We all know that there are many flavors of wireless (Figure 1); different frequencies, different data rates, different power requirements, different prices. This shouldn't be a problem when there's a lot of room, so we decided to measure what happens in a confined area with lots of wireless devices operating. We chose the show floor at ISA Expo 2005 for our test.
The majority of the many devices with antennas—seen by individuals as they walked around the show floor—operate within the ISM frequency bands at 900 and 2450 MHz. In a quasi-scientific process, we walked around the exhibit hall performing spectral measurements to ascertain ISM-band channel congestion. The results presented here do not represent some definitive conclusion on technical issues, such as a functional and coexistence comparison of (seemingly) esoteric (but very important) topics such as frequency-hopping spread spectrum or direct-sequence spread spectrum systems. Rather, we present some of our findings in this setting; in essence, a mini site survey.
Figure 1. A snapshot of types of wireless, ca. 2005
With more than 90% of all equipment/instrument vendors displaying some components and systems with wireless capabilities, wireless was all over the exhibit floor. In many cases, the wireless devices weren't actually operating; they were there as show pieces. The same could not be said for the exhibit hall's Industrial Wireless Pavilion. As the name implies, the vendors and organizations located in that section were busy displaying functional devices. A quick unscientific query of the exhibitors revealed that the vast majority of the products were set to operate in the unlicensed Industrial, Scientific, and Medical (ISM) frequency bands. While there are many ISM frequency bands available for use in the U.S., (Figure 2), most of the wireless traffic was concentrated in the 902–928 MHz and (+/-)2450 MHz frequency bands because of bandwidth and operational restrictions.
Figure 2. License-free ISM frequency bands
In standards-compliant wireless operation, most devices have gravitated to using either an IEEE 802.15.4-compliant wireless channel or an IEEE 802.11b/g-compliant channel. Please note that not all of the exhibited devices operated under IEEE compliance. Some were running their own protocol and broadcasting in the ISM bands. Numerous sensors/instruments/transmitters all attempting to operate in the same 900 and 2400 MHz channels resulted in considerable congestion and coexistence issues.
The channel assignments for 802.15.4 are shown in Figure 3. Figure 4 shows those for 802.11b. Devices operating in either of these bands do not need to comply with either of these standards—plenty of wireless devices are simply sensors/devices with a "conventional" wireless transmitter slapped on. As long as the transmitter's radiated power doesn't exceed the maximum value, the device is "compliant" with operating in an ISM band.
Figure 3. The 802.15.4 frequency channels in the 2450 MHz range
On to the Measurements
As part of our quasi-scientific approach, we obtained a background noise-floor spectral snapshot by walking onto the exhibit floor at 7 a.m. on Wednesday, October 11, 2005, turning on the Anritsu MS2721A portable spectrum analyzer, and tuning it into the 900–930 MHz range. Very few individuals were present, setting the stage for a "quiet" moment to take some measurements. The noise floor measured in the 900 MHz range is shown in Figure 5. The amplitude level was relatively flat at –82 dBm. As expected, no wireless signals were measured. (A similar case was found for the 2400 MHz band—this was somewhat surprising, since we predicted that someone would have left a WiFi base station activated, yet none were found.)
Figure 4. The channel assignments for 802.11 operation in the 2400 MHz range
A different electromagnetic "scene" was found at 12:30 p.m. on the same day. The Expo floor was buzzing with vendor displays operating and hundreds of people walking around. We picked up the spectrum analyzer and returned to the same physical location on the exhibit floor where the earlier noise-floor measurement was taken. With the analyzer set to the same sampling/measurement settings, we found significant congestion (Figure 6). The peak amplitudes shown in Figure 6 top out around 5 dBm, with local minima in the –25 dBm range. This shows a 67 dB increase (in regular numbers, it increased by a factor of over 5,000,000).
Figure 5. The very quiet noise floor at 900 MHz, measured early in the morning before the show opened
Later in the day we grabbed the analyzer and replicated the 900 MHz measurements. We performed a time average to ascertain the average channel activity, and the resultant graph is presented as Figure 7. It was still electromagnetically "busy" in the exhibit hall.
Figure 6. At 12:30 p.m., the 900 MHz band had reached saturation. The "background" noise floor has risen by >60 dB (a 5,000,000X increase)
The 2400 MHz ISM band also had its fair share of wireless traffic. As shown in Figure 8, the spectrum, measured at 12:30 p.m. at the same location as our previous measurements, showed substantial channel traffic.
Figure 7. An averaged view of the 900 MHz ISM band
Does This Matter?
It is difficult to present definitive information on the impact of such RF congestion on system performance. We obtained anecdotal "evidence" by walking around to various exhibitors and asking them discreetly how their demos were functioning. The results of this study verified what the actual measurements showed— there was significant channel congestion and the wireless systems didn't always function as the exhibitors had hoped.
Figure 8. Spectral measurements of the 2.4 GHz ISM band showing a profusion of radio traffic
We reiterate that this was, at best, a quasi-scientific study conducted in a somewhat unnatural setting of an exhibit hall with literally hundreds of vendors' wireless systems operational. On the flip side, if the analysts' forecasts are right, then with thousands of wireless sensors and networks being deployed throughout industrial settings, it is probably worthwhile to pay attention to congestion in ISM band traffic and the associated negative effects of co-channel interference.
Read Up and Get Involved
In addition to their duties at their respective companies, Fuhr, Kagan, Conant, Gutierrez, and Manges are on the Executive Teams of WINA (the Wireless Industrial Networking Alliance) and ISA's SP100 Industrial Wireless Standard committee.
The interested reader may wish to follow the developments of two IEEE standards groups that are working toward addressing the coexistence issue, namely P802.15.2—Recommended Practice for Coexistence in Unlicensed Bands and 802.19—Coexistence of Wireless Data Transport.
We specifically invite you to participate in the activities of WINA (www.wina.org) and ISA's SP100 Wireless Standard (www.isa.org/community/sp100). These dynamic organizations are attempting to bridge the chasms between IT, wireless, and industrial bus systems, with a focus on developing standards for coexistence and interoperability of networked systems in industrial settings.