Demystifying Voice Technologies on LTE Networks

Application Note 328

Mobile network operators (MNOs) must offer high-quality and reliable voice services to their long-term evolution (LTE) subscribers. Otherwise, over-the-top (OTT) players will continue to take bigger bites out of the voice and short message service (SMS) revenue. OTT players do this by making free use of the wireless broadband networks that operators deploy and maintain at great expense. Several avenues are available to operators who want to provide voice services on their LTE networks. To better understand the trade-offs that LTE operators face when providing high-quality and reliable voice services, each voice technology should be demystifi ed. This application note explains the technical differences among the voice technologies that can be deployed on LTE networks. In addition, the document explores what these different technologies are, why they are used, and what advantages and challenges are involved in their implementation.

WHAT ARE THE MAJOR VOICE TECHNOLOGIES?There are several voice technologies in the LTE landscape. This application note will specifi cally focus on voice-over-LTE (VoLTE) and voice over Wi-Fi (VoWi-Fi), since these are the cutting-edge technologies operators will be deploying for the foreseeable future. However, since LTE networks are often overlaid on legacy 2G/3G networks, most operators deploying LTE choose to start with circuit-switched fallback (CSFB) as a way to provide voice services.

CSFB: OVERVIEWFor operators deploying LTE, the most logical fi rst step to provide voice services is to “fall back” to their legacy 2G/3G circuit-switched networks. Since the 2G/3G networks have been fi ne-tuned over the years to provide reliable voice services, relying on proven technologies is justifi ed. This approach of falling back to an existing legacy network for voice services is called CSFB. However, it is not a sustainable model because most operators do not want the overhead of maintaining multiple networks. For them, it is more economical to decommission the legacy networks and re-farm the freed spectrum.

CSFB: TECHNOLOGY OVERVIEW With CSFB, the user equipment (UE) registers with the legacy network at the same time it is attached to the LTE network. A new interface between LTE and 2G/3G, called SGs, is used to coordinate between LTE and 2G/3G. This interface is also used to send and receive SMS without the UE has to leave the LTE network. When a subscriber receives an incoming call, the legacy network notifi es the LTE network, which then pages the UE. To accept the call, the UE has to disconnect from the LTE network and connect to the legacy network. Similarly, when an outgoing call is made, the UE has to detach from the LTE network and attach to the 2G/3G network. Once the UE is connected to the 2G/3G network, the traditional 2G/3G procedures for handling voice calls are followed. Any ongoing data sessions in the LTE network may or may not continue in the legacy network, depending on the network capabilities. When CSFB is applied to 3G, it is common to hand over ongoing data sessions from LTE to 3G. If data sessions cannot be moved, they are typically suspended and resumed when the voice call ends and the device returns to LTE.

Figure 1. CSFB in LTE

CSFB: PROS AND CONSCSFB is generally seen as a temporary, stopgap solution, which is acceptable because it enables the delivery of voice services in the short term with minimal network upgrades. However, due to the following reasons, CSFB should not be considered a permanent and long-term solution:

1. Legacy networks must be maintained. From a cost perspective, it is more effective to decommission legacy networks over time and reuse the freed spectrum within the LTE network.

2. User experience, such as longer call setup times and suspension of ongoing data sessions, is an important issue. Because there are many steps involved in the process of moving from LTE to 2G/3G to accept incoming calls or initiate outgoing calls, call setup times can be noticeably and often unacceptably high. Field measurements indicate that these times are sometimes twice as long as times in legacy networks alone.

VOLTE: OVERVIEWVoLTE is another voice technology that was developed to support voice/SMS services natively on LTE networks. Because voice is delivered as data directly on the LTE network, the legacy circuit-switched voice networks no longer need to be maintained. Practically, however, they must be preserved because legacy 2G/3G networks cannot be decommissioned overnight, at least not until LTE provides full coverage of the operator’s market. Thus, subscribers continue to move between LTE and 2G/3G coverage. This means that operators must continue to support traditional voice services in addition to their interworking with VoLTE.

Demystifying Voice Technologies on LTE NetworksNisar Sanadi and Uday Parida, Product Managers, Simulator, EXFO

© 2016 EXFO Inc. All rights reserved.


VOLTE: TECHNOLOGY OVERVIEWBased on GSMA IR.92 specifications, VoLTE has become the standard for providing voice services natively on LTE networks. Since VoLTE uses the IP multimedia subsystem (IMS) to deliver voice services, it relies on session initiation protocol (SIP) for call control and real-time transport protocol (RTP) for voice. Although the service is delivered by IMS, every element, from the UE and eNBs to the elements in the evolved packet core (EPC), plays an important role in delivering quality VoLTE service.

Different types of data have different levels of tolerance for network impairments such as packet loss, delay, and jitter. Since the network is shared by several types of data, it is imperative that the network prioritizes and processes all these data types differently. This is based on their respective tolerances and the importance of the associated service. Voice data, for instance, is particularly sensitive to any network impairments, and VoLTE service is an extremely important service for any operator. Thus, the network should give the highest priority to the voice data associated with VoLTE. Other services, such as fi le transfers or web browsing, are more tolerant and of lesser importance from the operator’s perspective. Data associated with such services can be treated as a lower priority. The policy and charging control (PCC) framework is responsible for this smart processing of data through prioritization, thus ensuring the most effi cient use of limited network resources by ensuring QoE. Hence PCC plays a dominant role in ensuring high-quality VoLTE service.

Single radio voice call continuity (SRVCC) is another very important feature associated with VoLTE. SRVCC is responsible for the seamless handover of voice calls from LTE to 2G/3G networks. When done correctly, this handover is completely transparent to the subscriber who may not even realize that he or she is no longer on a VoLTE call. Of course, to do it right, a lot of complex maneuvers have to be perfectly executed within the network and under stringent timing constraints.

Figure 2: VoLTE Architecture

VOLTE: PROS AND CONSThere are many documented advantages to VoLTE. Here are the most important:

1. High-defi nition (HD) voice quality is the biggest advantage for consumers, who will notice a better quality in the voice due to the use of advanced codecs like AMR-WB.

2. Efficient use of spectrum. Operators can free up spectrum currently used by their legacy networks and use it in their LTE networks.

3. Faster call setup times, which can be twice as fast as legacy networks.

4. Foundation of new services, such as conversational video and rich communication services (RCS) like video chatting, file sharing, etc. VoLTE is the tip of the iceberg of the IMS-based multimedia services operators can offer.

5. Seamless interworking between operators. A global standard will ensure that there are no interoperability issues between worldwide operators. Therefore, a VoLTE subscriber will be able to make voice calls on any VoLTE networks owned by any operators.

6. Advantages over OTT services, which enable operators to compete on quality and mobility, as well as emergency calls.

Although these advantages are clear and manifest, VoLTE remains an extremely demanding service, putting tremendous stress on the network, both in terms of voice data processing at the highest priority level and signifi cantly increased signaling traffi c associated with PCC.

VOWi-Fi: OVERVIEWVoWi-Fi is another voice technology that uses Wi-Fi as the access technology to extend service to areas where coverage and congestion is a problem. The Wi-Fi network’s use of an unlicensed spectrum provides the mobile industry with important fi nancial benefi ts. This extra spectrum—and its widespread availability—has helped operators to substantially reduce the cost per bit. The presence of access points, which are located indoors and, therefore, are closer to subscribers, ensures suitable signal strength.

These benefi ts led operators to consider Wi-Fi for voice services. Thus, VoWi-Fi or Wi-Fi calling was born. Since it enables operators to launch IMS-based voice and video calling, even early in the process of building out LTE coverage, VoWi-Fi can be used to complement VoLTE. VoWi-Fi, therefore, provides users with an enhanced experience compared to CSFB.



VOWi-Fi: TECHNOLOGY OVERVIEWVoLTE and VoWi-Fi have several similarities. Both require a session initiation protocol (SIP) stacked in the client smartphone to access the IMS network. They also share the same architecture. However, VoWi-Fi also has a few differences:

› Evolved packet data gateway (ePDG) and trusted wireless access gateway (TWAG) are new elements included in the operator’s network.

› Smartphones require Internet protocol security (IPSec) support to meet security measures that ensure data are carried securely over Wi-Fi.

› IMS handles the call termination differently (i.e., whether to terminate a call in IMS or to break out to central server).

The ePDG enables mobile operators to provide subscribers with secure access to the third generation partnership project (3GPP) EPC networks from untrusted non-3GPP IP access networks that are widely available in coffee shops, airports, hotels, etc. Indeed, the ePDG works as a security gateway that gives access to network security and control via IPSec tunnel establishment. The information obtained is based on 3GPP authentication, authorization, and accounting (AAA).

While ePDG offers offl oad using untrusted Wi-Fi access, TWAG does that over trusted Wi-Fi access networks. The requirements are similar in both cases, but there is no need to have a security function in the latter case, because the subscribers use trusted access.

Figure 3: VoWiFi Architecture

VOWi-Fi: PROS AND CONSAlong with the benefi ts of offl oading the network, it also provides other major benefi ts. With VoWi-Fi, operators have the ability to provide roaming subscribers with access to their home network via the Wi-Fi access of the visiting network. This means that subscribers can enjoy local rates even while roaming. Such service availability and cost savings are benefi cial for both subscribers and operators. Wi-Fi can cover areas without LTE coverage, thus eliminating the need to have support for legacy networks and complex technologies like CSFB and SRVCC.

While there are many benefits, carrying VoWi-Fi relies less on bandwidth and more on reliable and consistent connections between the device and the access point. Consequently, voice is properly transmitted in favorable Wi-Fi conditions, but issues may arise if there are interferences or overloaded access points. In these cases, techniques that can prioritize real-time media over ordinary data will be helpful and should be privileged in the network.

This concludes the description of the prominent voice technologies associated with LTE. Each is technologically challenging in itself. The challenge in wireless networks today is the fact that multiple voice technologies need to coexist and interwork seamlessly with each other. This is the challenge of Voice over X. Successful deployment of voice services in this environment requires extreme preparation. A signifi cant part of this preparation involves testing these services in real-world conditions. This could be done either in a lab environment or even in a production network prior to going live with a service. The next section covers how EXFO’s QualityAssurer solution meets this challenge of preparing for success with Voice over X.

EXFO QUALITYASSURER There are certain prerequisites a test solution must satisfy in order to rise to the challenge of testing Voice over X:

BREADTH OF COVERAGE Comprehensive testing of Voice over X is only possible if the test tool has extensive support of all the technologies involved, spanning 2G/3G, 4G, WiFi, and IMS. The test solution must support them all in a tightly integrated manner, and come from a common platform with a common front end. The necessary functionality for interworking among these different technologies must be built in to enable the testing of services, such as CSFB, and features, such as SRVCC.

PERFORMANCE AND CAPACITYLive networks typically service millions of subscribers who generate an ever-increasing volume of mobile data, thanks to the constantly growing selection of data-hungry apps. Replicating the same scale and volume of subscribers, network elements, and mobile data is a must for effective testing. Of course, a test tool must be able to do this with a reasonably sized system —the smaller, the better, from a manageability perspective. This implies that the performance/capacity density of the component modules within the test solution remains very high.

TRAFFIC SHAPING A variety of factors determine the traffi c within a live network such as the type of market, the time of day, special events, etc. It also changes due to differences in overall subscriber behavior patterns, popularity of new apps, introduction of new devices, and other reasons. Clearly, effective testing requires these traffi c patterns must replicate accurately. This requires the test solution to provide powerful traffi c-shaping capabilities with suffi cient granularity of control. Users of the test tool should be able to shape the traffi c based on busy hour call attempts (BHCA) for individual procedures, volume of data generated, types of data and their relative mix and other such factors.



FLEXIBILITY Comprehensive testing goes beyond just testing for blue-sky scenarios. A solid test program should include testing for negative scenarios, misbehaving devices or network elements, failure or overload of network elements and so forth. Such types of testing require extreme flexibility from the test tool. Another area where fl exibility is critical is in the handling of variations in the interpretation of the same specifi cation by different vendors. For instance, the SIP stack on an iPhone is likely to differ slightly from a Samsung handset. If not explicitly tested, even minor variations can prove potentially fatal. Flexibility is also necessary to keep up with the rapid changes taking place in the types of mobile data transmitted through the networks. The test tool should have the fl exibility to enable end users to adapt to these situations and not have to depend on the tool vendor.The capabilities listed above enable testing Voice over X in a variety of test scenarios and use cases. EXFO’s QualityAssurer solution delivers all of the above characteristics. Thanks to the extreme adaptability of the QualityAssurer solution, the possibilities are practically endless. The following sections examines interesting test scenarios and use cases.

TEST SCENARIOSIn this section, some of the most common test scenarios in which Voice over X is tested will be detailed. This list is by no means exhaustive.

ELEMENT TESTING In this case, an individual network element is isolated for testing. All other elements that this element communicates with are simulated. Practically any element can be isolated for testing. As an example, here is the evolved node B (eNB) testing.

Figure 4. eNB Testing

To validate the eNB’s VoLTE handling capability, the QualityAssurer can simulate the complete EPC and IMS in a single box. Either real VoLTE capable handsets or simulated VoLTE devices can be used to make VoLTE calls.

The following are the benefi ts of using EXFO EPC-SIM to test eNBs:

› Combined EPC and IMS simulation solution makes validation of eNB easy by eliminating the need for external elements like the policy and charging rules function (PCRF) and avoiding trouble-shooting of the Gx and Rx interfaces.

› Flexibility allows for validation of corner conditions and proprietary requirements.

› Support for advanced features like multiple packet data network (PDN), emergency calls, and Video over LTE make it future proof.

To extend the testing to also verify mobility scenarios like SRVCC, the QualityAssurer can simulate the circuit-switched 3G core network and interact with 2G or 3G access networks.

Similarly, any element within the wireless core or IMS, such as the mobility management entity (MME), signalling gateway (SGW), packet data network gateway (PGW), call session control function (CSCF), and session border controller (SBC), can be isolated for element testing.

EPC and IMS TESTINGIn this test scenario, the EPC and IMS together comprise the system being tested. QualityAssurer simulates the eNBs and UEs. Additionally, to test for SRVCC, MSC and serving GPRS support node (SGSN) from 2G/3G networks are simulated. Similarly, to test for VoWi-Fi, ePDG and TWAG elements are simulated. Thus, the EPC and IMS are completely surrounded by simulated elements. Users can then specify traffi c models with complete control over the following:

› The number of simulated elements and subscribers;

› Subscriber behavior that results in desired traffi c patterns;

› User plane traffi c mix, which could comprise Voice over X traffi c along with data related to other services and activities, such as video streaming, social media interactions, web browsing, etc.

This setup is ideal for ensuring that the core network can handle all variants of Voice over X traffi c and that it can deliver quality service for VoLTE and VoWi-Fi at the target load levels. It can also ensure that the core network can handle interactions between the different voice technologies resulting from subscriber mobility between 2G, 3G, 4G, and Wi-Fi.

END TO END TESTINGIn this test scenario, the system being tested is comprised of everything except the Radio Access Network (RAN) portion of the various technologies. The QualityAssurer simulates base station controllers (BSCs) and radio network controllers (RNCs) in 2G/3G, eNBs in 4G, and access points in Wi-Fi.

This setup comes closest to verifying the end user experience under full load conditions across the entire operator network. Again, this test setup helps to ensure the quality of not just individual voice technologies but also their interworking, as simulated subscribers move between 2G/3G, 4G, and Wi-Fi.

USE CASESWithin each of the test scenarios described above, it is possible to target specifi c use cases as part of the testing. Each use case has a very specifi c objective. There are many basic use cases that measure the quality of service under different load levels or the responsiveness of the network. However, due to the fl exibility of the QualityAssurer platform, along with some of its unique features and its capabilities around performance and capacity, some truly unique use cases are made possible. A few of these use cases are described below.

VoLTE/VoWi-Fi BREAKING POINTThe objective with this use case is to identify the VoLTE or VoWi-Fi load level that triggers an unacceptable degradation in the quality of service. This use case could be done in a variety of test setups, such as EPC and IMS testing, PGW isolation testing, etc.

EXFO Headquarters > Tel.: +1 418 683-0211 | Toll-free: +1 800 663-3936 (USA and Canada) | Fax: +1 418 683-2170 | [email protected] | www.EXFO.com

EXFO serves over 2000 customers in more than 100 countries. To find your local office contact details, please go to www.EXFO.com/contact.


APNOTE328.1AN © 2016 EXFO Inc. All rights reserved. Printed in Canada 16/05

As part of the setup for this use case, incremental steps are defi ned. Each step is comprised of a certain number of simulated subscribers and the load they generate. This load can include not just Voice over X traffi c, but also other types of activities, such as video streaming, email access, etc.

As part of the execution of this use case, load is gradually ramped up step-by-step. At each step, the system tracks key performance indicators (KPIs) that indicate the quality of Voice over X service. This could include mean opinion score (MOS) for voice quality, call setup times, jitter, etc. At the same time, the system can measure KPIs to track how other services are performing. For instance, the system could track http response times as a measure of users’ experience with web browsing or some other http-based application.

At some point, as the load is ramped up, the quality of Voice over X service will degrade. This is considered the breaking point, since operators do not ever want voice service quality to drop.

Time VoLTE MOS

HTTP Response Times

VoLTE Load

Figure 5. Nailing the breaking point of your VoX service

SERVICE RESILIENCY TESTINGThis use case focuses on the ability of the network to handle failures in network elements. Inevitably, over time some network element will fail, which is one of the reasons critical elements deploy in pools. Members of a pool share the load under normal conditions. However, should any element within the pool fail, other members within the pool take over for the failed element. This failover should happen fl awlessly and quickly to have minimal to no impact on subscriber QoE. The use case described here aims to quantify the impact of such events on the network and test how resilient it is in recovering from them. How long does the failover take? How many ongoing calls are dropped? How much longer does it take to set up a new VoLTE call? These are the types of questions this use case tries to answer. The diagram below illustrates this for a P-CSCF failure as an example. It could be any other element like the MME, SGW, PGW, and the like.

One of the challenges with this use case is that implementations for handling failovers are often vendor-specifi c. This is because the specifi cations generally provide only guidelines. One can imagine the potential danger of this in a multi-vendor network deployment. The fl exibility of the QualityAssurer solution is the key in the ability to deal with this. In fact, a tier 1 NEM evaluated every option in the market and concluded that the QualityAssurer was the only solution capable of handling these types of scenarios.

EUTRAN

EPC

IMS

PDNeNB

S1-U

S5

S11

eNBeNB

eNBeNB

EUTRANEUTRAN PDN

Server

EUTRANeNB

EUTRAN

MME

eNB S1-U

HSS

S1-U

HSS

SGW

Server

S11

PDN-GW

HSS

EPCMMESGW

PCRFPCRF

ATCF

P-CSCF

S5

EPCS5

SGW

PDN-GW

PCRF

PDN-GW

P-CSCF

SCC-AS

IMSIMSIMS

ATCF

P-CSCF

SCC-AS

IMS-MGW

S1-MME

S6a

Mb

Rx

Gm

SGi

EUTRAN

SYSTEM

UNDER TEST

P-CSCF

Figure 6. EPC and IMS Testing

These are a few sample use cases, but there are many other possibilities. For instance, use cases targeting specifically PCC functionality such as charging accuracy and handling of real-time policy updates have been proved valuable.

CONCLUSIONEach individual voice technology that can be deployed on an LTE network has been described and explained. They all have advantages for adoption and deployment. However, along with many benefi ts come many challenges.

From a voice service perspective, the biggest challenge awaiting LTE operators is having multiple technologies existing in the network at the same time. For instance, an LTE network might have to support both VoLTE and VoWi-Fi. In addition, the operator may also need to have a 2G/3G network to cover areas not yet covered by LTE. Subscribers on voice calls can then roam freely between LTE, Wi-Fi, and 2G/3G coverage.

This is the Voice over X challenge: now that these voice technologies have been introduced individually, they must coexist in the network.

While the story of voice services in wireless networks is decades old, we have now entered a new era of Voice over X. Voice is back, and this time it’s crossing all boundaries.

Demystifying Voice Technologies on LTE Networks

Documents