Previously, we saw that, to emit a transmission request, terminals must first register with the eNodeB. In this video, we’ll see how that works. In a network, rules must define how several users share the same channel preventing interferences. These rules are called access protocols. They serve to define who can talk, when, and how. There are 2 groups of access protocols. In the first group, “contention-based access” protocols, we speculate that the chances of a collision are very small. Everyone can transmit whenever they want. Because there can still be interferences, we create a way to know whether the message was received or not. For example, using acknowledgements. If a transmission fails, we can try again later. For example Wi-Fi works like this. The other group is called “reservation-based access” protocols. The resource is broken into small elements that are allocated to senders. For example, each user will have the right to a turn to talk for one millisecond. This allocation can be static, when it is defined at the start; or dynamic - it can be modified according to needs. LTE belongs to this group. And we have already seen that the resources are divided by time and by frequency and that they are allocated dynamically. On the downlink, there’s no problem, since all transmissions are made by the eNodeB. On the uplink, terminals must first send a request before being allocated resources. These requests go on the uplink control channel on which each terminal has the right to a turn to speak. This assumes that terminals are registered with the base station. For that, they must have been able to transmit once to make themselves known. The solution in LTE was to introduce a bit of contention within the uplink. On the uplink, a group of resources is reserved so that new terminals can indicate their presence. This group of resources constitutes a new physical channel called PRACH, or Physical Random Access Channel. It is made up of 6 adjacent resource blocks pairs and occurs every one to 20 milliseconds depending on the eNodeB parameters. The access to this channel is done using a particular access technique that is called CDMA Code Division Multiple Access. In CDMA, transmitters encode their transmissions with sequences. The mathematical properties of these sequences are such that, if 2 devices transmit simultaneously with different sequences, the receiver can still distinguish between the two transmissions. Another detail about this channel: on the allocated one millisecond, only one symbol is transmitted and this symbol uses all 6 resource blocks. So, when a terminal transmits on this channel, it actually send only one symbol. But, the base station detects that something was transmitted, and infers that a new terminal wants to connect. Even though the terminal does not transmit data, the base station can identify the particular sequence used by the mobile and associate this terminal with this sequence. There are 64 sequences available, so, there could be up to 64 simultaneous requests. We will explain later what happens if 2 users use the same sequence at the same time. Let's first see what happens when a new terminal enters the cell. We saw previously that, to allocate resources, the eNodeB gives each terminal an RNTI. This is the procedure: the terminal that arrives listens to the eNodeB to identify when the Random access channel is scheduled. It then chooses, at random, one of the 64 sequences, and transmits on the random-access channel, using this sequence at the right moment. The eNodeB detects this transmission and the sequence. It does not yet know the identity of the mobile, but will use this sequence to temporarily identify the terminal and will then assign it an RNTI. However, at this point this RNTI is only known by the eNodeB. So, the eNodeB cannot use it as an address to send a message to the terminal. Instead, it will use another identifier: the RA-RNTI or Random Access RNTI. This identifier is calculated using the sequence already used by the mobile. So, the terminal, just like the eNodeB, can calculate this RA-RNTI. Now, the eNodeB transmits a message to the mobile to send a few parameters, including the assigned RNTI and it will use the RA-RNTI as the destination address of this message. The eNodeB also allocates a resource to that terminal so that it can send a connection request. When the terminal receives this message, it can transmit its transmission request on the allocated resource using the given RNTI. The rest of the connection procedure is described in another week of this MOOC. I said I would explain what happens when 2 UEs access the contention channel with the same sequence. Each terminal goes through the random-access procedure thinking it is the only one. The eNodeB will note one request, thinking it is the only one and will then go through the procedure we just saw: it will assign an RNTI and allocate a transmission resource, and send this information to the mobile. The 2 phones will receive this message thinking it is meant for them, each will transmit a connection request and an interference will occur. As for interference, 2 cases are possible. In the first case, the base station is not able to decode either of the messages and will not respond. The terminals will then repeat the procedure later. In the other case, the base station can decode one of the messages. That’s what we see here. It believes there is only one user and continues the procedure. This is a problem because we don’t know which of the 2 terminals was registered. To solve this problem, the standard dictates that the eNodeB must send to the UE an echo of its connection request. Because the connection request contains the identity of the mobile the terminal can compare its identity to the one in this copy. If it is its own identity, it continues the connection procedure. Otherwise it stops the procedure and starts again later. In a nutshell, LTE is a reservation-based protocol. Allocation is managed dynamically by the eNodeB. The registered terminals have the right to use the control channel to send their transmission requests. To register in a new cell, the terminals use the random-access channel. This channel is contention based. And a protocol is used to resolve ambiguities resulting from collisions.
