Beyond test drives

Operators are being driven to reduce costs while at the same time rolling out complex new services to grow average revenue per user (ARPU) — they are stuck between a rock and a hard place. Jeff Atkins of Actix explains how software that uses key data from the radio access network can lead to significantly improved productivity, resulting in lower costs and an improved ability to roll out and optimise new services.

The once simple job of radio network optimisation, using just drive test equipment, is rapidly becoming obsolete as the challenges of debugging new data services bring new complexity to troubleshooting and optimisation processes. New and more advance equipment is needed to keep pace. Indeed, market analyst firm Frost & Sullivan expects the market for wireless test equipment to increase to $1.91billion by 2009 from $1.33bn last year, as the delivery of wireless services becomes more complex.

Because the radio link is known to be the weakest link in the wireless provider’s network, it is traditionally the first place to look when service problems with voice networks occurred. However, with complex new technology like 3G, subscriber perceived service problems such as: low data throughput can arise from a variety of causes including unplanned server downtime, internet congestion and core network dimensioning, as well as coverage or interference issues in the radio network. However, before engineers can begin to solve the problem, they must first be able to isolate the portion of the network responsible for the problem.
 
To achieve this it is necessary to have an “end-to-end” view of network performance and that is only possible by correlating data from a number of different sources — drive test equipment, infrastructure vendors’ proprietary call trace logs, protocol analyser logs from open radio and core network interfaces and IP “sniffer” logs. That’s why it’s more important than ever to be able to get access to these various sources in a single platform.

To optimise and troubleshoot a 3G network effectively, performance data needs to be collected from a variety of points in the 3G RAN. It is only by utilising data from a combination of sources that a full picture of the performance of the network can be obtained. For example, to find out that an FTP proxy is related to low throughput and to understand how best to correct the problem, TCP IP logs and drive test data need to be correlated on a common platform.

Information collection points

The most common sources can be seen in the image above and include: the air-interface (Uu), RNC-Node B interface (lub), RNC-MSC interface (lu CS), RNC-SGSN interface (lu PS), and Performance Counters and Measurement Programs (OMC).
Each information collection point offers different strengths and weaknesses in areas such as the type of information that can be obtained (e.g. radio link information, circuit call information, or packet data information), the availability of data collection devices like handsets, the granularity of data which effects its ability to be used to solve specific problems, the ease and cost of collection, and the volume of data that can be collected. Wherever that data comes from, once collected, it must be filtered and reduced before it can be used to make decisions on improving network performance. In addition, collecting and analysing various sources of data at the same time allows efficient utilisation of resources.

Subscriber perspective

Using equipment available from a variety of vendors, operators can drive around their network measuring performance from the perspective of the subscriber. The equipment needed to do this typically comprises a special test mobile phone and wideband scanner, connected directly to a laptop, or indirectly through an intermediate hardware device. The scanners are used to passively measure desired and interfering RF signals from base stations faster and with better accuracy than test mobiles and therefore they compliment the measurements available from the phone. In many cases, scanners can detect the underlying RF causes of the performance problems detected by test mobiles. Some vendors also offer drive test equipment that can be operated by remote control, allowing equipment to be placed in technicians’ vehicles or fleet vehicles (such as taxi cabs), for automatic data collection.

In addition, proprietary measurement programs that run on the switch or RNC enable operators to collect performance data for specified mobile phone numbers. The log files are often used to collect uplink performance metrics to complement the downlink performance measured during drive tests. These log files may be synchronised to drive test data or used independently.

Straight from source

Using protocol analysers available from a number of vendors, operators can collect performance data directly from key infrastructure interface points including the Iu CS, Iu PS, and Iub interfaces. Because these interfaces are based on open standards, the development of collection equipment and analysis software can be completed during infrastructure development. It is then available for use during the planning and lab/field trial phases prior to system launch. Protocol analysers come next and collect a wide range of data, from performance data on the packet and circuit interfaces, down to RF data as reported by the User Equipment.

OMC Performance counters are vendor-specific, proprietary statistical counters of key network events at a network element level of resolution (e.g. statistics for a cell). Operators have traditionally relied on performance counters to monitor the high-level performance of their networks, either using collection software provided by the vendor, third party software, or by building in-house systems. 

Once data has been collected from all sources, it must be processed, analysed and archived. The processing of the data can be challenging for a number of reasons. Firstly, operators typically have a number of vendors for different types of drive test and protocol analyser equipment, each with a unique interface format. Operators also often use measurement programs from different technology networks (e.g. GSM and WCDMA) and/or different infrastructure vendors, each with a unique interface format.

Data sets collected at different interface points may need to be synchronised so that they can be merged for troubleshooting across network elements. The sets may also be extremely large (many gigabytes), and key information must be filtered and reduced before it can be used to make decisions. Finally, formats are constantly being updated and the technology of the air-interface is constantly changing (e.g. 3G rolling out on the back of 2.5G technology). This means that many engineers have limited training and experience with newer technologies.

Multiple function support

To enable operators to use the data effectively, their data analysis platforms must support a range of functionalities. These include supporting interfaces to a variety of vendors of drive test equipment, protocol analysers, and measurement programs and providing support for open interfaces, which can typically be used to collect performance data well in advance of proprietary data sources, like test mobile and peg counter data. Other important functionalities include supporting multiple technologies on one platform simultaneously (e.g. GSM/GPRS and WCDMA), reducing data through binning and standard database type querying and filtering capabilities, and synchronising data collected from different network elements and sources to remove timing discrepancies. Providing interfaces into databases for storing collected data statistics and provide web-enabled reporting interfaces for extracting data is also a key function, as well as supporting the latest technologies and vendor formats, and providing a user interface that allows less experienced engineers to become effective quickly.  Finally, analysis platforms need to be able to embed engineering expertise into software to automate the process of analysing large amounts of data.

Data collection

A system can be designed to collect data from all available links from the air-interface through to the switch/SGSN. Data can be collected in discrete log files and processed through a desktop application for manual, on-the-spot analysis. It can be collected from any source and processed and loaded into a database system from which it may be served up through a web browser or other client.

Mobilkom is one of the first European mobile operators to launch its public UMTS network, having selected Actix’ RVS and ANR solutions to be at the heart of its network optimisation strategy. Testing using Actix’ solutions began in mid-December 2002 and the network went live on the 16th April 2003. Other operators that use Actix’ 3G technology include Swisscom and Hutchinson’s 3G networks in both the UK and Italy.

These operators are streamlining their process — doing it faster and better — ANR and RVS are being used to embed processes into software, accelerating the rate at which rollout can be accomplished, reducing the impact of not having many engineers that are skilled with 3G technology by embedding expertise into the platform, and offering better quality of service by performing complex analyses of problems not previously possible.

Ultimately, being able to bring 3G services to market faster and provide an enhanced end user experience will be critical to operators taking advan-tage of the only foreseeable long-term revenue growth opportunity.