Drop Sessions (Part 1 of 2)

There are several reasons why a session may drop in LTE. However, whether the
session is dropped or not depends on the particular vendor implementation. That is, the
drop may be caused by a UE message or by measurements carried out by the eNodeB.
Both the UE and the eNodeB may check if the radio link is in-synch. In this blog, we will
describe the activities that the UE carries out to determine if the radio link is in-synch
and their consequences. Part 2 of this blog, will present the activities that the eNodeB
may carry out to determine if the radio link is in-synch or not.
So. When is the Radio Link in-synch?
The UE is expected to monitor the RS in the downlink. Based on the signal strength of
the Reference Signals (i.e., the RSRP), the UE will determine if it can decode the
PDCCH based on a certain set of parameters that are provided in the specs. Each UE
will have a different RSRP threshold in which it will assume it cannot read the PDCCH.
If the Reference signals have enough strength such that the UE can decode
consistently the PDCCH, then the link is In-Synch.
How do we determine if the Radio Link is out of Synch?
The full procedure for determining if the link has failed due to being out of sync is shown
in the figure below. In the picture, there are three parameters shown:
n310: This parameter indicates the number of 200 ms intervals when the UE is unable
to successfully decode the PDCCH due to low RSRP detected. That is, this parameter
indicates the number of times in which the UE cannot successfully decode 20
consecutive frames in the downlink.
t310: It is a timer, in seconds, used to allow the UE to get back in synchronization with
the eNodeB.
n311: This parameter indicates the number of 100 ms intervals that the UE must
successfully decode the PDCCH to be back in-synch with the eNodeB. That is, this
parameter indicates the number of times in which the UE must successfully decode 10
consecutive frames in the downlink in order for the UE to assume the radio link is insynch.
If the UE detects n310 consecutive out-of-sync indications, it starts the t310 timer. If the
timer expires, the link has failed. If the UE detects n311 consecutive in-sync indications
prior to the t310 timer expiring, then the timer is stopped and the link has not failed.

So what happens after the UE detects that the link failed?

If the UE determines that the Radio Link fails, the UE will try to reconnect with an RRC
Connection Reestablishment Request message. There are a number of cases that
could occur based on vendor implementation.
What if the eNodeB does not support RRC Connection Reestablishment?
The case shown in the figure below is the simplest case where the eNB does not
support RRC Connection reestablishment. In this case, the eNB responds with an RRC
Connection Reestablishment Reject message. Simultaneously, the eNB will realize that
the radio link has failed and request the connection to be release to the MME. It first
requests to drop the UE Context or the connection to the UE. The cause value is set to
Radio Connection with UE Lost. The MME will respond with a UE Context Release
Command. At this point, the eNodeB will respond with the UE Context Release
Complete message to the MME and will release the RRC connection with the UE by
sending an RRC Connection Release to the UE. Depending on the RF conditions, the
UE may or may not receive this message.

What if the eNodeB does support RRC Connection Reestablishment?

If the eNodeB supports RRC connection Reestablishment, and assuming that the
eNodeB finds both the UL and DL in synch when it receives the RRC connection
reestablishment request message, two scenarios may occur: RRC connection
reestablishment success and failure.
In the case of an RRC connection reestablishment success, the following signaling is
carried exchanged.

If the RRC connection gets successfully reestablished, then the session does not get
dropped.
If the RRC connection reestablishment procedure fails in one of its steps, then the
eNodeB will send the UE context release request message to the MME. Note that the
RRC connection reestablishment process may fail in several steps. Below, in the figure,
only one case is shown.

If the RRC connection reestablishment fails, then the session is dropped.

Yes, you are right!! But think in the consequences first!
If you increase the power of the RS, If you increase n310, if you increase t310 or if you
decrease n311 to its minimum value, the then the number of drop calls will decrease.

Drop Sessions (Part 2 of 2)

In the previous blog, we described the activities that the UE carries out to evaluate the
condition of the radio link to determine if it was in-synch or out of synch. Depending on
the vendors implementation, an out of synch indication may result in a drop session.
In this blog, we will concentrate on the activities carried out by the eNodeB when it
detects that the radio link has failed.
The types of failure that the eNodeB may detect (again, these may be vendor specific)
are:
a) DL failure (RLC failures)
b) UL failure (Physical layer failure).

DL Failure at the RLC layer:

The RLC Layer has a failure when data or signaling that is sent over the air is
unsuccessful and the RLC Layer stops trying. When data is sent over the air, but is
received incorrectly, the receiver will send a NACK. Also, the transmitter can send a
request for an acknowledgement of all received packets, by setting the poll bit. The
receiver will then send a list of all received packets. If a sent packet is not received, it is
considered lost. In either case, the transmitter will retransmit. See figure below.

This procedure can repeat, but at some point the transmitter will give up on the packet.
If that happens, the transmitter declares that the radio link has failed and starts the
procedures to communicate that to the other side.
The parameter MaxRetxThreshold determines the number of times a packet is
retransmitted at the RLC layer in the downlink. If this number is reached, the eNodeB
declares a DL RLC failure and kills the context as shown in the picture below.

UL Failure at the Physical layer:

Not all vendor implementation support this type of failure detection. It essentially
consists in measuring the power of the sounding reference signals (SRS) sent by the
UE in the UL. If the power is below a given SINR threshold, a timer gets started. If the
SINR remains under the stated SINR threshold for the entire duration of the timer, then
the eNodeB declares the UL as out of synch and proceeds to kill the context. If the
SINR of the SRS goes above a second specified threshold during the timer duration, the
UL is said to be in-synch and no actions are carried out.

Below, the actions carried out by eNodeB are shown when an UL Physical Layer failure
is detected.

Yes, you are right!!! But think about the consequences again!
Yes, increasing the value of maxretxthreshold may result in a decrease in the number of
drop sessions due to RLC DL failures.
However, to avoid a large number of drops, the best thing to do is to clean the RF
environment in your network.