If I don’t use my terminal for a few tens of seconds, it will go back to standby mode with respect to the radio connection. If, after several minutes, a server wants to send me a packet, can it do it though my radio connection is no longer active? That is the question we will answer in this video. We saw in the previous videos that a cell phone that is attached to the network, in other words, that has an IP address, is in the EMM Registered state. If it is inactive, it is in the ECM Idle state. When the user wants to do something with his terminal, this triggers the transmission of a packet from the terminal to the network and the terminal makes a UE Triggered Service Request where we have the re-establishment of first the radio connection, second the S1-AP connection and third the S1 bearer between the eNodeB and the S-Gateway. There is another procedure called Network Triggered Service Request. When a packet arrives from the external network, the 4G network has to be able to re-establish the connection. For example, if you are using voice over IP, you are generally using the Session Initiation Protocol, or SIP, and this protocol defines a message for calling a terminal. We can also have applications that regularly generate messages to be sent by a server. Therefore, there is a need for this re-connection procedure where the initiative comes from the external network. At the end of the procedure, whether it’s the EU Triggered Service Request or the Network Triggered Service Request, the cell phone is in state EMM Registered and ECM Connected. Let’s look at the initial state. Let’s consider the following situation, with the tunnels indicated on this diagram. Between the S-Gateway and the P-Gateway, the tunnels are maintained. The tunnel between the S-Gateway and the MME is also maintained. And one end of the S1 tunnel is maintained: I mean the end on the S-Gateway side. The cell phone is in idle state regarding the ECM state. On the other hand, it is still attached to the network. And the temporary identities attached to it is allocated I mean the GUTI that includes the TIMSI. What triggers the mechanism is a server on the external network that sends a data packet. This data packet is received by the P-Gateway. The P-Gateway does have a data tunnel. So, it sends the packet to the Serving Gateway. As for the S-Gateway, even if it knows the TEID, it doesn’t have a tunnel that is actually active. So, it will request that the MME re-establish the connection. The MME, as we will see in week 6, has a vague idea of the location of the terminal. It doesn’t know exactly in which eNodeB the cell phone is. What it does, is send a message called a Paging message to all cells where the terminal might be found. This Paging Message contains the TMSI ( Temporary Mobile Station Identity). In any case, the terminal is continuously listening to the network, listening to what is sent. If ever it notices its TMSI accompanied by a Paging message, it will reactivate. The terminal, wherever it is, will respond to the paging. Consider the analogy with the dog and the master. The dog is running and is a bit far from the master. The master is just calling his dog. Here, the dog is obedient. It comes immediately. Let see the message sequence chart. The IP packet comes from the external network. We have the bearer established between the Serving Gateway and the P-Gateway and we have the TEID, for example, 102, which is still reserved for the S1 tunnel. The IP packet is transmitted by the P-Gateway to the S-Gateway This packet cannot be delivered immediately. So it is buffered. If there are more packets, those packets will be put in a queue: they will be buffered. The S-Gateway asks the MME to reactivate the connection by sending Downlink Data Notification. In other words, that means “I have data queued that I cannot deliver. Take care of finding the terminal for me.” The MME knows the general location of the UE and sends a message called S1-AP Paging, with the TMSI of the UE, to each cell where the subscriber can potentially be located. The eNodeB passes on this message on the radio interface, so there will be as many messages as there are eNodeBs to which the MME has sent the S1-AP Paging message. If we consider the cell where the UE is located. The UE will react to the S1-AP Paging and will carry out the UE Triggered Service Request procedure. That will cause the re-establishment of the radio connection, with the corresponding Radio Bearer and the S1-Bearer between the eNode B and the Serving Gateway. Of course, the bearer between S-Gateway and P-Gateway remains maintained. Some packets have perhaps continued to be sent to the Serving Gateway. And the Serving Gateway has buffered them, as soon as it can, it will transfer these packets on the re-established bearers. Depending on the operator’s strategy, there can be lesser or greater memory capacity in the SGW. It is possible that the operator decides not to store queued packets, thinking, “If there’s a higher protocol like TCP, it will take care of the recovery mechanism.” If the operator wants to offer a high quality of service, it is better to store the packets between the moment the first packet is received and the moment the bearers have been re-established. We can summarize with a few questions and answers. When is the Network-triggered Service Request procedure used? When the UE is in EMM-REGISTERED and ECM-IDLE states and when a packet should be sent to the UE. The main problem is to reach the UE because its exact location is not known. How does the procedure work? First, the MME pages the UE on a set of cells. Then the UE triggers the Service Request procedure.
