What are the differences between the 5G and our radio interface and LTE? That is to say, 4G as as regards access and protocol stack. That's what we will see in this video. As with 4G, transmission on the NR radio channel takes the form of blocks of data with one block occupying one or more PRBs. To enable simple identification of the UE on the radio channel, it's allocated an RNTI or Radio Network Temporary Identifier. For every terminal that has an RNTI, at the beginning of each slot The gNB indicates an allocation map. This allocation map consists of a set of allocation messages called downlink control information because they are transmitted from the network to the terminal. For each case, we indicate whether the allocation is downlink or uplink. The number of PRBs is indicated, 4 in this case, these are the number in the frequency space and what's new with 5G is that we indicate the number of symbols in the slot which constitute the PRB. All of the slot symbols can be used for data or as we have seen, just a few symbols for the PRB. Remember that beam forming is possible, as we have already seen. Beam forming is possible for the user data but the DCI messages can be sent over the entire cell or we can have transmission of the allocation message, DCI, also on a narrow beam, sent only to the relevant terminal from this allocation. The problem, since we said that resource allocation depends on the RNTI, is to allocate an RNTI to each terminal that carries out radio exchanges. This procedure is quite similar to that of 4G. Slots are periodically opened at all the terminals. A terminal that wants, for example, to transmit data on the radio channel will transmit a preamble by randomly choosing a preamble among a set of available preambles. The base station allocates an RNTI, which is temporary and will indicate this RNTI in response at the request of the terminal. Collisions are possible. Several terminals might use the same preamble and one may be received correctly by the base station. In this case, several terminals might think they have the same RNTI. To resolve these collisions, the first message is sent by the terminal contains a non-ambiguous identity of the UE. Accordingly, when the network responses, it will use a non-ambiguous identity of the terminal. If the UE detects an identity that is different to its own, it will reapply the principle of random access. If it sees the same identity, then the RNTI is validated. That's what we call the four-step handshake. Let's look at the different layers on the radio interface. We'll start naturally enough with the physical layer. The physical layer enables transmission of transport blocks depending on radio conditions and on the quantity of data to be transmitted. We dynamically adapt the modulation and coding scheme and the quantity of resource allocated. The physical layer will deal with all the signal processing. I mean, beamforming, MIMO, modulation and the error correction coding. On this point, in 4G, we used turbo codes at the error correction code. Whereas in 5G, the choice was made to use LDPC codes or low density parity check for the data. Another type of code called polar codes for the signaling messages. Since these messages are shorter and polar codes are therefore more suitable. Note that PRB contains a set of resource elements, but they are not all available to contain data. There are lots of reference signals to transmit. Above the physical layer we have the MAC layer or Medium Access Control. The MAC layer will transmit the blocks, and for the blocks received, transmit the acknowledgments. When an acknowledgment has not been received, the block is retransmitted until it is successfully received. This is known as HARQ for hybrid automatic repeat request. It's similar to LTE with more flexibility, because with LTE, there is a fixed time interval of four milliseconds between the acknowledgment and the transmission of the data. Here, the duration can be varied. We have a single MAC instance per frequency band. I mean, the MAC layer supplies one and only one transport channel. Above the MAC layer. We have RLC, or Radio Link Control. RLC has three modes. In the mode, we call transparent, the data delivered by the upper layer are transmitted directly without the addition of a header. In the unacknowledged mode, a header is added to detect data loss, but there is no acknowledgment and retransmission mechanism in the case of error. Of course, in acknowledged mode, the lost blocks are retransmitted until they are successfully received. RLC does not support concatenation, nor resequencing, which is specific to 5G. (resequencing was managed by RLC in LTE). At reception, let's consider a data flow where block one, block two, and then block three are transmitted. If blocks one and three are received satisfactorily, but not block two, in LTE, we have to wait until block two has been received for the RLC to deliver block two to the upper layers, and then immediately afterwards, block three. There's delay for block three which was not strictly necessary. In 5G, as soon as block three is correctly received, it is sent to the upper layers even if block two has not yet been received. There are several possible RLC instances. RLC provides what we call logical channels. Above RLC, we find PDCP Packet Data Convergence Protocol. PDCP manages the security function, encryption and integrity control. For PDCP, it's possible to duplicate the packets in a systematic way at transmission to ensure better reliability. Of course, at reception, any duplicate that may exist must be eliminated. PDCP can handle re-sequencing, and will also manage header compression and decompression. Here too, there can be several instances of PDCP simultaneously. The SDAP layer, or service data adaptation protocol, is new in 5G. SDAP will focus on the quality of service. It's capable of implementing rules of priority. If there are too few resources to transmit all the waiting blocks, SDAP will then give priority to the higher priority blocks. To ensure the same level of quality of service thourghout the entire network, including the core network, we add a QoS marker in the SDAP header. Again, several instances of SDAP may occur simultaneously. Note that SDAP is only present in the user plane. We can have a more concise view of the different elements presented. Let's imagine a terminal (a UE), which has a stream for video conferencing with both audio and video paths and at the same time, a file to transmit, as well as signaling messages to transmit. Let's suppose that we transmit everything in the same slot. Thanks to SDAP we can, for example, prioritize the audio over the video. We will thus, have two instances, two flows, and two different QoS. But for audio and video, we will use RLC in the unacknowledged mode, the files must be correctly received. In my example, RLC for the blocks that make up the file, is transmitted in acknowledged mode with transmission if lost. There is one MAC instance. All the data, whether they are video or audio, or files, are transported in one single transport block, which will be managed by the physical layer. To conclude, we have a protocol stack in NR, which is similar to what we had in 4G for LTE, a physical layer, MAC layer, and RLC layer, and PDCP layer. This is not a revolutionary design. We introduce SDAP for the user data to manage different levels of quality of service. One of the important aspects of 5G is to enable greater flexibility in the allocation of resources. But this induces greater complexity.
