A UE cannot transmit on PUSCH until the eNB gives it a grant, and the eNB cannot give a grant to a UE it has not heard from. Random Access is the procedure that starts from neither side knowing the other. The UE sends a preamble on a resource the cell has broadcast in advance, and the eNB answers on a resource the UE can calculate from the preamble it just sent. Everything else in the procedure follows from those two facts.

This page stays at the overview level. The full treatment, including the preamble formats, the PRACH parameters and a live capture of each case, sits on the RACH_LTE page.

• Random Access
• What are the four messages, and why is the fourth one needed ?
• When is the procedure contention-free ?

Random Access

Random Access process plays two main roles - establishment of uplink synchronization and establishment of a unique UE ID (C-RNTI) known to both the network and the UE.So Random Access is used not only for initial access, but also after periods of uplink inactivity when uplink sync got lost in LTE_ACTIVE states.

• i) UE initiate a Random Access Procedure on the (uplink) Random Access Channel (RACH).(The location of RACH in the frequency/time resource grid the RACH is known to the mobile via the (downlink) Broadcast Channel (BCH). The random access message itself only consists of 6 bits and the main content is a random 6 bit identity)
• ii) Network sends a Random Access Response Message(RARM) at a time and location on the Physical Downlink Shared Channel (PDSCH) (The time and location of RARM on PDSCH can be calculated from the time and location the random access message was sent. This message contains the random identity sent by the device, a Cell Radio Network Temporary ID (C-RNTI) which will be used for all further bandwidth assignments, and an initial uplink bandwidth assignment)
• iii) The mobile device then uses the bandwidth assignment to send a short (around 80 bits) RRC Connection Request message which includes it's identity which has previously been assigned to it by the core network

Only the step i) uses physical-layer processing specifically designed for random access. The remaining steps utilizes the same physical layer processing as used for normal uplink and downlink data transmission.

For more detailed explanation about this procedure, refer to RACH_LTE

What are the four messages, and why is the fourth one needed ?

Random Access is usually summarised in three steps, and that summary leaves out the fourth message. After the third message the eNB knows a UE is there. But the UE does not yet know whether the eNB answered it or answered somebody else who happened to pick the same preamble. A fourth message settles that question, and until it arrives the C-RNTI is only provisional.

Figure 1 shows the four messages in order, with the UE on the left and the eNB on the right. Each arrow carries the name of the message and the channel it travels on, and the note beside it names the identity that addresses it. Read those notes first. The identity changes at every step, and that change is what the procedure is really doing.

Figure 1. The contention-based Random Access sequence. The addressing identity changes at every step: no identity at all in Msg1, then an RA-RNTI shared by everyone in the PRACH occasion, and finally a C-RNTI that belongs to one UE.

• Msg1 is the only message with physical layer processing built for Random Access. Msg2, Msg3 and Msg4 use PDSCH and PUSCH exactly like ordinary data, which is the point the Random Access section already makes.
• The UE picks its preamble at random from 64 signatures, so two UEs in the same cell can pick the same one in the same PRACH occasion. The eNB cannot distinguish them from Msg1 alone.
• Msg2 is addressed by RA-RNTI, and the RA-RNTI comes from the time and frequency of the PRACH occasion rather than from the UE. So every UE in that occasion reads the same Msg2 and takes the same Temporary C-RNTI.
• Msg3 is the first message that carries the UE’s own identity. Every colliding UE sends Msg3 on the same grant, so the eNB decodes at most one of them.
• Msg4 echoes one identity back. Only one UE finds its own identity there, and that UE keeps the Temporary C-RNTI as its C-RNTI. The others return to Msg1 and try again.
• A contention-free procedure stops after Msg2. The eNB chose the preamble and told the UE which one to use, so two UEs cannot collide and Msg2 carries the real C-RNTI.

This is why the Temporary C-RNTI is called temporary. The eNB allocates it to a PRACH occasion rather than to a UE, and several UEs can hold the same one at the same time. It becomes a real C-RNTI only for the UE whose identity comes back in Msg4. The Random Access section describes the C-RNTI as known to both the network and the UE, and that becomes true at Msg4 rather than at Msg2.

• The fourth message is what makes the identity unique : Msg2 hands the same Temporary C-RNTI to every UE in the occasion, and only Msg4 reduces that to one UE.
• Only Msg1 is special at the physical layer : the other three use the same processing as ordinary traffic. So a failure after Msg1 is usually a scheduling or coverage problem rather than a RACH problem.
• The design expects collisions rather than preventing them : the UE picks one of 64 signatures at random, so two UEs can collide and the procedure resolves that afterwards.

When is the procedure contention-free ?

Collisions are possible only when the UE chooses its own preamble. When the eNB names the preamble instead, it reserves that signature for one UE, and no other UE will send it. So the question of contention is really a question about who chose the preamble, and that depends on what started the procedure.

Six situations start Random Access in LTE. Who picks the preamble is the column to read first, because it decides whether contention is possible at all.

Two things follow from the table. A contention-free procedure needs the UE to be in RRC_CONNECTED already, because the eNB has to reach it with the preamble assignment before the procedure starts. And initial access can never be contention-free, because there is no connection yet on which to send that assignment.

The saving is not only the fourth message. A contention-free procedure can put the real C-RNTI in Msg2, so the UE becomes addressable one round trip earlier. That is why handover uses it whenever the target cell has a dedicated preamble available, since the handover interruption is measured in milliseconds.

The RACH_LTE page covers each of these cases as a separate call flow. The trigger list is under When RACH Process occurs ?, and the two procedures are compared under Two types of RACH process.

• Contention-free needs a connection that already works : the eNB has to deliver the preamble assignment somehow, so a UE in RRC_IDLE always uses the contention-based procedure.
• Handover is the main contention-free case : the target eNB names the preamble in the handover command, which removes both the collision risk and Msg4 from the interruption time.
• Losing uplink sync does not mean losing the C-RNTI : a connected UE keeps its C-RNTI when it runs Random Access again, and what it is missing is a valid Timing Advance.