Digital radio supports wireless communications at 1996 Olympic Games

In what became an inspiration for runners everywhere, Pheidippes, a Greek soldier, ran 23 miles from Marathon to Athens to relay news of victory against the Persian forces of Darius the Great. That was in 490 BC. To honor this hero, the modern Olympic Games created the marathon competition, forever establishing the symbiotic relationship between athletic achievement and communications. During the past century, since the rebirth of the Olympic Games in Athens in 1896, the Games have consistently grown in size and scope. This summer’s Centennial Olympic Games are remarkably different from the first modern Games. Only 311 athletes from 13 countries attended those first events. More than 10,000 athletes from 197 nations will participate in Atlanta in 1996. And although only about 100,000 fans watched the original Games in Athens, more than 2 million will enjoy the Games in Atlanta. In addition, it is estimated that two-thirds of the entire world population–3.5 billion people–will watch some portion of the Games on television.

To put it in perspective: The Centennial Olympic Games in Atlanta will be the largest sporting event in the Twentieth Century. Managing the event will command resources equivalent to staging 12 Super Bowls in one city–in one day–for 17 days.

Complementing this increased athletic participation and larger world audience, wireless communications technology has become an indispensable tool in executing the Games, training the athletes and communicating their triumphs. My employer, a partner-level sponsor of the Centennial Olympic Games, has been active at various levels of Olympic sponsorship since 1972 and has been a primary catalyst in the development of wireless communications for the Olympic movement.

The heart of our partner-level commitment to the 1996 Olympic Games is embodied in a state-of-the-art FM digital two-way radio network. It will be the largest and most sophisticated two-way radio communications network ever to be employed at an athletic event.

The Olympic Games network is designed to meet the extensive communications needs of the tens of thousands of staff and volunteers who are responsible for security, transportation, event management and countless other concerns. The “backbone” radio system will be supplemented with our pagers, cellular phones, computer modems and secure two-way communications equipment.

Customer needs analysis To get the most accurate sense of what demands would be placed on the communications network, extensive groundwork had to be laid early in the design process. Working in concert with members of the Atlanta Committee for the Olympic Games (ACOG), we conducted extensive research into ACOG’s requirements for the successful operation of the Centennial Olympic Games. This needs-analysis defined and clarified ACOG’s basic operational concerns and communications objectives and revealed its key performance requirements before the process of network-building could begin. Essentially, a network designed for a large-scale event such as the Olympic Games has to effectively manage communications among thousands of users.

We started the needs-analysis process for defining the wireless communications network more than four years ago. Network designers and engineers met with ACOG staff, and two of our technology executives were “loaned” to the organization to drive technological development.

The network required meticulous planning across the board, and it was very much a “learn-as-you-go” situation. Because advancements in two-way communications technology have paralleled the growth in size and scope of the Games themselves, we could not assume this network merely required refining previous solutions, such as the one used in Los Angeles for the 1984 Summer Games.

The Los Angeles network served competitive venues and facilities spread out over an area with a 100-mile radius. At the time, the availability of radio frequencies was not a pressing issue. Trunking technology was just coming into its own during that period, and the concept of a two-way trunked radio network, which we implemented at the 1984 Summer Games, was an innovation slightly ahead of its time.

The Atlanta Games’ needs were more complex to define than those in Los Angeles 12 years ago. The technological solution that worked for Los Angeles would not work for the Atlanta Olympic Games.

Seventy percent of Atlanta’s Olympic Games activity will be concentrated in the Olympic Ring, an area in the city with a 3-mile radius. The remaining events will occur throughout the state and across the Southeast. A network had to be designed to accommodate operations in both the metro area and in the remote venues.

In addition to immediately apparent differences in the Games’ geographic locations, the Atlanta Games’ communications network also had to contend with a relative scarcity of available radio frequencies and, at the same time, had to accommodate more than twice as many users as the Los Angeles Games network. These factors created a whole new set of operating characteristics that would have to be addressed when in the network design.

Once these initial factors were determined, network designers conducted extensive customer research to obtain an accurate impression of ACOG’s vision for the network’s performance. In October 1994, our engineers and executives began interviewing the future network users, getting an immediate sense of their communications needs and operational plans. This gave us essential impressions of ACOG’s operational plan and ACOG’s intended use of its communications equipment to its own effectiveness.

As we were clarifying ACOG’s communications needs, we also determined that, in addition to being extremely efficient, the network would also have to interoperate with area public safety agencies. Such interoperability capabilities will allow ACOG to interact with law enforcement and emergency response forces at every venue of Olympic Games competition.

The needs analysis considered all these elements. Using this information as the starting point for the network’s development, we began building the communications system that would meet ACOG’s needs. The resulting recommendations called for a flexible, highly sophisticated network built on our Astro FM digital two-way radio platform. Similar to the many large-scale public-safety systems that we have designed for federal, state and local governmental agencies in the United States and worldwide, it is the foundation for ACOG’s wireless network.

The essential systems operating in the network are Smartnet II digital trunked simulcast systems. A six-site system will serve the extended metro Atlanta area. Another two-site system will operate throughout the Olympic Ring. A single-site system will be installed in the Olympic Village at the Georgia Institute of Technology to coordinate on-site logistics and operations. These systems, although independent of one another, will be tied together through our SmartZone technology for full, wide-area connectivity. For example, a user in Stone Mountain Park, 16 miles away, will be able to speak with a user at the Olympic Stadium.

In more remote locations, stand-alone radio systems will serve the needs of each venue outside of the metro Atlanta area. Finally, two Motorola Advanced Systems Communications Trailers (MASCOT) will provide transportable trunked radio systems for maximum flexibility to handle communications at various locations prior to and during the Games.

System integration, site preparation Once our company and ACOG had determined the specifics for the network, many technicians and engineers began the process of integration. A key to this successful phase assembly was our Customer’s Center for Systems Integration (CCSI).

The 38,000 square-foot CCSI, located at our Schaumburg, IL, headquarters, allowed us to perfect the network to ACOG’s exacting specifications while it essentially still was on the factory floor, close to all necessary technical personnel and resources. The CCSI facility allows customers to completely test wide-area communications systems exactly as they will reside in their installation locations before they are shipped from the factory.

The CCSI was fundamental to the successful development of the network, which we began constructing on Sept. 5, 1995. In less than three months, the network was staged and tested to ACOG’s satisfaction. Without the ability to stage a network of this complexity at the CCSI, it would have taken more than twice the amount of time and personnel to complete the Olympic Games’ network.

With the network fully tested and shipped, ACOG and our company have started preparing the infrastructure installation sites. Overall, three wireless communications towers and six equipment shelters will be constructed for the network, supporting network controllers, 250 repeaters and ancillary equipment such as antennas and transmission lines. The tower construction was scheduled for completion in January. The remaining equipment storage sites and remaining infrastructure components were scheduled for completion in February, with final testing of all sites scheduled for completion by April.

Preparation for the Games With the network designed, tested and in the initial phases of installation, now we are preparing for the Games themselves. With the summer approaching rapidly, the network’s users must be trained, devices must be programmed and the equipment must be field-tested to ensure overall satisfactory operation.

To complete this phase, nearly 10,000 mobile and portable radios functioning on the network were to be shipped to their operation sites by early spring. Once there, they were to be programmed by a team of 20 technicians to match their operability to a pre-established “fleet-map,” that defines the devices’ areas of functionality.

After the devices have been programmed, users are taught how to operate them. Throughout April and May, ACOG staffers, designated as communications technology “champions,” are to be trained so they can in turn train the remaining network users. We also are designating training videos for each particular communications device to provide users with easy-to-follow directions on how their communications devices perform, and how each device will enhance users’ effectiveness.

To complement the training and the video, each device–two-way radios, pagers and cellular phones alike–will have an accompanying help-card listing tips for easy device operation. A radio distribution center will be located in each venue for additional backup and support. Controlled by the venue’s radio communications manager, the center will have spare parts on hand for immediate replacement and a spare battery for every radio issued. It will facilitate efficient, decentralized inventory control and distribution management.

To support network operations, we will also have about 100 service technicians on-site working with ACOG staff throughout the 17 days of the competition to manage network operations. System managers will monitor network activity around the clock, and additional employees will be on hand to lend technical support.

Incorporating related technologies In addition to this cutting-edge digital two-way radio network and nearly 10,000 mobile and portable radios, we are incorporating other related advanced communications technologies into the overall Centennial Olympic Games communications effort. We will also supply 6,000 alphanumeric pagers, 1,500 cellular phones, 1,500 computer modems and secure two-way communications equipment.

Included among these communications technologies is our Integrated Dispatch Enhanced Network (iDEN), which incorporates four communications services (voice dispatch, full-duplex telephone interconnect, text messaging and future data capabilities) into one network–operating on one device. For the Games, its wide-area capability was an ideal way to meet the needs of the Olympic Family transportation system.

The cellular phones and pagers will be used throughout the Games, meeting the diverse communications needs of each event venue. These responsibilities will include security and transportation, as well as event management needs such as medical support, event scoring and timing, judging, and food and beverage services.

The computer modems will link the computers throughout each venue, connecting them to provide reliable, rapid information exchange, while the secure two-way communications equipment will serve the Olympic Game’s security forces via the landline telephone network.

Looking ahead to summer Throughout the last century, the complexity and effort required to execute the Olympic Games has paralleled the Games’ ever-increasing exposure and appeal. Fortunately for the world’s eager Olympic fans, the technologies that make it possible to stage such a dramatic event on such a grand scale have evolved to keep pace with the Games’ popularity. Without the ease of reliable communications, coordinating this global event on such a scale would be extremely difficult. The innovative wireless communications system that we and ACOG developed is a big reason why the Centennial Olympic Games in Atlanta will showcase some of the world’s finest athletes to the largest world audience in history.

Simulcast paging operators can now take advantage of a significant advancement in system test and measurement: signal performance mapping as seen by an actual pager. Signal performance maps can display the bit error rate (BER), codeword error rate (CWER) and two forms of peak edge (PE) error (errors in the location of bit cell edges) for the received paging signal throughout the geographic area measured.

The minimum measurement system consists of an Advanced Signal SignalPro signal analyzer and a modified Motorola ProEncore or NEC MessageMaker Exec pager. As this article was being written, we were aware of no other combinations of commercially available products that would allow this type of performance mapping to be conducted. The pager is modified to provide an output for the receiver’s recovered audio signal. This signal is fed into the signal analyzer for decoding and signal analysis. The signal analyzer also logs the analysis results along with a physical location determined by an internal Global Positioning System (GPS) receiver.

A fundamental goal of our entire industry is to ensure that subscribers receive reliable, error-free data. Performance maps are a quantitative representation of a simulcast system’s ability to deliver error-free data. The visual presentation of the bit error, codeword error and peak edge error on a map provide the operator with a clear indication of system coverage as seen by an actual pager.

The maps offer insight into the reasons for poor system performance. The maps also provide an ideal method for evaluating the effect of changes in operating characteristics or equipment in a simulcast system, such as deviation, deviation offsets and effective radiated power (ERP). How do you measure the effect of a change in a simulcast system offset plan? It is easy to evaluate if “before” and “after” performance maps are used.

The performance mapping method is an easy, cost-effective way to verify a system’s message delivery capability. It can be used in conjunction with our channel loading product and with the channel’s existing traffic.

Methodology In the June 1995 issue, a test and measurement methodology was described that uses the signal analyzer, a modified pager and an RF communications test set. [“Data Analysis Aids Setup of Flex Paging Channels,” June 1995, page 10.]The article stressed the importance of measuring a system’s bit and codeword error performance as seen by an actual pager. It also recommended using an RF test set in areas with poor performance to record RF levels and viewing the recovered audio signal on an oscilloscope. It identified the biggest challenge in implementing Flex protocol as locating and analyzing areas of poor performance. The following information describes the data that are made available with that methodology.

Equipment used to collect the data included a SignalPro signal analyzer with a built-in GPS receiver, a modified Motorola ProEncore pager, a Hewlett-Packard Omnibook 600C (486DX2, 75MHz) laptop computer, an HP8920A RF communications test set with a battery pack and a portable power inverter. The inverter produces 115Vac from the car battery through a cigarette lighter adapter. The equipment was placed in a car with the pager clipped to the passenger-side sun visor, and a magnetic-mount GPS antenna was affixed to the roof. The inverter was used to power the signal analyzer and personal computer during the three days of data collection.

The mapping software used to create the accompanying figures was ATLAS GIS from Strategic Mapping. ATLAS GIS allows text files that identify data-collection points to be imported and overlaid on detailed street maps. A text file with the latitude and longitude for each transmitter site was created for display with the street maps to depict the expected coverage area. Next, the mapping software was used to produce regional maps within the coverage area. A 1-mile grid pattern was marked on the regional maps to identify potential measurement locations. The potential measurement points were adjusted to coincide with street intersections, and the cross-street names were obtained from the mapping software database. Identifying most of the points prior to the test minimized the time spent collecting data.

At the start of each day, the signal analyzer and console personal computer were switched on, and the internal GPS acquired a position fix. The position fix took about the same time that was required to travel to the first measurement location. Once at a measurement point, the signal analyzer uses GPS position data to determine that it is at a fixed location. The signal analyzer then clears all statistics and begins analyzing paging data received from the pager. Performance data are accumulated on the live customer traffic already present on the channel. Communication with the office to request more test pages is not required. If customer traffic is not yet active, our channel loading product can be used to generate traffic for the channel. Bit error and codeword error rates for each format detected are accumulated over a four- to five-minute interval. This amount of time allows the signal analyzer to see a sufficient number of bits for a meaningful measurement.

Also, during the measurement interval, the worst-case maximum peak edge error and peak edge 75th percentile are logged. Peak edge error is a measure of the error seen in the location of bit edges for a given data rate. The max peak edge error value is the worst edge error that is measured, whereas the 75th percentile value indicates that 75% of the edges measured had an edge error at or less than the value displayed. An example of the logged data is shown in Table 1 on page 60.

The signal analyzer produces a series of tones to indicate that a data point has been obtained. When the analyzer is connected with a personal computer, data also are displayed on the channel metrics display window. Once the data point is obtained, the operator is free to drive to the next location and repeat the process. At the conclusion of every eighth data point, the signal analyzer logs the previous eight points to non-volatile memory. As many as 190 points can be logged in this manner. This method is useful if a console is not used in the collection process. At the end of the day, the signal analyzer can be instructed to download its analysis log to a console.

Periodically throughout the test interval, an HP8920A was used to obtain a measure of the RF field strength for the desired channel and the strength of adjacent channels. Also viewed was the recovered audio signal on the oscilloscope display of the HP8920A. Care was taken to make this observation for locations where the bit error was non-zero. Signal strength, and whether the signal exhibited fades or severe spikes on the data edges (which would indicate simulcast misalignment), were noted on the regional maps. This information also plays a role in the analysis of poor performance areas.

Displaying data on a map The first step in displaying the performance data is to sort the data per format. This sorting is done with a utility provided with the signal analyzer console program. In this case, the sort utility produced two files, one containing the POCSAG 2,400-baud data and another containing the Flex 6,400-baud data. These files were then copied to the computer where the mapping program resided.

Table 2 on page 64 provides a sample of each sorted file. Next, the mapping program was started and the POCSAG 2,400-baud data file was opened. Within ATLAS GIS, the user selects columns in the data file to indicate latitude, longitude and the variable to be plotted. A theme is then specified that determines how the data is presented on the map. A “range plot” on the POCSAG 2,400-baud CWER column was selected. (See Figure 1 on page 58.) A range plot places a colored symbol at the specified coordinates based on the range in which the value in the CWER column falls. As an example, for each location where the CWER was less than 317 X 10^-6 (317 per million), a green circle is placed. A different map can be generated by selecting another variable or a new set of ranges.

Analyzing the performance maps The performance maps presented here are for a 900MHz system operating in the St. Louis metropolitan area. Conditions in St. Louis are generally good for successful message delivery. The transmitters for this system are on building tops about 100 feet high, except for the downtown site, which is on a taller building. The use of any high-gain or pattern-altering antennas is unknown. The terrain is generally flat with some hills and valleys about 100 feet above or below the average elevation. The following is a brief review of the four factors affecting performance in this system.

* RF signal fades or low signal level — RF signals fade because of multipath and the characteristics of narrowband FM noise. Fading conditions constantly change with time. The depth of a given fade generally is modeled as a random variable with a Rayleigh or Rician probability density function. Fades of 15dB or more occur in most simulcast systems. Consistently low signal levels exist primarily because of shading from the terrain or other large obstructions.

* Simulcast misalignment — Simulcasting of multiple RF signals occurs when their RF levels (as seen by the receiver) are within roughly 10dB of each other. Decoders of simulcast signals generally can tolerate a misalignment between the signals of one-fourth of the bit time. For a 2,400-baud signal, this interval is 104 microseconds, and for a 3,200-baud signal, the interval is 78 microseconds. Also consider the affect of the RF exciter’s rise time on the tolerable misalignment. For a 3,200-baud signal, an exciter rise time of 88 microseconds consumes 28% of the bit time. If this interval is taken into account, the signal spends only 225 microseconds at its desired deviation. One-fourth of this time is only 56 microseconds. Depending on which value you subscribe to, the tolerable propagation difference between two signals, simulcast at 3,200 baud, is 10 to 14 miles.

* Combinations of fades and simulcast misalignment — Many consider simulcasting to occur only at locations roughly equidistant from two or more transmitters. Consider the case when the primary (or nearest) signal fades or is otherwise blocked by a hill or obstruction. If secondary (or more distant) transmitters are capable of supplying a signal at this location within 10dB to 15dB of the primary (prior to a fade), then some percentage of fades will result in a simulcast condition between the signals. If the propagation difference for the distant transmitters approaches a quarter of a bit time, simulcast misalignment situations will exist. This condition can, and does, exist in many simulcast systems.

* Interference — When a signal level in a neighboring channel is significantly higher than the desired signal, energy from the neighboring signal may not be sufficiently rejected by the IF filters within the receiver. The remaining energy disrupts the decoder’s ability to accurately decode the data.

The ranges selected for each of the maps indicate where excellent, marginal and poor message delivery performance can be expected. Green symbols indicate excellent performance, yellow and orange symbols indicate marginal performance, and red symbols indicate poor performance. The codeword error rate map is used for the following discussion because it most closely represents the probability of a missed or garbled message. Codeword error rate represents the proportion of codewords received that contained three or more bit errors, rendering the codeword uncorrectable. Codeword error rates take into account the bit error correction embedded in the paging format. The peak edge error ranges were selected to place a red symbol where peak edge error reached 25% of the bit time. The green, yellow, and orange symbols subdivide the remaining error range. The RF levels measured by the RF test set were added manually to the maps.

Note that several measurements were taken at some locations. This was done because the pager antenna can be highly directional. If the pager’s orientation had an effect on the bit error rate, additional readings were taken at different orientations. This is why some locations have more than one colored symbol on the error maps.

The POCSAG 2,400-baud codeword error rate map identifies six poor performance areas. (See Figure 1.) Area A is subject to interference from a channel 100kHz away. The interferer is 30dB to 40dB greater than the desired signal. The interference was so strong that when the channel was keyed, its signal was audible along with the desired signal. Area B is adversely affected by a consistently low signal level. A -95dBm RF level is too low for the pager and the signal analyzer to decode POCSAG 2,400-baud reliably. “C” identifies an area that is subject to 10dB to 15dB RF fades. Because the signal is weak to begin with in these areas, fades of 10dB to 15dB are enough to drop the signal level too low for reliable decoding. Any visible, or audible, evidence of simulcast misalignment in these areas was not noticed.

The data displayed on the POCSAG 2,400-baud peak edge error map for each Area C show errors of 70 microseconds or less. (See Figure 2 on page 58.) This supports the conclusion that simulcast misalignment did not play an important role in degrading the performance at 2,400 baud.

The Flex 6,400-baud codeword error rate and peak edge error maps identify additional marginal or poor performance areas for the higher data rate format. (See Figures 3 and 4 on page 68.) To minimize clutter, the Areas A, B and C from the POCSAG 2,400-baud maps are not repeated on the Flex maps. These areas perform poorly for Flex 6,400-baud for the same reasons as POCSAG 2,400-baud. On the Flex maps, there are two additional A areas where an interfering signal was able to affect Flex decoding. The interferer(s) were 75kHz to 100kHz away at a level 25dB above the desired channel. Errors were indicated on the front panel of the signal analyzer when the other channels were keyed. There is one additional Area B where the RF level was -93dBm. This level is too low for the pager and the signal analyzer to decode Flex protocol reliably. Five additional areas are marked solely with a C. These areas are affected by RF fades that dropped the signal level below that for which Flex protocol can be reliably decoded. A “D” identifies those areas where an RF fade resulted in the simulcast of misaligned signals. As the primary signal fades, it simulcasts with the signals from more distant transmitters. For Flex 6,400-baud, the distant transmitters are far enough away to result in errors due to misalignment.

The Flex 6,400-baud peak edge error map also supports this conclusion with measured peak edge error measurements from 30 microseconds to 70 microseconds in those areas. (See Figure 4.) There are two areas marked with a C and D. These areas have a slightly lower peak edge error reading, but the combination of a faded signal and misalignment work together to degrade the performance at these locations.

Both the POCSAG and Flex peak edge 75th percentile maps were all within green limits. This indicates that there are no consistent simulcast misalignment areas, but rather conditions that produce a brief fade of the primary signal, which now simulcasts with a signal from a distant site. This condition, however, does not strongly affect the 2,400-baud data because the distant sites in this system are well within the 21-mile limit for propagation differences. The 3,200-baud data are altered because the sites are a net distance of 10 to 14 miles away from the affected areas. The signal strength data in Figure 2 bring to light the fact that signal strength is not always a clear indicator of system performance at higher baud rates.

Conclusion Performance maps provide a visual representation of system capability over a geographic area. Error rates are obtained using customer traffic. The operator does not have to watch for test pages that only account for a small portion of the channel’s transmission time. Data are acquired quickly and accurately using the signal analyzer with a modified pager. The RF test set (in this case, the HP8920A) provides additional data that, when combined with the maps, give the operator unparalleled insight into the root causes for poor performance.

An ideal use for performance maps is to establish a baseline for judging the effectiveness of changes to system operating characteristics or equipment. Any substantial change (such as a system offset plan) can be followed up with a new map to confirm its effect in a quantitative manner. Maps give the operator a means of addressing problems with the entire system in clear view. The operator can address multiple coverage problems by using a system-wide strategy. This will lead to a lower-cost solution than would be the case if each coverage problem were addressed individually.

The data and figures presented here represent what can easily be done when using a signal analyzser, a modified pager and an RF test set to measure the performance of a simulcast paging channel. Using a signal analyzer and a modified pager allows you to identify problematic areas, as seen by a pager, within a system. A signal analyzer, a modified pager and an RF test set allow you to identify problematic areas and to determine a cause. Essentially any system problem can be identified with this test and measurement methodology.