The Drive for Improved Performance

Mobile communications networks truly have changed the way we live. Many years ago, lots of dimes and a memorized list of public phone locations were critical to a big-city salesman’s success. You had no other way to inform a client that you were tied up in traffic or that your car had broken down. Today, cell phone service is ubiquitous. However, for network operators, meeting ever-higher expectations for both voice and data communications is a continuing challenge.

Network planning must balance a wide range of factors to achieve the required coverage and quality of service (QoS). On the one hand, base stations are expensive so you want to minimize their quantity. On the other hand, unless a company’s cell phone service is at least as reliable and fulfilling as traditional landline phones, customers may become frustrated and try a different network.

So, a network operator’s job has distinct phases. The first is to determine the extent and nature of the network. Then, it must be built and managed. Given the high rate at which new technologies are being developed, it’s easy to argue that there is yet another phase—the redevelopment of the network.

Throughout, one factor distinguishes this type of communications network from the traditional phone system: mobility. In plain old telephone systems (POTS), terrain didn’t matter. Dealing with hills and valleys and existing infrastructure in cities added cost when POTS were being built. But once the networks were completed, these factors didn’t affect service. That’s not the case with cell phone networks.

A mobile communications network operator faces a very dynamic situation. Customer satisfaction can change with the seasons because foliage affects signal strength. Similarly, new road layouts and large construction projects can alter signal propagation patterns that have been stable and acceptable for years. Of course, concentrated business development implies that a greater network capacity both inside and outside buildings may be required to handle higher peak traffic, which also is a problem for POTS.

To deal with all of these considerations, you need to know how well the voice or data network actually is performing. This means measuring performance at many locations and times throughout the network area. And, typically, this is accomplished via drive testing.

Driving Down Dissatisfaction

Planning
Basically, drive testing is a form of mobile data acquisition. The objective is to gather information that can be applied to initial planning, QoS troubleshooting, or simply routine performance monitoring. It’s accomplished, as the name implies, by driving around the area of interest in a vehicle equipped with the necessary instrumentation. But, the type and level of instrumentation you need depend on your test purpose.

In the initial planning stage, it’s important to have accurate knowledge of signal propagation losses. Network planning software such as the Actix Radioplan application can estimate path loss based on the local terrain. This 2G/3G automatic RF optimization software is claimed to quickly improve path-loss accuracy based on drive testing data that is used to fine-tune the planning database.

MapInfo from Pitney Bowes is one of the leading sources of mapping information. Many of the telecom-oriented planning and drive test analysis packages interface to maps from MapInfo. For example, the Rohde & Schwarz ROMES Software has a modular architecture that allows any type of data to be collected and analyzed. The graphical subsystem contains a map section that deals with imported data from MapInfo files and uses the Pitney Bowes MapX utility to display measured data at the appropriate map location.

Signal path-loss measurements can be performed by using a crane as a base station. You can dangle a transmitter or antenna from the crane and determine how well its signal propagates by measuring received signal strength as you drive various routes around the area. Data is tagged with location by a built-in GPS. From that data, you can tell how well the planning software represents the actual signal propagation conditions and if the proposed number and locations of base stations are suitable.

Berkeley Varitronics Systems provides several transmitter models including the dual-channel 45-W Tortoise and single-channel Gator with output power from 10 W to 45 W depending on frequency. These both are Class A transmitters with low spurious and harmonic content. Output power and frequency can be remotely programmed.

In addition, you need a receiver that can accurately measure signals from your mock transmitter as well as interference from other sources. A small, portable spectrum analyzer such as Berkeley’s YellowFin for signals from 2 GHz to 5.9 GHz can do the job for WiMAX networks. These tests must be done with the frequencies your network will use.

Propagation loss drive test data typically is plotted on top of a conventional map or a projection similar to a GoogleEarth view of an area. As Figure 1 shows, signal strength varies along a given route. Several post-acquisition processing applications are available that present drive test data as part of a graphical map; Forecaster from Berkeley is one example.

Figure 1. Drive Test Signal Strength With Base Stations Highlighted Courtesy of Rohde & Schwarz

The locations of your network’s base stations are indicated, and you can elect to show foreign network towers as well. Drive testing validates the proposed network design. It establishes reference signal levels or benchmarks against which the actual network performance will be compared as its construction is completed.

Because of the interactive nature of drive testing early in the planning phase, reviewing the acquired data on-site may be important. There are PC-based drive test systems that support immediate post-acquisition analysis or analysis during acquisition, so this type of instrumentation could be appropriate. Nevertheless, combining data acquisition and analysis implies changes to the number and abilities of the personnel doing the drive test.

The person driving the car needs to concentrate on driving. He can set up data acquisition conditions beforehand but will have little time during the actual drive test to interact with the data-collection process. If he also has sufficient technical abilities, he can review the collected data at frequent stops during the test. Alternatively, a second, technically trained person could operate a PC to analyze the data on the fly during the test and recommend changes in transmitter frequency.

The Distributed Measurement and Testing System from Focus Infocom and distributed by JDSU is an example of this distinction between acquisition and analysis. According to the product brochure, the system is designed for benchmarking applications and one-person, nonexpert drive or walk tests.

During operation, system status is monitored with audiovisual feedback without distracting the driver’s attention. Driver-only operation is called student mode. Full access to real-time data, called the expert mode of operation, also is provided.

Device Installation and Data Collection
Most drive testing equipment manufacturers refer to Lee testing or Lee sampling in product literature. They’re talking about a 1974 technical paper co-authored by William C. Y. Lee and Y. S. Yeh that directly relates to signal strength measurement.¹

The objective in mobile radio signal strength measurement is to determine the effect that terrain-based fading has without including the separate effects of Rayleigh or fast fading. Rayleigh fading is caused by close-in reflections and produces drops in signal strength that are only a fraction of a wavelength in size.

Lee’s paper found that eliminating the effect of Rayleigh fading on signal strength measurements only requires that the data be averaged for a time period equivalent to the time required to drive a distance of 40 wavelengths. There should be at least 36 points measured within that time.

At 800 MHz, for example, the wavelength is about a foot. So, at 60 mph, or 88 ft/s, sampling at 88 S/s would provide one sample/ft or 40 samples in the time it took to travel 40 ft.

Lee’s criterion, known as 40 Lambda Sampling, is spatial, not time-based. As a result, many drive test receivers have triggering capabilities that accept an input from a special sensor that monitors rotation of a car’s wheel.

The triggering is based on distance traveled, not time. This is the meaning of a receiver specification such as 6 in. at up to 125 mph. The receiver can make accurate measurements as fast as once every 6 in. even at a vehicle speed of 125 mph. In this case, the equivalent rate is 367 measurements/s.²

In addition to a wheel sensor, a car also may require special wiring and power distribution depending on the type of drive test equipment used. Some equipment is designed to be plugged into the car’s accessory socket. This is convenient, but typically this socket only can supply up to 10 A.

Larger drive test chassis with several phones and receivers can draw more than 10 A and usually are mounted semipermanently in the car’s trunk. For example, when fully loaded, Agilent Technologies’ E6474A Wireless Network Optimization Platform draws up to 17 A and requires option 050, the receiver power distribution module. This device has fused outputs, includes undervoltage detection and ignition sense line support, and is reverse-power protected.

Quality of Service
By far, the most common use of drive testing is to investigate QoS problems. After a network has become operational, in addition to basic signal strength, all the nuances of its particular protocols affect performance. Regardless of the underlying cause, customer complaints are based on calling experiences: Was the audio clear, was the call dropped, could the conversation be continued without interruption as they drove through the coverage area?

Because customer problems are based on phone performance at certain locations and times within a network, it makes sense to attempt to replicate those situations with actual phones. Using real phones ensures that you are interacting with the network protocols in the same ways as the customer. Most drive test equipment allows simultaneous use of several different types of phones during a single drive test. Typically, scanners also are supported.

Nevertheless, some problems are external to the cell phone network, and a phone-based test system can’t address these. Instead, you need a wideband receiver or scanning receiver. It’s important to identify possible sources of interference including base stations in other networks.

Agilent’s E6474A Platform can be used with up to four phones as well as up to four W1314A Measurement Receivers. The phones assess protocol-related impairments and interact with the network, but the receivers operate outside of the network protocols and determine parameters such as signal strength, frequency, and timing. Receivers can perform spectrum analysis across a frequency band, regardless of network protocols.

They also can demodulate and decode the received signals to determine the base station identity code (BSIC) or digital color code (DCC). One W1314A Receiver can measure up to eight frequency bands.

The GSM version of this receiver measures power at a rate of up to 1,400 channels/s; but when decoding BSIC, the rate drops to 45 channels/s. The integrated dispatch enhanced network (iDEN) version scans at up to 3,000 channels/s, dropping to 70 channels/s with DCC decoding. A large range of options covers Mobile WiMAX, GSM, iDEN, CDMA, and UMTS technologies as well as specific frequency-band selection.

Most recently, software has been released that provides scanning capability for a WiMAX PC card (CPE) when used with the E6474A Platform. The CPE can be used in phone mode or in scanning mode, which measures RF parameters on up to 10 channels.

With the E6474A, a receiver can be assigned to track a phone and create an independent measure of the signals the phone is working with. This means that you can correlate dropped calls with events occurring in the overall RF environment.

In addition, if a problem is reported in call hand-off, you can determine that the list of neighbor base stations is correct. The affected phone might not have a base station in its neighbor list, but the receiver data may show that the station should be.

Comarco’s Marketing and Product Manager William Ortiz explained some of a receiver’s advantages. “In addition to providing a picture of the overall RF activity in an area, receivers provide more accurate information than a phone in the presence of strong interferers. A strong signal can saturate the front end of a receiver, and in a drive test system that uses both phones and receivers, the phones can interfere with the receiver.

“A receiver designed to have a high third-order intercept point (IP3) level minimizes nonlinear mixing caused by strong signals,” he continued. “With a high 1-dB compression point as well, measurements of strong signals remain linear and accurate.”

The Symphony QVP Software from Comarco and Ascom Qvoice consists of several modules that deal with administration and analysis of network data. The Pitney Bowes MapInfo program provides map details upon which drive test data can be displayed.

In-car hardware ranges from the Symphony Multi that supports up to 24 test phones, data cards, and scanning receiver to the Symphony Opti Lite that consists of a laptop and one data card or up to four phones (Figure 2). The USB bus is used in these and many other manufacturers’ systems to interconnect the hardware elements.

Rohde & Schwarz offers several types of scanning receivers that transfer data to a PC via FireWire. The company calls these products radio network analyzers. The latest model, the TSMW Universal Radio Network Analyzer, features an FPGA-based software-defined architecture that provides two independent RF and signal-processing paths, each with 20-MHz bandwidth.

As Stefan Schindler, product manager for drive test, explained, “All Rohde & Schwarz TSMQ Drive Test Scanners are broadband and can be used in all bands between 80 MHz and 3 GHz independent of the network technology. TSMW operates to 6 GHz. These scanners demodulate all system information blocks for GSM, WCDMA, and CDMA2000 to localize problems with cell ID, location area code, mobile network code, and mobile country code.”

In addition, the TSMW model includes access to I/Q streaming data via a 10/100/1,000Base-T Gigabit Ethernet interface. Together with a flexible MATLAB interface and an equivalent C++ function interface, this instrument supports customer-specific measurements and algorithm development.

X-TEL Communications offers a drive test system with host controller hardware that supports up to six USB devices and eight serial devices. The X-2 GSM and WCDMA scanner cover GSM 900/1, 800-MHz and WCDMA 2,100-MHz bands. Data acquired by these devices can be displayed on street networks from any MapInfo-compatible source. User-configurable graphs, charts, and text displays of GSM and WCDMA parameters are shown.

The company recently introduced Rush Street Mobile, a data analysis and viewing application that allows engineers to plot and view multiple metrics during the drive test. This software provides the capability to make decisions in the field without returning to a central post-processing location.

GL Communications’ Senior Manager Robert Bichefsky made the distinction that his company’s phone-based drive test system is used for voice-quality testing while the receiver-based system performs measurements from the RF signal point of view. “Each system complements the other and corroborates during analysis of drive test results to develop a complete system understanding,” he said. “Testing determines voice quality using perceptual evaluation of speech quality (PESQ), perceptual analysis measurement system (PAMS), and perceptual speech quality measure (PSQM) ITU algorithms. We also measure round-trip propagation delay and plot mean opinion scores along with street maps, GPS coordinates, and voice quality test scores.

ZK Celltest has taken a somewhat different approach to drive testing in the ZK-SAMp, a portable system access monitor. This instrument consists of a small, 2-lb controller that accommodates up to five phones or phone-based scanners. A separate color display with control and navigation buttons supports data replay while still in the field.

The controller has its own internal battery for approximately four hours of operation and also can be powered externally from a 12-V source. This instrument is not PC-based and has been designed to operate from -20°C to +60°C.

The data is logged to a compact flash card and compatible with map source files such as MapInfo. In the 23 years that ZK Celltest has been providing drive test equipment, the company has become convinced that non-PC based products are easy to operate and provide inherent safety, reliability, and productivity advantages over PC-based products. The ZK-SAMp can be used to map network performance inside a building as easily as outside, and the instrument has a big brother, the ZK-MPS, with a Comarco Seven.Five Scanner.

Summary

Drive testing is a well-established way to measure actual mobile network performance. On the other hand, it’s expensive and inconvenient to have one or more cars and drivers tied up on the road. Maybe there’s a better way to achieve the same results.

Comarco’s Mr. Ortiz described how his company uses the increased processing power in today’s smart phones to accomplish at least a large part of the data acquisition previously addressed by drive testing. “A QuOTA client is downloaded to a Windows Mobile- or Symbian-based phone and runs there as an application, adding the capability of a remote measurement probe to those the phone already has. The QuOTA client is remotely controlled by a centralized Web server, which allows the probes to be viewed and managed remotely to gather voice, data, and video services performance information.”

Autonomous testing such as with QuOTA is relatively new but very attractive economically. Rather than have one or a few cars measuring network performance, with the QuOTA system, literally hundreds or thousands of probes can be used.

According to Mr. Ortiz, “They may be deployed in fixed locations such as base stations, airports, or retail stores. Probes could be installed in taxis, buses, or delivery vehicles and even carried by friendly cellular customers. Of course, this last opportunity is by far the best application because the network’s performance is being measured as the phones are being used where the customers live and work.”

References

1. Lee, W. C. Y. and Yeh, Y. S., “On the Estimation of the Second-Order Statistics of Log Normal Fading in Mobile Radio Environment,” IEEE Transactions on Communications, Vol. 22, pp. 869-873.
2. Bush, J., “The Significance of Raleigh Fading in Coverage
Measurements and 40-Lambda Criteria,” www.bvsystems.com

November 2008