2.9.2 3G-2G INTER-RAT HANDOVER FOR CS AND PS SERVICES
2.11 CALL ESTABLISH SUCCESS AND FAILURE RATES
Basically it seems to be easy to calculate call establish success and failure rates, but there are some tricky options. It is important to measure the number of successfully established con- nections exactly, because it is an important input to call drop rates as well.
The first question that needs to be discussed is: What is the definition of a call in a UTRAN environment? According to chapter 1 a call is an active connection between the UE and the network used to exchange signalling and data. The exchange of signalling is a prerequisite to establishing a voice or data call. Hence, analysing the set up of signalling connections represented by SRBs is as important as analysing the establishment of voice and data connections represented by radio bearers (RBs) and/or radio access bearers (RABs).
2.11.1 RRC CONNECTION ESTABLISHMENT
Signalling radio bearers are established when an RRC connection between the UE and the SRNC is set up. This RRC signalling connection establishment is requested by the UE and executed/controlled by the SRNC as shown in Figure 2.41.
In this scenario an RRC connection setup success rate can be defined as follows. It does not matter if SRBs are mapped onto common transport channels (RACH/FACH) or dedi- cated channels (DCH).
RRC Connection Setup Success Rateẳ
PRRC Connection Setup Complete
PRRC Connection Request 100%
ð2:21ị If the set up of the RRC connection fails there are three different cases. In case 1 the RNC is – due to load conditions – not able to offer a sufficient quality for the desired service in
Figure 2.41 RRC connection setup procedure
the cell initially selected by the UE. Based on the establishment cause included in the RRC Connection Setup Request the RNC has already received the information about why signalling radio bearers are to be established: is it only to register to the network or does the UE wish to set up a voice or data call? If there are not enough resources available for the desired service the RRC connection establishment is blocked by the RNC that sends the RRC Connection Reject message. Rejecting the desired establishment of an RRC connection is also known as blocking.
The so-called RRC blocking rate is now defined as follows:
RRC Blocking Rateẳ
PRRC Connection Setup Reject
PRRC Connection Request 100% ð2:22ị
This RRC blocking rate is an important feedback for radio network planners, because typically they have designed network topology and availability of network resources in a way that the blocking rate per cell is approximately 1%. If in reality this value is higher or lower measurement results can help to optimise topology and resource planning.
If the set up of signalling radio bearers is blocked, the establishment of an RRC connec- tion must not immediately fail. Depending on the feature availability in RNC software an RRC Connection Reject message may contain redirection information that is used to redi- rect the UE to a neighbour cell. This neighbour cell could work on a different UTRAN frequency or it is also possible to send the UE to GSM cells. The procedure is called RRC redirection. If there is no redirection information included in RRC Connection Reject the UE will perform cell reselection as described in 3GPP 25.304.
The second failure case is that the UE’s RRC Connection Request is not answered at all although it is sent several times to the same RNC. The RRC connection setup procedure on the UE side is guarded by timer T300 and counter N300. The default value of T300 is one second, the default value of N300 is three. This means that if the RNC does not answer, the UE will send three RRC Connection Request messages using a time difference of one second (as shown in Figure 2.41). If all three attempts fail the UE falls back into IDLE mode.
In this case of multiple attempts the messages protocol analysis is confronted with the philosophical question of whether each unanswered RRC Connection Request is to be counted as a failed procedure or if only the final result of the procedure counts. If only the final result counts the overall success rate of the procedure is computed as 100%, if RRC Connection Setup and RRC Connection Setup Complete is monitored after the third RRC Connection Request sent subsequently by the same UE. Since there are different points of view even within operator service groups a final answer cannot be given in this book. Certainly there is no impact on the customer perceived quality of service if the UE needs to send more than one RRC Connection Request – it just takes one second longer to set up a call. On the other hand if this behaviour of network is monitored more often it must be guessed that something is wrong either in RNC software or in the transport network (because it is also possible that although RRC Connection Request has been monitored at a certain point it has not been received by the RNC, e.g. due to problems in ATM routers).
As another failure case it must be taken into account that the UE is not able to respond to the RRC Connection Request sent by the RNC, because it has again lost radio contact with cell. Since this failure happens relatively often a number of NEMs have implemented a
proprietary counter/timer function similar to the one in the UE in their RNC software. In a typical configuration scenario if the UE does not respond with RRC Connection Setup Complete within one second, then RRC Connection Setup will be sent again. If a second try is not successful either the RNC will send RRC Connection Setup a third time and then stop the procedure. From the perspective of performance measurement software the procedure must be detected as failed if after the second repetition of RRC Connection Setup plus waiting a typical maximum response time (not more than one second) there is still no answer from the UE as illustrated in Figure 2.42.
RRC Connection Setup ‘no answer’ failure rate can now be computed as follows:
Pno Answer to RRC Connection Setup
PRRC Connection Setup 100% ð2:23ị
There are no vendor-specific restrictions for this KPI. Timer and counter values described for the ‘no answer’ failure case might be different in that time periods are shorter and the maximum counter value is less. These timers and counters are not described in 3GPP standards.
Usually there are 10 times more ‘no answer’ failures measured in the case of RRC Con- nection Request establishment cause ‘interRAT cell reselection’. A typical network scenario shows approximately a 10% RRC Connection ‘no answer’ rate for this particular establish- ment cause, while the same failure rate for all other establishment causes is typically measured in a range from 0.5 to 1% depending on the RNC manufacturer. Root causes are different Ec/N0 thresholds defined for the inter-RAT cell reselection procedure.
Figure 2.42 UE does not answer RRC Connection Setup
Knowing all failure cases the total RRC connection setup failure rate can be computed as follows:
PRRC BlockedþP
RRC Request not answeredþP
no Answer to RRC Connection Setup
PRRC ConnectionRequest 100%
ð2:24ị The RRC cell update procedure is performed in a similar way and also guided by the timer/
counter as described for the RRC connection setup procedure. For this reason cell update success/failure rates are not explicitly described in this book. However, one important difference still needs to be mentioned: instead of an RRC Connection Complete message, an RRC UTRAN Mobility Information Confirm, RRC Physical Channel Reconfiguration Complete, RRC Transport Channel Reconfiguration Complete or RRC Radio Bearer Reconfiguration Complete message is used – depending on which kind of specific information was embedded in the previous RRC Cell Update Complete message that can be used e.g for the set up of dedicated transport channels/dedicated physical channels for transport channel type switching.