.
  •   viii) SIB deconding (SIB1 should be decoded first and then SIB2 and then remaining SIBs)
  •   ix) < Cell Selection > : UE may find multiple suitable cells, but it try camp on to HPLM cell with the highest priority
  •   x) < Initial RACH Process >

Note 1 : UE may repeat the step iii)~viii) multiple times before it finds a suitable cell. This sequence is part of Cell Search process.

Note 2 : PCI (step v) will be automatically derived as a result of step iv).

  • The list is a chain rather than a sequence : each step consumes something the step before it produced.
  • A step that fails is often a step whose input never arrived : the useful question is which earlier output the failing step needed.

What Each Step Hands to the Next

Read as a list, these are ten things to do in order. Read as a chain, they are ten things that each need something the previous step produced. The second reading is the useful one when a step fails. It turns a question about SIB decoding into a question about frame timing.

The table below gives each step its output and names what later needs it.

Step

What it produces

What needs it later

Frequency Search

A carrier worth looking at

Everything after it

PSS detection

Subframe timing, and the identity within the group

The SSS search window, and the PCI

SSS detection

Frame timing, and the identity group number

The PCI, and the MIB window

PCI

The CRS sequence and where it sits in the grid

Channel estimation, and PDCCH descrambling

MIB decoding

dl-Bandwidth, and the frame number

The full resource grid, and the SIB1 window

CRS and channel estimation

A channel estimate across the band

Equalization of PDCCH and of PDSCH

PDCCH and DCI

The allocation that carries SIB1

SIB decoding

SIB1

The scheduling of the other SIBs, and the cell access information

SIB2, and cell selection

SIB2

The PRACH configuration

The initial RACH

The timing chain is the tightest part

Two scheduling rules in 36.331 clause 5.2.1.2 fix the shape of the middle of this chain. The MIB uses a fixed 40 ms period, and its first transmission sits in subframe #0 of the radio frames where SFN mod 4 = 0. SIB1 uses a fixed 80 ms period, and its first transmission sits in subframe #5 of the radio frames where SFN mod 8 = 0.

Both rules are written in terms of the frame number. A UE that does not yet know the frame number cannot tell which subframe #5 is the right one. That is the dependency the list hides. SIB1 stays out of reach until the MIB has been read, and the unknown bandwidth is only half the reason.

The MIB closes that gap itself, in a way worth knowing. Its systemFrameNumber field is eight bits wide while the SFN runs to ten. The two missing bits come from which of the four transmissions inside the 40 ms window the UE managed to decode.

The identity chain is shorter than it looks

Note 2 above says the PCI is derived from the synchronization step rather than measured on its own, and 36.211 clause 6.11 is why. There are 504 physical-layer cell identities, grouped into 168 groups of three. The primary synchronization signal carries the identity within a group, and the secondary one carries the group number. A UE that has detected both already holds the PCI.

That matters for the step after it. The cell-specific reference signal sequence and its position in the grid both depend on the PCI, so channel estimation cannot start until the identity is settled. An error in SSS detection therefore does not appear as an identity error. It appears as a channel estimate that will not converge.

  • The middle of the chain is held together by the frame number : 36.331 schedules both the MIB and SIB1 against the SFN, so neither can be found before timing is settled.
  • A wrong PCI does not announce itself : it surfaces one step later, as channel estimation that fails on a cell the UE can otherwise hear.

Why the Bandwidth Bar Rises as the Sequence Runs

The note beside the MIB step makes a point that deserves more room than it gets. Decoding the MIB proves the centre of the band is good and proves nothing about the rest. Each step that follows asks for more of the spectrum than the one before it. That is what makes this sequence a diagnostic rather than a pass or a fail.

The table below puts each step beside the part of the carrier it needs.

Step

Where the signal sits

Clause in 36.211

PSS and SSS

The centre of the carrier

Clause 6.11

MIB, on PBCH

The six resource blocks at the centre

Clause 6.6

CRS

The whole system bandwidth

Clause 6.10.1

PDCCH

The whole system bandwidth

Clause 6.8

SIB, on PDSCH

Wherever the DCI allocates it

Clause 6.4

The first two rows need only the middle of the carrier, and they have no choice. The UE does not know how wide the band is at that point, so the specification cannot ask for more than the centre. The system bandwidth arrives in the MIB. Every row below it can therefore spread across whatever that bandwidth turned out to be.

That split produces a failure signature worth recognising. A UE that reaches the MIB and then stops has good signal in the centre six resource blocks and poor signal somewhere else in the band. Nothing about the first three steps would have warned of it.

The step between the two halves is the one that fails most visibly. PDCCH spreads its control channel elements across the full bandwidth. The UE therefore has to equalize the whole band before it can find the DCI pointing at SIB1. A carrier with one damaged edge can pass every step up to the MIB and fail at exactly this one.

The note beside the channel estimation step makes the other half of the argument. RSRP and RSRQ are measured on the reference signal, and the reference signal spans the whole bandwidth. Those two numbers therefore describe the band, while a successful MIB decode describes only its centre.

These steps have been run on captured data elsewhere on this site

The steps in this list are abstract by design, and two other notes carry them out on real signal. The LTE PHY DSP note and the Downlink RX note work through those steps on baseband I/Q captured from an eNB. They cover PSS detection, SSS detection, the resource grid, channel estimation and equalization. The pictures for these steps live there rather than here.

  • The sequence is a ladder of bandwidth, not a single test : the first steps need the centre of the carrier and the later ones need all of it.
  • MIB decoded and SIB failed is a band-edge symptom : the centre six resource blocks were good enough and the rest of the band was not.
  • RSRP and RSRQ describe more of the band than a MIB decode does : they are measured on a reference signal that spans the whole bandwidth.

Reference

Three specifications carry the rules quoted above. The clause numbers sit beside each claim so the wording can be checked rather than taken on trust.

  • 36.211 - Evolved Universal Terrestrial Radio Access (E-UTRA); Physical channels and modulation. Clause 6.11 defines the synchronization signals and the 504 physical-layer cell identities in 168 groups of three. Clause 6.6 is PBCH, clause 6.8 is PDCCH, clause 6.10.1 is the cell-specific reference signal and clause 6.4 is PDSCH. The clause headings and the opening text of clause 6.11 were read; the resource element mapping in those clauses is given as equations that the text extraction did not preserve.
  • 36.331 - Evolved Universal Terrestrial Radio Access (E-UTRA); Radio Resource Control (RRC); Protocol specification, v19.3.0. Clause 5.2.1.2 carries the fixed schedules for the MIB and for SystemInformationBlockType1 quoted above. Clause 5.2.2 is system information acquisition. The ASN.1 in clause 6 gives MasterInformationBlock its eight bit systemFrameNumber.
  • 36.304 - Evolved Universal Terrestrial Radio Access (E-UTRA); User Equipment (UE) procedures in idle mode, v19.2.0. Clause 5.2.3 is the cell selection process that the cell selection step above performs, and clause 5.2.3.2 is the criterion it applies.