In LTE, timers and constants such as T310, N310, and N311 are critical parameters used in the management of radio link failure (RLF). They help the UE (User Equipment) monitor and recover from physical layer issues in RRC_CONNECTED state.

The diagram below shows the whole mechanism on one time line. The lower layer of the UE sends out-of-sync indications to the higher layer. After N310 of them in a row, the higher layer starts T310. In-sync indications may then arrive, and N311 of them in a row stop T310 before it expires. The two notes under the diagram give the author's simplified meaning of one indication.

T310 started by N310 consecutive out-of-sync indications and stopped by N311 consecutive in-sync indications. RRC in the higher layer counts the indications, and the physical layer only reports them.

Note 1 : one "Out of Sync Indication" in this diagram means "20 subframes of consecutive PDCCH decoding failure.

Note 2 : one "In Sync Indication" in this diagram means "10 subframes of consecutive PDCCH decoding success.

A note on the two notes above. They are a simplification for the diagram. 36.133 clause 7.6 does not define an indication by a count of decoding failures. It defines it by the BLER of a hypothetical PDCCH over an evaluation window, as the last section of this page explains.

Followings are the topics to be covered in this page.

• T310
• N310
• N311
• Radio Link Failure Management Workflow
• Reconfiguration of T310, N310, and N311
• Interaction Between T310, N310, and N311
• How does the physical layer decide out-of-sync and in-sync?
• Reference

T310

T310 is a timer started by the UE upon detecting consecutive "out-of-sync" indications for the PCell (Primary Cell) from lower layers (physical layer). It supervises the time the UE waits for conditions to improve or a recovery response from the network before declaring radio link failure.

When T310 is Started:

T310 starts only in RRC_CONNECTED and only for the PCell. The UE does not start it while another procedure already supervises the connection, because that procedure has its own timer and its own failure handling.

• Upon receiving N310 consecutive "out-of-sync" indications for the PCell while:
• T300, T301, T304, T311, or T316 is not running.
• For DAPS (Dual Active Protocol Stack) handovers, T304 is running, and N310 consecutive "out-of-sync" indications are received for the source PCell.

When T310 is Stopped:

Recovery through N311 is the normal way T310 stops. It is not the only way, though. Any procedure that replaces the current PCell link also makes T310 unnecessary, so the UE stops it at that point.

• Upon receiving N311 consecutive "in-sync" indications for the PCell from the lower layers while T310 is running.
• This indicates recovery of the radio link without the need for additional signaling.
• T310 is also stopped upon triggering the handover procedure, upon initiating the connection re-establishment procedure, and upon initiating the MCG failure information procedure.
• In a discontinuous coverage scenario, T310 is also stopped upon expiry of t-Service or when the UE is out of the current serving cell coverage.

Expiry of T310:

T310 expiry is the most common trigger of radio link failure, but it is one of several. A random access problem, the maximum number of RLC retransmissions and T312 expiry lead to the same RLF handling. What the UE does next depends on its security state and its dual connectivity configuration.

• If T310 expires, the UE declares radio link failure (RLF) and:
• Initiates the RRC connection re-establishment procedure as per clause 5.3.7.
• Stores relevant RLF information for reporting (e.g., measurements of the failed cell and neighboring cells).

Let's be exact about the recovery path. 36.331 clause 5.3.11.3 initiates re-establishment only when AS security is active. Without AS security, the UE leaves RRC_CONNECTED with release cause 'other', apart from some NB-IoT cases with CIoT optimisation. With (NG)EN-DC and T316 configured, and with the SCG still working, the UE sends MCGFailureInformation over the SCG instead of re-establishing. In every case the UE stores the failure in VarRLF-Report with connectionFailureType set to rlf. 36.133 clause 7.6.2.1 also requires the UE to turn off its transmitter within 40 ms after T310 expiry.

• T310 expiry declares RLF on the PCell : it is one of several RLF triggers in clause 5.3.11.3.
• Re-establishment needs AS security : without it the UE goes to RRC_IDLE.
• MCG failure information can replace re-establishment : with (NG)EN-DC, T316 and a working SCG.

N310

N310 is a configurable constant that specifies the number of consecutive "out-of-sync" indications the UE must receive from the lower layers before taking action (e.g., starting T310).

Behavior:

N310 filters out short fades. A single bad evaluation period does not start T310, because the UE needs N310 indications in a row. The larger N310 is, the longer the link must stay bad before the UE even starts to count time.

• Upon receiving N310 consecutive "out-of-sync" indications:
• The UE considers the radio link to be in a degraded state.
• Starts T310 to monitor whether the situation recovers within the timer's duration.

The count must be consecutive, but gaps do not reset it. 36.331 clause 5.3.11.2 notes that periods in which layer 1 reports neither in-sync nor out-of-sync do not affect the count. Only an indication of the opposite kind breaks a run.

N310 has a twin for dual connectivity. The PSCell uses N313 and T313 in the same way, and the UE starts T313 after N313 consecutive out-of-sync indications for the PSCell, while T307 is not running. T313 expiry leads to SCG failure, not to RLF of the whole connection.

• N310 counts consecutive out-of-sync indications : from 1 to 20, broadcast in SIB2 or set in rlf-TimersAndConstants.
• A silent period does not reset the count : only an in-sync indication does.

N311

N311 is a configurable constant that specifies the number of consecutive "in-sync" indications the UE must receive from the lower layers to consider the radio link as recovered.

Behavior:

N311 protects the UE from a false recovery. One good evaluation period is not enough, so the link must look good N311 times in a row. Then the UE keeps its RRC connection as if nothing had happened.

• Upon receiving N311 consecutive "in-sync" indications while T310 is running:
• The UE stops T310.
• The UE considers the radio link to have recovered and maintains the current RRC connection without additional signaling.

Recovery also stops T312, if it is running. T312 is the Release 12 timer for early RLF detection. The UE starts it only while T310 is running, when a measurement report is triggered for an event with useT312 set to true. So a recovery through N311 cancels both timers at once, and the UE keeps the entire radio resource configuration.

For the PSCell, N314 plays the role of N311. N314 consecutive in-sync indications for the PSCell stop T313 while it is running. The two pairs are counted separately, so the PCell and the PSCell can be in different states at the same time.

• N311 counts consecutive in-sync indications : from 1 to 10, and it matters only while T310 is running.
• Recovery needs no signalling : the eNB does not learn that T310 was running.

Radio Link Failure Management Workflow

Let's put the three parameters together as one process. The UE moves through three phases: it detects a degraded link, it waits for recovery, and it either recovers or declares RLF. Each phase below names the parameter that controls it.

Detection of Degraded Link - N310 and T310:

The first phase starts in the physical layer. Layer 1 compares the downlink quality with the Qout threshold and reports out-of-sync to RRC. RRC counts the reports and starts T310 when the count reaches N310.

• The UE monitors the "sync" status provided by the physical layer:
• "Out-of-sync": The link quality between the UE and the PCell is insufficient.
• Upon N310 consecutive "out-of-sync" indications, the UE starts T310 to supervise the degraded state.

Recovery Monitoring - N311:

The second phase is a race between two events. The link may improve so that N311 in-sync indications arrive in a row, or T310 may reach its configured value first.

• While T310 is running, the UE continues to monitor for recovery indications:
• "In-sync": Indicates improved link quality.
• If the UE receives N311 consecutive "in-sync" indications, T310 is stopped, and the radio link is considered recovered.

Radio Link Failure - RLF:

The third phase happens only if T310 wins the race. From here the UE no longer tries to save the current link. It tries instead to save the connection, through re-establishment or through MCG failure information.

• If T310 expires without receiving N311 consecutive "in-sync" indications:
• The UE declares radio link failure.
• It initiates the RRC connection re-establishment or takes appropriate failure-handling actions.

Reconfiguration of T310, N310, and N311

The values for T310, N310, and N311 can be dynamically reconfigured by the network via the rlf-TimersAndConstants parameter. This allows the network to adjust failure detection sensitivity based on conditions like load, mobility, or interference.

• Release Scenario:
• If rlf-TimersAndConstants is set to release:
• The UE uses the default values specified in ue-TimersAndConstants in SystemInformationBlockType2.
• Reconfiguration Scenario:
• If rlf-TimersAndConstants provides new values:
• The UE updates the timers and constants to the received values.

There are two sources of the values, and the listings below show both. The first is ue-TimersAndConstants in SystemInformationBlockType2, which every UE in the cell reads.

Following is based on 36.331 v19.3.0 (Release 19)

UE-TimersAndConstants ::=           SEQUENCE {
    t300                                ENUMERATED {
                                            ms100, ms200, ms300, ms400, ms600, ms1000, ms1500,
                                            ms2000},
    t301                                ENUMERATED {
                                            ms100, ms200, ms300, ms400, ms600, ms1000, ms1500,
                                            ms2000},
    t310                                ENUMERATED {
                                            ms0, ms50, ms100, ms200, ms500, ms1000, ms2000},
    n310                                ENUMERATED {
                                            n1, n2, n3, n4, n6, n8, n10, n20},
    t311                                ENUMERATED {
                                            ms1000, ms3000, ms5000, ms10000, ms15000,
                                            ms20000, ms30000},
    n311                                ENUMERATED {
                                            n1, n2, n3, n4, n5, n6, n8, n10},
    ...,
    [[  t300-v1310                      ENUMERATED {
                                            ms2500, ms3000, ms3500, ms4000, ms5000, ms6000, ms8000,
                                            ms10000}        OPTIONAL,   -- Need OR
        t301-v1310                      ENUMERATED {
                                            ms2500, ms3000, ms3500, ms4000, ms5000, ms6000, ms8000,
                                            ms10000}        OPTIONAL    -- Need OR
    ]],
    [[  t310-v1330                          ENUMERATED {ms4000, ms6000}
                                                            OPTIONAL    -- Need OR
    ]],
    [[  t300-r15                        ENUMERATED {ms4000, ms6000, ms8000, ms10000, ms15000,
                                            ms25000, ms40000, ms60000}  OPTIONAL        -- Cond EDTorPUR
    ]]
}

The second is rlf-TimersAndConstants in RadioResourceConfigDedicated, which the eNB sends to one UE. The Release 9 IE carries all five values. The Release 13 IE adds the longer t301-v1310 and t310-v1330, and E-UTRAN configures it only if the UE supports ce-ModeB.

Following is based on 36.331 v19.3.0 (Release 19)

RLF-TimersAndConstants-r9 ::=           CHOICE {
    release                                 NULL,
    setup                                   SEQUENCE {
        t301-r9                             ENUMERATED {
                                                ms100, ms200, ms300, ms400, ms600, ms1000, ms1500,
                                                ms2000},
        t310-r9                             ENUMERATED {
                                                ms0, ms50, ms100, ms200, ms500, ms1000, ms2000},
        n310-r9                             ENUMERATED {
                                                n1, n2, n3, n4, n6, n8, n10, n20},
        t311-r9                             ENUMERATED {
                                                ms1000, ms3000, ms5000, ms10000, ms15000,
                                                ms20000, ms30000},
        n311-r9                             ENUMERATED {
                                                n1, n2, n3, n4, n5, n6, n8, n10},
        ...
    }
}

RLF-TimersAndConstants-r13 ::=          CHOICE {
    release                                 NULL,
    setup                                   SEQUENCE {
        t301-v1310                              ENUMERATED {
                                                    ms2500, ms3000, ms3500, ms4000, ms5000,
                                                    ms6000, ms8000, ms10000},
        ...,
        [[  t310-v1330                          ENUMERATED {ms4000, ms6000} OPTIONAL    -- Need ON
        ]]
    }
}

T310 runs from ms0 to ms2000, N310 from n1 to n20 and N311 from n1 to n10. A value of ms0 means that RLF follows directly after N310 out-of-sync indications. The extension t310-v1330 adds ms4000 and ms6000. E-UTRAN includes t3xy-v1330 only in the BR version of the SIB, and a UE that supports CE mode B uses it instead of t310. As long as a dedicated configuration is in place, the UE does not update these values from SIB2, except T300.

• SIB2 gives the cell default : ue-TimersAndConstants applies until a dedicated configuration arrives.
• rlf-TimersAndConstants overrides it per UE : release returns the UE to the SIB2 values.
• CE mode B gets longer T310 values : ms4000 and ms6000 through t310-v1330.

Interaction Between T310, N310, and N311

The three values set the sensitivity of RLF detection together, so the network cannot tune one of them in isolation. The list below states the role of each one, and the paragraph after it works through a numeric example.

• N310 and N311 thresholds act as triggers to start and stop T310.
• T310 acts as a time boundary:
• Ensures that the UE doesn't wait indefinitely for recovery when N310 is met.
• Allows sufficient time for N311 to indicate recovery when conditions improve.

Let's take an example. With N310 = n10 and no DRX, the UE needs ten out-of-sync reports before T310 starts. Layer 1 sends the first one only after the 200 ms Qout window has become bad, and two reports are at least 10 ms apart. With T310 = ms1000 the UE then waits up to one more second. So the time from the start of a deep fade to RLF is roughly the Qout window, plus N310 indication periods, plus T310. A larger N311 makes recovery slower but safer, because every in-sync report needs a 100 ms Qin window that is good.

• N310 decides when to start counting time : it adds delay before T310 starts.
• T310 decides how long to wait : it adds delay before RLF is declared.
• N311 decides how sure the UE must be : it adds delay before recovery is accepted.

How does the physical layer decide out-of-sync and in-sync?

RRC counts indications, but it never measures the radio link itself. The physical layer does that job, and 36.133 clause 7.6 sets the rules. The UE monitors the downlink quality of the PCell on the cell-specific reference signal. It then compares the quality with two thresholds that are defined through a hypothetical PDCCH, not through a real one.

Qout is the level at which the downlink cannot be received reliably. It corresponds to 10% BLER of a hypothetical PDCCH with DCI format 1A. Qin is the level at which reception is clearly more reliable. It corresponds to 2% BLER of a hypothetical PDCCH with DCI format 1C. Both include the PCFICH errors. The two thresholds are far apart on purpose, so that a link near the boundary does not switch between the two states all the time.

Without DRX, layer 1 evaluates Qout over the last 200 ms and Qin over the last 100 ms. Two successive indications are at least 10 ms apart. With DRX the evaluation periods become longer and depend on the DRX cycle, as 36.133 Table 7.6.2.2-1 specifies. So the same N310 value means a longer detection time for a UE in a long DRX cycle.

• Out-of-sync means worse than 10% hypothetical PDCCH BLER : evaluated over 200 ms without DRX.
• In-sync means better than 2% hypothetical PDCCH BLER : evaluated over 100 ms without DRX.
• DRX stretches the evaluation : RLF detection is slower for a UE in a long DRX cycle.

Reference

[1] 3GPP TS 36.331 v19.3.0 - clause 5.3.11, Radio link failure related actions, clause 6.3.6, UE-TimersAndConstants, and clause 7.3, Timers

[2] 3GPP TS 36.133 v19.5.0 - clause 7.6, Radio link monitoring