Special Section: Wireless
Experts answer users’ questions on tracking wireless assets in industrial settings
By Peter Fuhr and Nacer Hedroug
Looking for your car keys, the remote, or the cat is fast becoming analogous in an industrial facility to finding a vehicle, sensor, or tank car. What was originally associated with the radio frequency identification (RFID) world has evolved into the more appropriate real-time location service (system), or RTLS, arena. While you could think of RFID as a database management system with readers who could inform the database as to where it last observed a tag, RTLS brings the realm of tracking devices into the industrial world.
Vendors are touting several technologies as optimal for industrial asset tracking, especially end users in petrochemical (on-shore/off-shore), pharmaceutical manufacturing, power systems (generation and distribution), and pulp and paper processing.
An industrial setting presents specific radio frequency (RF) and environmental ambient conditions that may be more taxing of a wireless system than seemingly any other site. In each of these locales, the amount of reflective and absorptive surfaces associated with the aptly named canyons of metal lead to variable attenuation and multipath conditions, which are frequently dovetailed into requirements for the wireless devices to operate in a non-line-of-sight situation. In practice, a radio signal may encounter many objects in its transmission path and undergoes additional attenuation depending on the absorption characteristics of the objects. There are many types of objects, including fixed, mobile, and transient objects that absorb RF energy and cause RF attenuation. Similar to the free-space propagation loss, higher frequencies attenuate much faster than lower frequencies. Therefore, 5 GHz RF signals typically have higher attenuation than 2.4 GHz, though there are a few exceptions.
The table shows how various objects introduce attenuation to RF signals at the 2.4 GHz and 5.8 GHz bands. Note the attenuation values for concrete. The large disparity in the values comes from different types of concrete materials in use in different parts of the world. In addition, the thickness and coating differ depending on whether it is used in floors or interior or exterior walls. Brick walls usually have attenuation at the lower end of the range shown here.
Not included in the table are attenuation values for a number of other materials commonly found on site. Measurements of RF signal attenuation caused by water reveal higher levels of relative attenuation of signal transmissions in the 5.7 GHz ISM bands than in the 2.4 GHz band. However, rain, snow, and fog attenuations are very small for frequencies under 10 GHz. The rain attenuation at 5 GHz is barely noticeable (< 1 dB per kilometer). Human bodies, composed of predominantly (70%) water, attenuate RF signals nearly 3dB in the 2.4 GHz range and 5dB in the 5.8 GHz range.
Four candidate technologies associated with industrial RTLS include: RFID, GPS-based, chirped frequency-based, and received signal strength indicator (RSSI)-based. Each technology has its own unique requirements for a supporting infrastructure to provide a somewhat comparable location resolution with, of course, an accompanying system cost.
RFID-based asset tracking
Different flavors of RFID range from systems with no active components within the tag to battery-powered semi-active tags, to fully active tags (microcontroller-based wireless devices, including the possibility for sensing). The key aspect of RFID systems is they require a reader, or interrogator, and the tags themselves. By using ISM (license-free) frequencies ranging from 125 kHz to 5.8 GHz, it is possible to tailor the entire system for specific needs. The separation distance between the reader and tag plays a significant role in the actual deployment of RFID systems. Directional antennas frequently see use to limit the field-of-view, and therefore the location resolution zone, for the readers.
Numerous variations exist to the above classic RFID system. In a passive system, the tag has no battery but reduces the reader-tag separation distance to less than 1 meter. Adding a battery into the tag provides an on-board source to power a radio, which leads to a situation where the separation distance may be 100 (+/-) meters. While innumerable matters might arise with an active tag (what radio to use, what frequency to use, what power to output, what protocol to use, and directional antennas), in this instance, the tags and readers topology is similar to that of any sensor and gateway topology. Such higher power active tags typically have considerable on-tag memory and a processor that allows for more information, perhaps even sensor-related information, to transmit upon receipt of an interrogation signal from the reader. Having a battery implies at some time you will probably have to replace it, hence the increased servicing needs, particularly in an industrial facility that may be hazardous or even present an explosive environment.
Hundreds of companies are currently offering RFID components and systems, but the general situation for an RFID-based system is the database can inform the user about which reader last observed the tagged asset. A requirement for higher resolution as to the asset’s location requires more readers to deploy.
GPS-based asset tracking
In order to track the asset using GPS, we must first determine the four variables associated with the asset, namely its position coordinates, Ax, Ay, Az, and the time of the determination, At. At , related to a value called clock bias Cb. As basic algebra instructs, in order to solve for four variables, you need four equations, then perform a simultaneous solution. The procedure used for GPS-based tracking requires the receiver on the asset to pick up signals from four satellites measuring the time it took for the signals to arrive. Using the simple equation R = V × t, where V is the velocity of the signal (in this case a constant, namely the speed of light) and t is the time it took to receive the signal; it is trivial to calculate R, the range from the receiver (asset) to the satellite.
Within the signal transmitted from the GPS satellite are its X, Y, Z coordinates at the time the signal went out. Having this positional information for four GPS satellites, coupled with the range between the asset’s receiver and each satellite, as well as the clock bias information, yields the four equations necessary to determine the position of the asset.
With simultaneous data received from four satellites, we can calculate the asset’s location (latitude, longitude, altitude, and time). Under ideal conditions, namely direct line-of-sight to the satellites (no multipath) and minimal ionosphere-induced variations, we can determine the location precisely and accurately.
The physical environment of an industrial setting is far from ideal. Non-restricted GPS signals transmit at 1.575 GHz, a frequency blocked by steel and concrete structures (buildings and tunnels), and attenuated by passing through trees and leaves. As such, receivers that just meet this specification are not reliable for use in forests or even tree-lined streets. Add to that the potential complexities and challenges of tracking assets mobile deep within the bowels of the industrial site or go in and out of buildings, and you can see fundamental limitations associated with using traditional GPS for industrial asset tracking.
Chirped frequency-based asset tracking
Since IEEE 802.15.4-2006 radio is the choice for the wireless underpinnings for ISA100.11a’s wireless field transmitter standard (not yet ratified), the end user community has sent out a call to ascertain if there is an easy way to obtain location and sensing information from an installed wireless field transmitter. While discussions pertaining to the actual need for such capability continue (the transmitter is bolted onto the tank, it is not going anywhere), there is a variant on the ISA100.11a radio technology that may readily provide such multifunction capability, an 802.15.4a radio. This radio relies on a technique for changing the center frequency of the transmission in a linear manner. The radio output begins at one frequency, and over time at a predetermined rate of change (slope), it changes linearly to another frequency, F(t) = F0 + (a × T), where F(t) is the output of the radio at time t, F0 is the radio’s starting frequency, a is the change of frequency with time (the slope), and T is time.
The system performance and range determination are easy to see using the analogy of chirped-frequency radar. Consider the case where a chirped-frequency radar transceiver that outputs a certain frequency, F1, at time T1. This signal proceeds down its merry way a distance X, taking time T, until it encounters a surface and is backscattered toward the radar transceiver. After another time, the radar transceiver receives T, the reflected signal. Remembering that distance = velocity × time (X = V × T), it has taken the signal a time interval 2T to travel down and back from the reflector. Since the speed of the electromagnetic signal is C, the speed of light (a constant), it is easy to find the distance, or range, from the radar transceiver to the reflector. This is how classic radar systems work. In the chirped frequency case, the situation is similar but back at the transceiver, instead of monitoring time, T, the frequency that the radar transceiver is currently outputting, F2 at time T2, is recorded. The difference in frequency ΔF = F2 – F1 is calculated and then by knowing the slope of the linear frequency ramp function, a, the time difference ΔT is determined. Knowing ΔT allows you to determine X, the distance (range) to the reflector.
While it may sound complicated, the reality is using a chirped-frequency radio for location information relaxes the time resolution requirements for range measurement resolution by shifting the measurement into the frequency domain. Back in the industrial field transmitter realm, there is a requirement for multiple devices to have connectivity to the asset under measurement (to provide the X, Y, and potentially Z coordinates of the asset). You can achieve this by having the asset within range of multiple gateways, or by using relative location information for other field transmitters within range of the asset. Typical location accuracy is in the cm range with an overall asset-to-asset or asset-to-gateway separation in the 1-100 meter range. An infrastructure or network/system connection must be in place to transport the range information to the appropriate software application.
Received signal strength-based asset tracking
Numerous techniques for RTLS are based on the strength of the signal received by the asset’s attached radio changing, and associating that received signal strength variation with a change in the separation distance between the gateway/access point and the asset’s receiver. The further a receiver is from a transmitter, the less the received signal strength is. This fundamental principle is based on the 1/r2 EM field law (sometimes referred to as the inverse-square law). In terms of communication systems, this means the received signal strength (RSS) follows:
where R is the receiver-transmitter separation distance.
In this scenario, the variation in the RSSI is predicated on all other parameters remaining constant so any change must be due to a change in distance between receiver and transmitter. Therefore if the receiver knows the transmitter’s output level, you can determine the distance R. This in turn leads to contour lines for the signal strength from a base station. It becomes an easy matter for the wireless device, which is attached to the asset, to report to the access point that it is associated with its RSSI value (along with its unique radio/tag ID). That information processes either locally in the AP (if it is intelligent) or goes on to a software application for it to determine the location of the tag with respect to the (typically) fixed location of the AP.
Knowledge of the tag’s RSSI value for a transmission from a single AP puts the tag within a certain range of the AP (in radial coordinates, the distance is known but not the angle). In order to determine the X, Y, and potentially Z coordinates of the tag requires the tag to be in communications range of a number of APs.
For improved location determination, the size of the APs’ RF footprints should be reduced (less transmitted power) with a higher density of APs being deployed.
A measurement based solely on intensity may be problematic; any change in the received signal strength may be interpreted as a change in separation distance between the AP and the tag. While in reality the variation may be caused by a decrease in the transmit power, decrease in the receiver sensitivity, or introduction of a new object that attenuates the signal. There are other issues associated with RSSI-based asset tracking. RSSI is not measured in specific units. Instead each wireless device vendor uses an arbitrary set of numerical units. It is incorrect, therefore, to attempt to match a given RSSI value with, say, a power unit such as mW, which leads to serious problems if devices from multiple vendors are supposed to interoperate. While not strictly for RTLS, RSSI is of great interest to any Wi-Fi device since much of the perceived performance of a wireless network is based upon inferences made via the use of RSSI.
The higher the RSSI, the higher the transmit data rate (up to a maximum); client devices tend to monitor the RSSI on a channel (frequency); when this value drops below a certain threshold, the device assumes the channel is clear to send and transmits data; the association and disassociation of client roaming between multiple APs is almost entirely determined by RSSI.
A popular tracking method is to have the X, Y, Z coordinates of a gateway, then use some flavor of communications ranging to determine the radio’s distance to the gateway. Mesh-networked topologies and associated protocols have demonstrated with each yielding some location information of the wireless device (asset) relative to the gateway. A similar approach sees use sometimes for cellular-based RTLS where a message transmits from numerous cell towers (base stations). The cell phone associates with a cell tower (base station). If the location of the cell tower is known along with the tower’s RF footprint, it achieves RTLS-like localization of the cell phone (it is nearby this certain tower). Higher spatial resolution requires more cell towers each with a smaller RF footprint. Variants on this approach may use a common message transmitted from a number of cell towers. If the cell phone is within range of at least three towers, then similar to the GPS simultaneous solution of multiple equations, using the difference in times for the similar message the cell phone will receive, you can determine the location of the cell phone.
End users have heard several statements regarding how wireless systems based on ultra-wide bandwidth may provide better RTLS performance. With no specific standards to follow, these systems tend to rely on proprietary communication ranging techniques similar to those we just described. The use of the spectrally much wider bandwidth allows for more sophisticated, and longer communications messages to transmit (leading to adaptive correlation-based receiver systems), as well as to reduce the multipath induced performance variation (the attenuation and reflectivity values for materials changes with frequency).
ABOUT THE AUTHORS
Peter Fuhr, Ph.D. is chief technology officer at Apprion in Moffett Field, Calif. Contact him at email@example.com. Nacer Hedroug , P.E., Ph.D., is a GPCE-Automation Project Manager at Eli Lilly in Indianapolis, Ind. Contact him at HEDROUG_NACER_E@LILLY.COM.
Fuhr and Hedroug are also co-chairs of ISA100.21 (www.isa.org/isa100).