VoLTE Monitoring Using Skylight

Overview

Securing the quality of the backhaul network is critical when rolling out packet-based VoLTE services. Traditional QoS methods, such as port or queue utilization monitoring, are insufficient to assert end-to-end service delivery for voice over packets.

The Skylight system by Accedian provides the technology to access extensive packet delivery metrics specifically aimed to monitor VoLTE delivery. By continuously monitoring voice, signaling, and data service classes throughout the IP RAN infrastructure, Skylight reports metrics that include one-way delay variation, packet loss, QoS remarking errors, and QoE indicators, such as MOS-values and R-values.

Measuring VoLTE Delivery with Skylight

The Mean Opinion Score (MOS) and R-Factor are network quality measurements that enable operators to determine whether a network link or path fulfills performance requirements. For decades, the ITU-T E-Model has been and remains the industry standard test used to calculate MOS-values and R-values for VoIP systems. The calculation factors in several network parameters, such as packet loss and one-way delay, and the codec setting. The resulting MOS-value is a QoE rating expressed on a 5-point scale (1 for least, 5 for best). The resulting R-value is a QoE rating expressed on a 0-129 scale (0 for least, 129 for best).
Using this model, Skylight measures network quality for VoLTE systems. It gathers network packet metrics from Layer-3 (RFC 5357 UDP/v4 or UDP/v6) TWAMP streams between the Skylight sensor: control and TWAMP responder. Skylight uses these metrics and the relevant codec setting to calculate MOS-values and R-values. Moreover, Skylight uses the metrics to provide SLA-based alarming, generate performance visualization graphs, and enable northbound data export.

Figure 1 shows a typical MOS value chart.

Figure 1: Sample MOS Chart

Figure 2 shows how packet loss affects the R-value depending on the codec setting.

Figure 2: Packet Loss Impact on Narrow-Band Codecs

While MOS-values and R-values are both useful to network operators, the R-value offers a more precise QoE rating. This is particularly evident when looking at TWAMP measurements for wideband codecs, such as AMR-WB. ITU-T recommendations treat all R-values over 100 as MOS 4.5.

Figure 3 shows how R-values vary slightly in the region above 100. Figure 4 shows the MOS chart for the same data. The MOS-value only goes below 4.5 if the R-value drops below 100.

Figure 3: R-Value Chart Variations Caused by Light Packet Loss

Figure 4: MOS-Value Chart Constant at 4.5

LTE Monitoring with Skylight

A single Skylight system can simultaneously monitor thousands of endpoints. Scale is achieved by deploying Skylight sensor control instances that each can handle up to 4,000 classes of service. By combining multiple Sensor control instances with the Skylight orchestrator, up to 52,000 separate network paths or classes of service can be simultaneously monitored in an LTE system.

Further scale can be achieved by combining multiple Skylight orchestrator platforms into Skylight performance analytics for a unified view of hundreds of thousands of network paths.

Figure 5: Parallel Measurement Flows

Figure 5 shows three parallel measurement flows used to gather metrics for three distinct S1 interfaces in the IP RAN. Skylight monitors call control, media and MME signaling classes in the network (toward the RBs) with three separate TWAMP flows.

At the remote site, the built-in reflector in an eNodeB, site switch or router serves as the responder. If there is no built-in reflector, the Accedian Skylight SFP compute sensors or Skylight sensor: module units can be used to add TWAMP responder capabilities.

It is important to note that the probe traffic (TWAMP/UDP) is not actual signaling or media packets, but pure UDP measurement packets that validate the network path carrying these services.

The responder, whether it is built-in to the eNodeB or residing in a plug-in SFP compute sensor, usually does not have the capability to terminate GTP flows. This means that the measurement packets are always transported outside any GTP tunnels (using the same class of service as the GTP-encapsulated packets). As such, the metrics show the network KPIs for the transport of the GTP tunnel but do not factor in GTP termination functions.

Configuring Skylight for VoLTE Monitoring

Configuring Skylight for VoLTE monitoring involves preparatory steps to ensure connectivity, codec configuration, and session configuration. Sample data is provided to show the resulting measurement topology.

Basic Setup

To set up basic connectivity for the Skylight PM system:

1. Verify that the Skylight orchestrator can connect to the Skylight sensor: control(s).

2. Configure the Skylight sensor: control traffic interface and IP route entries to use for TWAMP measurements.

3. Use the Skylight sensor: control CLI command util-tool twamp to verify TWAMP connectivity with the reflector endpoints.
   # util-tool twamp interface e2 10.2.3.4 tos 184 port 6000

Codec Configuration

Skylight orchestrator supports configuring the E-Model packet loss robustness factor (Bpl) and equipment impairment factor (Ie) separately using CLI commands.
mgr-config name UDPBpl value < VALUE >
mgr-config name UDPimpairmentValue value < VALUE > There is a separate set of codec values configurable for IPv6 TWAMP. mgr-config name UDP/IPV6Bpl value < VALUE >
mgr-config name UDP/IPV6impairmentValue value < VALUE >

The following table shows recommended values for the codecs that are currently supported.

Session Configuration

Use the Skylight orchestrator deployment tool to set up TWAMP sessions. Prerequisites include an endpoint CSV file and session definition CSV file.

• Endpoint CSV file—lists all endpoint (eNodeB or SFP compute sensors) names and IP addresses
  Format:
  ENDPOINT_NAME, IPADDRESS
  Example with three reflector endpoints: MTL-31442-ENB-4G,10.1.1.22
  MTL-31423-ENB-4G,10.1.1.42 MTL-31422-ENB-4G,10.1.1.41

• Session definition CSV file—describes the required session naming and TOS/DSCP settings
  Format:
  TYPE,SNAME,SLA,AF,ANAME,IF,SPORT,DPORT,PPS,INT,TOS,TTL,VPRIO,PSIZE
  Example session definitions CSV file outlining three classes of monitoring identified by IPTOS (DSCP):

TWAMP,EPC1-S1UPL-$RXE,,ipv4,E1Probe,e2,4000,6000,10,300,184,255

TWAMP,EPC1-S1UCC-$RXE,,ipv4,E1Probe,e2,4000,6000,10,300,88,255

TWAMP,EPC1-S1M-$RXE,,ipv4,E1Probe,e2,4000,6000,10,300,56,255

The following table provides field descriptions for the session definition CSV file.

Testing the Sample Configuration

Load the sample files onto the Skylight orchestrator and run the deployment-tool command. A total of nine sessions are created (3 x 3) because three endpoints were defined and three sessions are created for each of the reflector endpoints.

Figure 6 shows the resulting measurement topology.

Figure 6: Measurement Topology Result