LTE Retainability KPI Optimization and Counter Mapping

LTE Call Drops KPI Explanation.

In an all-IP always-on E-UTRAN (LTE) environment, a “call” is defined as a single radio connection between a UE (User Equipment) and the network. When this connection is unexpectedly interrupted, such as by signal loss on the radio interface, it is considered a “call drop.”

Call drop events are signaled by the S1AP UE context release request message sent from the eNB (eNodeB) to the MME (Mobility Management Entity). By the time this message is sent, the radio connection with the UE has already been terminated at the RRC (Radio Resource Control) layer, and both UE and eNB revert to the E-UTRA RRC IDLE state. However, the PDN (Public Data Network) connection between the UE and the IP network application server usually remains active.

To calculate the total number of active calls, consider:

• Initial contexts successfully established.
• Incoming handovers (identified by the S1AP handover notify message).
• Minus the number of outgoing handovers (UE Context Release due to “successful handover”).

The KPI formula for a call drop ratio on the cell aggregation level can be defined as follows:

$\frac{\sum UE Context Release Request (abnormal cause)}{\left[\sum S1Ap Initial Context Setup +\sum S1Ap Handover Notify – \sum UE Context Release [cause ‘‘successful handover‘‘]\right]}\times 100$

In the all-IP E-UTRAN and EPC (Evolved Packet Core) network, not all call drops are perceived as dropped connections by subscribers. For non-real-time services like web browsing or email, users may experience temporary data transport interruptions, delays in accessing websites, or reduced data transmission rates. If the network re-establishes the lost radio connection quickly enough, users might not notice the drop.

However, for real-time services such as VoIP calls, users immediately notice the loss of connection as their ongoing conversation is abruptly cut off. To maintain user experience, the network must re-establish the radio connection within 1-2 seconds.

To accurately measure user-perceived QoE (Quality of Experience), it is essential to compute call drop ratios not only per cell but also per service type (or per QCI – QoS Class Identifier).

The root causes of call drops are diverse and cannot be conclusively determined by the cause value in the S1AP UE context release request message alone. Identifying the root cause requires a detailed analysis of all involved transport and signaling protocols, focusing on call state transitions and radio quality parameter changes. Figure below illustrates various root causes that can be inferred from common call drop S1AP causes like “failure in the radio interface procedure” and “radio connection with UE lost.”

RRC Re-Establishment.

As we know, it is mandatory for the UE to attempt RRC re-establishment before considering a call dropped and returning to IDLE state. RRC re-establishment can help recover connection problems in areas with rapidly changing radio conditions. By counting the number of RRC re-establishment messages and computing a meaningful KPI for this procedure per cell, we can identify such problem areas in the network.

The messages shown in Figure above provide a basis for defining the RRC re-establishment success ratio. This KPI helps in evaluating how effectively the network can recover from connection issues and maintain the radio link. Here’s how the RRC re-establishment success ratio is defined:

$RRC re–establishment success ratio=\frac{\sum RRC\_Connection\_Re–Establishment\_Complete}{\sum RRC\_Connection\_Re–Establishment\_Request}\times 100$

Call Drop Rate KPIs – Retainability LTE KPIs.

VoIP Call Drop Rate.

This KPI evaluates the call drop rate for VoIP services (QCI 1) in a cell or cluster. A call drop occurs when the eNodeB initiates the release of an active E-RAB (with data in the buffer) via an E-RAB Release Indication or a UE Context Release Request message with an abnormal cause. Abnormal causes exclude: Normal Release, Detach, User Inactivity, CS fallback triggered, UE Not Available For PS Service, or Inter-RAT redirection. All other E-RAB releases are considered normal.

$VoIP\_CDR=\frac{VoIP\_ERAB\_AbnormalRelease}{VoIP\_ERAB\_Release}\times 100\%$

Call Drop Rate (ALL)

This KPI evaluates the call drop rate for all services, including VoIP, in a cell or cluster. It considers all abnormal E-RAB releases across different services.

$CDR \left(ALL\right)=\frac{ERAB\_AbnormalRelease}{ERAB\_Release}\times 100\%$

Call Drop Rate (Always Online)

This KPI evaluates the call drop rate for all services, including VoIP, when the “always online” state is active. In this state, the number of E-RABs transitioning from synchronous to unsynchronous states is considered.

$AlwaysOnline\_CDR=\frac{\mathrm{ERABAbnormalReleaseOfAlwaysOnline}}{\mathrm{ERAB\; Re\; leaseOfAlwaysOnline}}\times 100\%$

Counter Mapping for Call Drop Rate.

Call Setup Complete Rate – Retainability KPIs.

VoIP Call Setup Complete Rate (VoIP CSCR).

This KPI evaluates the completion rate for VoIP call setups (QCI 1 service) in a cell or cluster. It measures the rate at which VoIP services are successfully terminated after an attempt to set up the service. The phases considered are:

1. RRC setup phase
2. S1 Signaling connection establishment phase
3. VoIP E-RAB setup phase
4. VoIP service E-RAB release phase

$VoIP\_CSTR=\frac{RRC\_Connection Succes{s}_{Service}}{RRC\_Connection Attempt{s}_{Service}}\times \frac{S1SIGConnectionEstablishment Success}{S1SIGConnectionEstablishment Attempts}\times \frac{VoIP\_E–RAB\_Setup Success}{VoIP\_E–RAB\_Setup Attempts}\times \frac{VoIP\_E–RAB\_Normal Release}{VoIP\_E–RAB Releases}\times 100\%$

Call Setup Complete Rate (ALL).

This KPI evaluates the completion rate for all call setups across different services in a cell or cluster. The phases considered are:

1. RRC setup phase.
2. S1 Signaling connection establishment phase.
3. Service E-RAB setup phase.
4. Service E-RAB release phase.

$CSTR\left(ALL\right)=\frac{RRC\_Connection Succes{s}_{Service}}{RRC\_Connection Attempt{s}_{Service}}\times \frac{S1SIGConnectionEstablishment Success}{S1SIGConnectionEstablishment Attempts}\times \frac{E–RAB\_Setup Success}{E–RAB\_Setup Attempts}\times \frac{VoIP\_E–RAB\_Normal Release}{VoIP\_E–RAB Releases}\times 100\%$

Call Setup Complete Rate (Always Online).

This KPI evaluates the completion rate for all call setups in the “always online” state. The phases considered are:

1. RRC setup phase
2. S1 Signaling connection establishment phase
3. Service E-RAB setup phase
4. Service E-RAB release phase.

$CSTR\left(AlwaysOnline\right)=\frac{RRC\_Connection Succes{s}_{Service}}{RRC\_Connection Attempt{s}_{Service}}\times \frac{S1SIGConnectionEstablishment Success}{S1SIGConnectionEstablishment Attempts}\times \frac{ERAB\_Setup Success}{ERAB\_Setup Attempts}\times \frac{\mathrm{ERABAbnormalReleaseOfAlway\; sOnline}}{ERABReleaseOfAlwaysOnline}\times 100\%$

Counter Mapping for CSCR.