Pursuing Agile Signals in a Crowded Spectral Environment

For the capable RF engineer, continuous- wave (CW) and predictably repeating signals are no great challenge. However, design and troubleshooting become difficult when dealing with agile signals, and the challenges grow when these signals exist in an environment densely populated with similarly agile signals. Examples include applications such as radar, electronic warfare, wireless connectivity, and wireless communications. Indeed, some engineering jobs must be performed where two or more of these technologies intersect — sometimes accidentally and sometimes deliberately.

Figure 1: With multiple agile signals sharing a 100-MHz frequency band, it can be difficult to understand signal behavior using a swept spectrum analyzer — even with the peak-hold function over a long period.

To help engineers keep pace with evolving requirements, new types of signal analyzers and application software have emerged in the past several years. This article presents recent tools and techniques that are useful for not only discovering elusive signals, but also taking the vital next steps toward pinpointing and solving complex RF problems.

Working in a Dynamic Signal Environment

Around the world, the industrial, scientific, and medical (ISM) band at 2.45 GHz is perhaps the most varied and dynamic signal environment. In many locations, it is lightly regulated and heavily used, and typical occupants include WLANs, Bluetooth PANs, cordless phones, and high-power microwave ovens.

Because transmissions are not generally coordinated, there are many opportunities for complex interactions due to interference, collisions, retransmissions, and so on. At a certain channel loading, collisions and any increase in channel occupancy due to retransmissions can cause a sudden drop-off in effective channel throughput (i.e., a “cliff effect”).

Behaviors such as channel scanning may be a combination of brief, wideband, and rare — a fraction of a second long, tens of megahertz wide, and occurring every few seconds or minutes. This makes these behaviors difficult to see with traditional swept- or FFT-analyzer technologies, and this is when real-time spectrum analyzer (RTSA) capabilities are especially useful. Fortunately, RTSA is no longer limited to costly, single-purpose instruments — it is now an upgradable option to widely used signal analyzers.

Going Beyond Peak Hold

Figure 2: With a setup similar to that of the swept spectrum analysis approach, an RTSA density display quickly reveals detail about the spectral occupancy of this band.

The 2.45-GHz ISM band is thus both dynamic and complex, and a good example of the challenges inherent in agile signal analysis. Traditional swept spectrum analysis is not an effective way to understand the activity in this band, as shown in Figure 1.

Depending on the degree of spectrum/ time occupancy, a single sweep of a spectrum analyzer may show either nothing or only a portion of one or more signal bursts. It can be difficult to interpret such measurements, especially because the dynamics of the analyzer’s sweeping resolution bandwidth (RBW) filter interact with the dynamics of the signal itself.

Peak hold is a useful tool for understanding some aspects of the signal environment, and a long measurement with peak hold will eventually catch most of the signals in the band. However, long peak-hold measurements often result in some signals obscuring others, as shown at the right in Figure 1.

Figure 3: A spectrogram display of the ISM band summarizes signal behavior over a period of seconds, revealing mostly WLAN and Bluetooth signals.

A real-time analyzer is a highly effective alternative. Fast processing and advanced displays make it a good fit for exploring dynamic signal environments, as shown in Figure 2.

This density display from a real-time analyzer provides an immediate understanding of the ISM band and the signals it contains. Because the measurements are gap-free and all signal samples are represented in the display, it is possible to see most of the signals in the band either at a glance or over a short measurement time.

Density displays are very data-dense and quite dynamic, updating about 30 times per second and with an adjustable amount of persistence applied to fade older data. With an FFT rate of almost 300,000 per second, each display update in an RTSA represents about 10,000 spectra. The result is a responsive display able to keep up with in-band activity and show subtle details such as signals inside other signals, and signals near the analyzer noise floor, even when such signals are small and infrequent.

It is worth noting, however, that the action of combining 10,000 spectra into one display update can cause signals present at different times to be displayed in the same display update. For example, the signals that appear to be multitone in Figure 2 are actually repeated Bluetooth frequency-hop patterns.

Observing Changes Over Time

Figure 4: Selecting an acquisition time of 1 ms rather than 30 ms provides increased time resolution, revealing more about the structure of the WLAN bursts and Bluetooth hops

The real-time spectrum display (Figure 3) is another way to understand signal behavior over time. The vertical axis of the spectrogram is time, which can reveal important aspects of signal behavior. Here, many Bluetooth hops form a repeating pattern, and other bursts appear to be isolated, mostly in the upper half of the spectrogram. Also note the diagonal bars that move between the wide WLAN channels in the lower half of the spectrogram. These may have been the result of channel scanning and were occasionally visible in the density displays.

At the upper right of the display, the acquisition-time setting controls how individual spectra are combined into spectrum updates (top trace) and individual spectrum lines to form the display. A longer acquisition combines more spectra into each line and causes the spectrogram to update more slowly. This allows a single spectrogram display to represent a longer time period.

Selecting a shorter acquisition time for each update — spectrum display or spectrogram line — provides better time resolution, as shown in Figure 4. In such cases, the buffer and display will cover a proportionally shorter span of time and therefore may not show some longer-term phenomena. However, the additional time resolution can reveal important spectral behavior that would otherwise be obscured.

In this case, increased time resolution shows more detail about individual WLAN bursts and Bluetooth hops. Two things are clear: Bluetooth hops that overlap WLAN bursts in frequency often do not overlap in time and, thus, collisions are not as frequent as the previous display would suggest. Note, however, that each spectrum line still represents hundreds of individual FFT results from the real-time measurement engine.

Going Deeper with Vector Signal Analysis

Figure 5: FFT analysis of a short time record (top) is likely to provide very low POI. For most agile signals, higher POI is more likely with a much longer time record (middle) and enhanced displays such as color persistence (bottom).

Spotting an elusive signal or event is often just one step in finding and solving problems or in optimizing performance. In these cases, vector signal analysis (VSA) software is a logical and powerful complement to RTSA.

VSA solutions generally take advantage of the same RF and signal-processing architecture used in a swept analyzer with a digital IF. VSAs add gap-free signal-capture and vector-processing capabilities such as analog and digital demodulation.

Vector signal analysis often begins with FFT analysis of a digitized IF signal. For agile signals or dynamic environments, FFT analysis removes the measurement variability and uncertainty caused by a sweeping RBW filter.

The entire spectrum measurement is calculated from a block or “time record” of a digitized IF signal provided by a compatible spectrum/signal analyzer or RTSA. In the VSA software, the data used for the FFT can be further refined through the use of time gating to select any portion of the time record for analysis.

A typical default time record in vector signal analyzers is relatively short at 1,000 time samples; this is similar to the longest time records used in real-time analyzers. The resulting spectrum is shown at the top of Figure 5.

The longer time records possible with a VSA provide a better view of many signals and a much higher effective probability of intercept (POI). Using a long time record and enhanced displays such as persistence and density can create signal views that are highly informative and significantly closer to real time on a continuous basis (Figure 5).

Getting a Comprehensive View

Figure 6: A spectrogram created from playback of a large, gap-free time capture shows every event in the ISM band in detail, with a high-resolution time scale.

The VSA’s ability to capture large blocks of gap-free vector signal samples is particularly useful in understanding agile signals on their own and in complex signal environments. The ISM band viewed previously is shown again in Figure 6, where the large capture buffer is organized in playback as a gap-free spectrogram covering about 26 ms. WLAN bursts, Bluetooth hops, and wandering signals from microwave ovens are shown clearly. Any individual signal or burst from the large capture memory could be selected for analysis or demodulation.

The complete time and frequency behavior of all signals in the ISM band is shown clearly, with “slice” markers positioned at a 16.7-ms (60 Hz) interval to verify the power-line frequency of microwave oven signals. Band-sharing successes and failures are easy to see, and post-processing in the VSA software can be used to select, from the single capture, any single signal or burst for measurement and demodulation.

Continuing to Evolve

Complex and dynamic signal environments make it difficult to keep data payloads moving. Fortunately, tools such as real-time spectrum analyzers and VSA software help RF engineers address evolving challenges in design and troubleshooting. Going forward, these tools must continue to evolve with measurement capabilities that keep pace with signal behaviors that are increasingly brief, wide, and rare.

This article was written by Ben Zarlingo of Agilent Technologies, Santa Clara, CA. For more information, Click Here