Altran recently hosted two webinars about the top ten challenges when doing testing for LTE EPC. A lot of good questions were asked by attendees, and we are sharing them here along with the answers from our panel of experts: Srimanta Kumar Purohit, Tauseef Hasan, and R. Ezhirpavai. We hope they are informative for those of you seeking to better understand the nuances of LTE EPC testing.
Q1. You mentioned about the different options for LTE EPC testing. Can you highlight the different phases of testing LTE EPC nodes?
Yes, in the slides we depicted various LTE EPC testing options. To ensure robustness and reliability, any LTE EPC node may go through different phases of testing, like:
- Conformance testing (protocol or interface conformance),
- Functional testing
- Load and Performance testing (the key focus here is scalability and benchmarking the behavior of the node at different load patterns)
- System integration testing (behavior of the node in an end-to-end test environment)
- Interoperability testing
- Field testing.
Q2. Why is the use of simulated nodes to be encouraged in LTE EPC node testing?
It is preferable to use simulated nodes and interfaces at least in the earlier phases of testing to achieve wider test coverage, thereby ensuring the robustness of the device under test (DUT).
The following examples are easy to test using simulators:
- Test the behavior of a GTPcV2 supported node, when its peer node supports only GTPcV1.
- Test a DUT that receives a trigger to delete a bearer for which the context is not found.
- Test a peer node that sends a request with a mandatory information element missing.
Simulators give better control to trigger different invalid and inopportune scenarios which the DUT may encounter after deployment.
Q3. Which are the different 3rd party test tools available in the market for LTE EPC testing?
Apart from Altran’s EPC test tools, some of the other test and measurement products available in the LTE space are:
- Signallion SORBAS as an LTE Test UE, Aeroflex TM500/EAST 500 for UE simulation, Anite, Anritsu (MD8435A), and R&S (TS 8980) are the other major players in providing test UEs.
- Aeroflex, Anritsu, and Spirent have a product line for eNodeB simulation.
- IxCatapult M500, NetHawk EAST EPC, Polaris LTE Functional Test Tool, Spirent Landslide, and Spirent Test Centre are few of the tools used for EPC node testing.
The selection of the tool depends on different aspects like the node to be simulated, testing phase to be simulated for, the scope of testing, cost of the tool, etc.
Q4. Why do you see proprietary algorithms being used by vendors for ciphering and integrity and does this create interoperability issues? Can one have proprietary algorithms in the consumer network?
Operators require two levels of support for ciphering and integrity protection. One is for the user keys and the second is related to proprietary algorithms so that hacking becomes difficult. The UE and HSS may have proprietary algorithms without affecting authentication, but ciphering and integrity can have interoperability issues as the visiting network’s MME needs to support the same algorithm as the home network’s MME.
Q5. Do you see a challenge in simulating different user speeds for handover in EPC testing within a lab environment?
Yes, there is a challenge in simulating high data throughput which the system under test has to support from a set of simulators. The main challenge in a simulated lab environment is to initiate high and varying data throughputs during handover across different eNodeB simulators, which have to be synchronized with the handover procedures.
Q6. You mentioned about the remote testing of LTE EPC nodes? Can you elaborate on this?
In remote testing, the system under test is remotely controlled and tested. This is possible for IP-based interfaces only. The setup requires a dedicated tunnel/VPN connection with sufficient bandwidth and controlled round trip delay. Based on our experience, we have seen that functional testing can be carried out remotely. However, load and performance are difficult to test remotely due to limitations of the tunnel throughput and latency issues.
Q7. How are Altran’s testing capabilities different from those of the LTE equipment provider?
Altran has LTE EPC offerings, in-house simulators, and test tools to facilitate LTE EPC testing. These can be used in various combinations to achieve different test objectives. Altran has also rich expertise and experience on many 3rd party test tools available in the LTE space.
Altran’s LTE EPC offering includes MME, SGW, PGW, HSS, and PCRF. These can be used for functional and end-to-end positive scenario testing. Altran’s in-house test simulators (specifically used for functional, load, and performance testing of our LTE EPC offerings) include combined UE and eNodeB, PCRF, and charging units.
Altran also has a proprietary software-based tool, called STATE, which is a test enabler for LTE EPC testing. Specifically, it is useful for conformance and functional testing. It is a proven test framework with sample test scenarios available for testing different LTE EPC nodes.
We provide comprehensive services across the testing lifecycle from test consultation to any phase of test execution (including load, performance, and interoperability testing).
Q8. What is the difference in the testing scope for an LTE EPC node getting tested as an individual node and as a combinational node?
Testing for an individual node focuses on that single node under test and verifies the external interfaces for all successful and failure scenarios. Since different procedures are handled by different nodes, the test scope for testing an individual node revolves only around the procedures of the node under test.
On the other hand, combined testing not only involves the external interfaces exposed by the nodes, but also procedures such as handover from local to external nodes, optimization when interfacing with internal nodes, and load balancing between different internal and external nodes. Apart from these, any external protocol extensions that need to be supported by internal nodes also have to be tested.
Q9. How is combined node or individual node test cycle different from end-to-end testing?
End-to-end testing involves system integration in the first phase. All nodes can be from the same or different vendors. End-to-end testing also involves the simulation of external interfaces that are exposed for handovers to other EPC nodes. The complexity of combinational testing increases in end-to-end testing, which involves executing synchronized test cases for multiple external interfaces with multiple triggers. Specifically, load testing in end-to-end setups becomes even more challenging due to the synchronization of interfaces.
Q10. How do you see DPI testing as a challenge?
Deep packet inspection (DPI) testing involves testing different kinds of packets. The level of DPI offered by each vendor is different, and hence a common test tool or test data packets for all kinds of DPI testing is very difficult to find. DPI testing inspects different kinds of bearer creation, synchronized with signaling flows. This involves the proper integration of the DPI engine with the policy engine. Different kinds of policy controls and charging also come under focus in DPI testing.
Q11. LTE has conformance specifications for the UE. What do you do about conformance for the other elements in the network: eNB and MME/SGW/PGW?
The conformance of all nodes has to be tested against specifications. However, except for a few interop cases defined by the MSF for LTE, the conformance suites for EPC are still under development by the MSF. We have studied the interface definitions and procedural requirements for the conformance of all nodes and developed in-house conformance test suites that are most apt for current scenarios.
Q12. Can you please point us to conformance specs in the standard as part of your links? This way I can follow your slides as well.
The conformance specs are still under development by the MSF. One can follow the MSF for the latest updates and developments:
Q13. You listed 10 challenges, are these in priority? If not, how would you prioritize them?
No, the challenges listed in this webinar are not in any order of priority. Strictly speaking, the priority depends mostly on the kind of phase, features, and market segment targeted by the system under test. Having said that, in our view, this could be one common priority order:
- Proprietary ciphering and integrity procedures
- Carrier grade and scalability aspects.
- Policy, charging, inter-RAT, non-3GPP access
- Optimal coverage, interoperability, and voice support
More Answers About LTE EPC Testing
Following up on our first webinar for the Americas on “Addressing the Top 10 Challenges of LTE EPC Testing”, we ran a second webinar for Europe. This again generated more questions from the participants all of which could not be answered during the webinar itself. So, we’ve collected and answered all the questions below.
Q.1. What are the KPIs for testing the performance and quality of LTE networks?
Validating the performance and quality of an LTE network requires a high rate and varied mix of data traffic to judge the end user experience. Traffic mix can be based on type of subscribers, applications/services used, usage etc. Some of the key KPIs which need to be validated are:
- Latency and packet loss
- Perceptual video and voice quality
- Number of active sessions, users, and transactions per second
Q.2. What are the challenges in testing IPv6 in an LTE network?
Although IPv6 is mandated in LTE, as of now, most of the existing networks and their components use IPv4. As a result, we have nodes with support for both IPv4 and IPv6 stacks in the network. The integration testing of IPv4/IPv6 poses a big challenge because all network services will need to be tested for interoperability without any issues between IPv4 and IPv6. Added to this, some tunneling schemes from IPv6 to IPv4 are another area of focus.
Q.3. What are the key attributes to focus on during testing, to minimize the issues with LTE EPC nodes after deployment?
Few of the key attributes to focus on for better solution reliability are:
- The LTE EPC node should be validated with an optimized set of invalid scenarios. Prior experience in interop and field testing is a definite help for identifying such scenarios.
- While testing for the scalability and benchmarking, next to real time scenarios should be simulated, such as, different mix of traffic, simultaneous calls, new calls per second etc).
- Inter-RAT handover or inter-working with non-3GPP must be tested based on the existing coverage in the deployment area
- Redundancy and resiliency should be given more focus during testing from a deployment perspective.
Q.4. There are diameter based interfaces in earlier technologies. How is testing for those different from LTE EPC testing?
There are around 12-14 diameter based interfaces in LTE. There are many new interfaces like SWx, STa, and SW that have been introduced to support non-3GPP interworking. At the same time, there are certain interfaces like Gx, Rx, and Sh which also exist in IMS for interfacing with the PCRF. But in the LTE context, there is a multitude of additional AVPs that require the test tools to be upgraded and complicate testing. The test scenarios are quite different in terms of message flow, handling, and parameter validation.
Q.5. Do you recommend that conformance testing as per the specs by GCF/MSF is enough for an LTE node?
No, it is not enough. Conforming as per these specifications is necessary but not sufficient. It is advisable to conform to these specs as they are standardized and different stakeholders expect nodes to comply with these. But at the same time, based on our experience, the node should also be tested against many field-like complex scenarios, optimized negative scenarios, and interop testing scenarios.
Q.6. What are the challenges in identifying the right traffic mix for load testing?
The traffic mix for LTE testing needs to consider different signaling events as well as data formats.
There are multiple interfaces that can initiate signaling messages with the EPC. Testing for these requires synchronization of triggering the messages from these interfaces. There are different frequencies for each kind of message and the number of such triggers for an attached UE is also different for different events. The messages also depend on the number of subscribers normally active at any moment and signaling events differ for different numbers and kinds of subscribers.
The area where the EPC is going to serve also results in different kinds of traffic. In an urban area, there would be a high rate of handovers, but in rural areas, the rate of handovers may be less. In an enterprise deployment, the number of active users may be more than in a consumer market, and so on.
The data formats also differ a lot based on the usage of different applications. Video download may result in large packets and high data throughput, but email usage may result in larger active duration, but smaller packet sizes.
Identifying the different use cases and traffic mixes for different deployment thus become a challenge.
Q.7. Can you give some examples of applications that would use dedicated bearers? Do these applications typically use network-initiated dedicated bearers or user-initiated?
Some examples of using dedicated bearers can be found in voice and video traffic. Applications like IPTV and video on demand, which require guaranteed bandwidth, may also require dedicated bearers. The dedicated bearers can be user-initiated or network-initiated. Mostly for voice and video applications the network would initiate dedicated bearers based on the SDP exchanged for the media for IMS calls. For applications like IPTV and video on demand, it could be either UE or network-initiated.
Q.8. Can you share some of the important metrics collected for a high available EPC solution in carrier-grade networks?
The important metrics for carrier grade EPC solutions validate the level of reliability and switchover time. Switchover could happen for both the control and the data plane. These important metrics are:
- Switchover time for control plane
- Switchover time for data plane
- Amount of memory for buffered packets during switchover
- Transmission time for buffered packets
- Level of reliability
- Check all levels of redundancy for all resources, interfaces, and software solutions
- Signalling and data throughput after switchover
Q.9. Is there a way to monitor the UE messages from a simulated eNodeB?
Yes. Although the UE messages are encrypted, with a simulated eNodeB it is possible to collect the UE keys from the eNodeB simulator and UE exchanges. Alternatively one can use pre-shared information. The eNodeB simulator can also decipher the NAS messages exchanged between the UE and the EPC.
Q.10. How is the voice quality measured in EPC testing and what aspects are covered as part of voice testing through EPC?
Voice quality is measured using multiple voice test tools to test the following:
- Voice quality
- Echo cancellation
- Comfort noise generation
- Voice break at handovers
Q.11. What are the essential scenarios which must be tested for an MME?
Following are the necessary scenarios which need to be tested for an MME:
- Default bearer creation/deletion/modification
- Dedicated bearer creation/deletion/modification
- Network initiated triggers (default/dedicated bearers)
- Handover (S1, X2)
- TA updates
- Release/Idle mode
Q.12. What is major challenge in testing IOT for voice scenarios like CS fall back?
For voice testing, all the different scenarios like CS fallback, IMS calls, SRVCC, etc. have to be tested. The different voice application nodes like the IMS network, CS network of 3G network, SRVCC handling IMS application server, and SCP nodes have to be simulated such that they can receive and initiate triggers.
The nodes need to be tested for different user capacities and different kinds of voice applications, like emergency calls, voice and video calls, multiband audio; and require multimedia application support in the simulators. This poses a major challenge.
Q.13. How it is ensured that voice and data go on simultaneously; is it possible on CS fall back?
Voice and data can go on simultaneously in CS fallback, depending on the network and UE capability. If the PS session is going on in LTE and voice is transferred to 3G, then the UE needs to support two kinds of radio at the same time. This may or may not be possible depending on the UE’s capability. Even in the case where both PS and CS are handed over to 3G, depending on the UE’s and the network’s capability, either the PS session may be suspended when the CS session is ongoing or both PS and CS sessions may go on simultaneously.
Q.14. What are the best commercially available test tools for EPC core interoperability, inter-RAT, or non-3GPP testing?
Normally for interop, customers prefer to test with a real node from another vendor, rather than using commercially available test tools.
Altran has frameworks for different LTE EPC core elements (i.e. MME, SGW, PGW, HSS, and PCRF) and eNodeB. We have used these for interop for some of our customers.
Additionally, IxCatapult M500, NetHawk EAST EPC, Polaris LTE Functional Test Tool, Spirent Landslide, and Spirent Test Centre are some of the other tools used for EPC node testing. The selection of the tool depends on different aspects like the node to be simulated, testing phase to be simulated for, the scope of testing, cost of the tool, etc.
Q.15. Are there any end-end network monitoring solutions you would recommend? Which vendors?
From the monitoring or protocol analyzer perspective many tools are there which focus mainly on the access network side. For example Anritsu, R&S, Tektronix, and Aeroflex.
From an end-to-end perspective, the few vendors that can be considered for network monitoring solutions are JDSU and Tektronix. Often, Wireshark or other tools used for testing can fulfill these requirements.