1 February 2002

Wireless R&D Aims to Boost Traffic

By Michael R. Moore

Oak Ridge National Laboratory research targets sensor networks' bandwidth—and more.

Wireless sensor networks are potentially among the premier mechanisms to improve manufacturing industries' efficiency. By eliminating cabling, they significantly lower costs.

To advance their own wireless networks, manufacturer users constantly demand more bandwidth, and suppliers constantly try to provide it.

Bandwidth is the easiest variable to discuss. However, bandwidth by no means addresses all the issues of delivering information to the locations where it's needed. For instance, if delivering bandwidth for one task degrades bandwidth dedicated to another, you gain little or nothing.

The basic issue is this: As radio frequency (RF) technologies try to keep pace with the information demands, wireless infrastructure and communication systems have become interference limited, rather than noise or power limited. Therefore, the goal should be to achieve more cooperative use of bandwidth, rather than merely optimizing the range (sensitivity) of isolated systems.

More cooperative use of bandwidth has political, commercial and technical implications. On a political level, entities such as the Federal Communications Commission (FCC) have to continue tailoring spectrum usage to allow the maximum number of users per hertz. On the commercial side, companies can share bandwidth on a cooperative basis when it is mutually beneficial. Technologically, some advances reduce the amount of data needed to communicate intelligence.

Seeks Open Standard

At Oak Ridge National Laboratory (ORNL), research in sensor technologies aims to increase industry's ability to share spectrum with other users. Objectives include building systems to an open (nonproprietary) standard, developing architectures that allow systems to adapt over time, using modulation schemes that cause less interference, and building systems that are more robust against interference from other sources.

These technologies and standards not only address the issue of efficient use of spectrum but also target sensor networking. ORNL's sensor focus includes features not commonly built in on standard information technology (IT) networks: submillisecond synchronization, transducer electronic data sheet (TEDS, online sensor information), mechanisms, and calibration traceability.

Using a well-planned, tethered sensor network standard as a starting point for future wireless sensor networks, users will not have to force fit them onto protocols not meant to accommodate them in the first place.

Therefore, IEEE P1451.3 is a good starting point. It covers the necessary features of a sensor network and also coordinates all communications over a single transmission line, making it readily adaptable to wireless applications.

Most general-purpose wireless systems target the FCC-defined unlicensed portions of the spectrum. Therefore, we must be aware of the potential for overcrowding, especially in popular bands allocated around 915 megahertz and 2.45 gigahertz.

The challenge becomes one of not only building an RF system that will communicate a specified distance against a reasonable background of natural and man-made noise sources but also guaranteeing the system will deliver the specified bandwidth against a background of ever-increasing competing sources. While this reality comes from work in the unlicensed bands, the engineering principles apply across the whole spectrum.

The U.S. Department of Energy's Office of Industrial Technologies (DOE/OIT) has funded a project to develop technologies that will help industries meet these challenges. Researchers have examined and reviewed several technologies, protocols, and standards for industrial application.

Long-term solutions include using modulation methods that cause less interference but remain robust to other electromagnetic sources and employing robust standards widely used by various manufacturers.

Spread spectrum is one successful modulation method that gained popularity during the past few years. It has two main implementations: frequency-hopping spread spectrum (FHSS) and direct-sequence spread spectrum (DSSS).

In FHSS, for a fraction of a second the RF spectrum is identical to traditional narrowband methods. However, the carrier is "hopped" about every millisecond to another frequency. This makes the signal harder to block or intercept and provides some multiple access capabilities.

In DSSS, the carrier's energy is continuously "spread" by artificially increasing the clock rate of the data modulating the carrier. The spreading process uses a bit sequence having noiselike properties. This noiselike sequence multiplies the conventional data stream that modulates the carrier so the spectrum of the spread signal looks more like broadband noise. If properly applied, these two types of spread-spectrum transmissions provide a level of immunity to interfering sources.

DSSS may lead to more users than FHSS, if applied correctly.

Robert C. Dixon, in his book Spread Spectrum Systems with Commercial Applications (Third Ed., John Wiley & Sons, Inc., New York, 1994, pp. 367–376), gives a thorough comparison of these two methods: "Overall, the conclusion must be that while frequency hopping is generally superior for military systems, direct sequence methods will always outperform frequency hopping in commercial/consumer multiuser applications."

DSSS Tolerates Interference

DSSS implies the use of spread-spectrum codes. DSSS uses orthogonal pseudorandom noise sequences to widen the spectrum of an otherwise narrowband signal. By spreading the signal over a wider frequency spectrum, this method rejects narrowband interferers. Spreading the signal also reduces the probability that the signal of interest will interfere with other receivers by reducing the power spectral density. The correlation process of the receiver "despreads" the desired communication signal and effectively spreads out the energy from the interference signal.

DSSS-based systems provide the most common basis for CDMA systems. That is, several users can simultaneously use the same portion of the frequency spectrum. Thus, in systems where time coordination among all users is not possible, a mechanism gives all users simultaneous access to the network.

Because the spreading codes are noiselike, each user looks like a random noise source to other users. Therefore, there are practical limits to the number of simultaneous users, depending on the length of the codes and other parameters of the system.

In DSSS CDMA systems, power control is very important. The "near/far problem," as it is called, refers to the fact that one DSSS transmitter that is very near the receiver can dominate the correlator so that a more distant DSSS transmitter is unable to communicate. Therefore, most DSSS systems employ some form of relative power control. By one mechanism or another, the output power of the various transmitters scales such that the received power at the central controller (e.g., base station) is approximately the same from all transmitters.

While this obviously increases the complexity of DSSS systems, it actually provides the mechanism for several other benefits, such as increased battery life and, concurrently, reduced interference with other systems.

A standard developed for tethered sensor communications, IEEE P1451, is in the process of adding a wireless physical layer, which is not yet defined. It differs from the tethered version of the standard IEEE P1451.3, Smart Transducer Interface Standard for Sensors and Actuators. The main difference is that individual nodes in a wireless application are not powered from the controller as they are in the tethered version.

IEEE P1451.3's goal is to develop a transducer bus architecture that accomplishes the following on a single coaxial or twisted-pair cable: synchronizing data acquisition of a large array of transducers; communicating simultaneously with a large array of transducer interface nodes; and providing power for operating all transducers and associated electronics.

This transducer network allows scalability of architecture and a range of cost and capability options. It allows time division multiple access, frequency division multiple access (FDMA) and CDMA multiple access schemes. It may provide the option of combining two or more physical layers on a single coaxial cable.

Tested at Paper Plant

In the DOE/OIT project, the protocols and architecture of IEEE P1451.3 work in conjunction with a wireless DSSS physical layer. Three of these units built and demonstrated as a network in an operating paper plant successfully transmitted sensor data without causing interference or being hampered by interference. This is important because a previous wireless demonstration using narrowband technology caused an interruption of the plant process.

In follow-on work, chip sets are being developed that will enable these units to be size and cost effective. The application-specific integrated circuit (ASIC) development is targeting the 5–6 gigahertz band, where there are three 125-megahertz widebands available for unlicensed applications. When these custom ASICs are developed, they will be available for technology transfer to interested manufacturers.

Marrying a technology such as DSSS to a well-planned sensor network standard leads to systems that will provide industry with a large amount of total throughput.

Additionally, these systems will be inherently robust to interference and will reduce the chances of interference with other plant processes. Therefore, we should be looking for ways to promote standards, technologies, and policies that will lead to an efficient use of the available RF spectrum. IC