As mobile markets become saturated and the telecommunications world gets ever more competitive, retaining customers and reducing churn become both harder tasks to achieve and more important that they are achieved. Paul Rasmussen investigates how test and measurement can help. 

Building a cellular network is not only fraught with problems and fabulously expensive, but ensuring that it performs and keeps pace with customer expectations is now critically important as competition between mobile operators becomes intense.

While tariffing has traditionally been a key weapon for many operators to gain market share, as margins come under increasing pressure those responsible for devising clever marketing plans need to look elsewhere for innovative ideas.

One concept that continues to gain credence is the notion that quality of service (QoS) is becoming a recognised differentiator as to why subscribers switch or remain with a mobile operator. If Jan Lindeberg, the product marketing manager of Anritsu’s service assurance division, is to be believed, consumers are becoming more knowledgeable with regard to telecoms services - and their failings.

“The customer base is more mature and increasingly demanding with regard to all communications services, and there is a growing expectation that fixed and wireless services will, or should be, equivalent in QoS terms,” states Lindeberg.

Anritsu, a major vendor of wireless test equipment, claims that mobile operators need to find and establish differentiators in the marketplace as voice revenues continue to decline. The company maintains that the only way to sustain revenues is for operators to establish a perceived level of QoS that will position them above other wireless service providers.

“Competing on price with WiFi and fixed line internet providers will not work for mobile operators wanting to offer mobile broadband services, they must look towards QoS and overall performance - this is perhaps the only way they’ll survive,” says Lindeberg.

However, defining meaningful QoS parameters that are recognisable to the user is hard to establish.

Agreeing KPIs
The accepted route to classify the KPIs (Key Performance Indicators) of a network are for the test and measurement (T&M) firms to help the mobile operator agree a set of QoS parameters for a network service or application. These are then translated into KPIs which, in theory, should describe the performance of the service.

“This is much easier said than done,” claims Othmar Kyas, director of strategic marketing for network diagnostics at Tektronix – a long-establish developer of T&M equipment. “Agreeing meaningful and relevant QoS parameters that are truly representative for the user experience is difficult.”

“But, when we have these KPIs described we can then start to implement an optimisation, management and monitoring strategy - we then have a fixed target to aim at, albeit in an environment that is very dynamic.”

The struggle to pin down realistic KPIs is being made increasingly more troublesome by the burgeoning growth of non-SMS data traffic. Kyas maintains that this will impact how KPIs are structured due to the complexities of measuring how data applications are transported over the cellular network, and not helped by the transition to high-speed services based upon HSPA technologies.

“Voice mixed with HSPA traffic is a very dynamic and difficult environment in which to conduct repeatable measurements to establish a workable recipe of how to test and optimise an HSPA cell,” maintains Kyas. “The testing issues in HSPA networks are very different to those within UMTS R99, which is what we have today.

This is due to some architectural differences, such as the high-speed capabilities available within each cell, and could mean 10 or 20 HSPA users per cell operating at 14.4Mbit/s, together with the cell also needing to support voice and conventional data. And it’s this ‘moving target’ that is the reason that test and optimisation issues are different with HSPA networks.”

While Kyas believes that existing HSPA upgrades are performing satisfactorily, their utilisation remains very low and the true test has yet to come. “In those areas where we have seen HSPA traffic increasing, a number of optimisation issues and handover glitches are being observed. For example, calls going from HSPA to UMTS R99 have resulted in a higher number of dropped calls, whereas in more heterogeneous HSPA environments the technology seems to be working well.”

What worries Tektronix more are the forecasts from some industry analysts that HSPA and LTE data traffic could increase 100-fold over the coming years. “This will make it significantly more difficult to reduce the huge quantity of data to the few bits that provide actionable information to manage, configure and optimise the network, or to accurately determine a root cause of a problem area.”

Testing data
If these forecasts for huge increases in data traffic are only partly correct, operators will need to be confident that the user experience is more than satisfactory. Measuring how a mobile data application is performing is already underway, according to Colin Davies, from Rohde & Schwarz’s network operator division.

“We believe it’s possible to measure the QoS of a data application by tasking the handset to carry out a sequence of events such as download a file, access a certain web page or send an email, etc,. Whilst this is underway we can closely monitor the performance of the network.”

“When we’re trying to quantify the time taken to download a page there are certain ETSI-defined trigger points within the signalling process to help establish the performance. So, the download time can be accurately measured by start and end triggers. These ETSI-defined triggers can be used to measure if there are differences between other networks, cells, handsets, etc.”

While the handset vendors will conduct their own compliance testing prior to releasing their products into the marketplace, operators will typically want to perform additional testing to ensure that any handset they qualify for use with their networks works correctly with the specific services they might offer.

These tests will use scripts developed by the operator to check for service interoperability and that the handset operates in a manner that is acceptable. This will also include testing the roaming capabilities to ensure that the handset selects the desired partner operator when the user travels away from their home network.

Handsets themselves have become more sophisticated in their ability to help operators understand what might be happening within their network – without the need to conduct ‘drive testing’ or the more traditional method of taxi cab-based monitors.

“There are software-based solutions within the handsets that are overtaking these more traditional network monitoring methods,” says Lindeberg. “Some vendors are providing software agents integrated into the handset to capture basic registration and other transmission metrics which are then sent to a central data monitoring system.”

This handset-based monitoring can be augmented by additional data starting from the air interface through to the operator’s core network.

“We achieve this by deploying monitoring probes at strategic points within the network. From the base station, to backhaul, through the internal network and up to the demarcation points between the operator and its surrounding partners,” claims Lindeberg. “You are then able to measure how users are accessing the network, how the production systems are working within the network and the efficiency of the content or services partners - and this includes interconnect and backhaul providers.”

Value chain
Understanding what is happening within this network value chain could be fundamental to the success of data applications. Lindeberg maintains that managing this chain from a QoS and performance aspect is one of the biggest challenges for the operators today. “Of course the radio coverage is important and high-speed data adds complexity to this, but the whole value chain is a bigger problem. But, this cannot be ignored given that the user will measure the operator on the service they receive on the handset.”

Of perhaps equal importance is that the information gathered by the operator during this monitoring can be analysed from a marketing angle.

If the operator is able to use data warehousing techniques to gain a comprehensive understanding of how their network is performing - who is using what service and when, they should be able to model their service portfolio so as to remove those that are unprofitable while boosting the promotion of those that are gaining acceptance and generating incremental revenues.

This analysis can be applied to any service level agreements (SLAs) and KPIs that existing enterprise customers wish to see, but more importantly the data can be used to retain valued subscribers or open negotiations with large enterprises that will only sign up with an operator if they can commit to SLAs – and have demonstrable evidence that they can, and are, being achieved elsewhere.

This increase in importance of KPIs and SLAs has provoked an increasing number of the more established operators to examine the functions of their network operation centres (NOCs). “These NOCs will continue to be of high importance in the running of the operator’s network, but many are now adding an extra unit to measure QoS. These service operating centres are more focused on working closely with the operator’s product and marketing managers to monitor what is being successful,” says Lindeberg. “There was a gap between what the NOC was able to see and communicate within the company and what the user experienced on the handset.”

As operators are pulled towards focusing more on monitoring and improving the customer experience, their ability to keep pace with the technological challenges looks increasingly more difficult. To relieve them of some of this burden the larger infrastructure providers are actively seeking outsourcing contracts that will see significant elements of the network management and organisation move under this external administration and control.

Regardless of who is actually responsible for running the network, the operator community is, according to Anritsu, wanting to understand in greater detail than ever what their customers are experiencing. “Networking monitoring systems only tell you part of what is happening – but there is no guarantee that the customer is happy with the service,” maintains Lindeberg.

“We’ve seen many times that on the surface everything appears to be working correctly, but dig into the cellular network and you find that the customer experience is poor. For us the issue is to provide operators with a system that clearly illustrates the customer experience.”