The Next Step
The potential market for wireless sensors is almost limitless. Everything from home automation to battlefield coordination could benefit from mature wireless sensor networks.
Michael R. Moore and Stephen F. Smith, Oak Ridge National Laboratory
IEEE 1451 is a family of standards that links sensors to users, similar to the way that IEEE 802 (Ethernet) provides connectivity for information systems. Currently, all working groups under the IEEE 1451 umbrella provide standard interfaces for sensors on tethered networks. But the demand for a wireless physical layer is growing.
A wireless IEEE 1451 standard should provide seamless connectivity among sensors and users, no matter what distance separates them. And it must do this without requiring the installation of new wires and with reasonable cost and size additions at each sensor node.
What Does Wireless 1451 Need to Accomplish?
The general requirements of any wireless network include throughput, range, reliability, and power consumption. Some users need tens of megabits per second while others require only a few bits per day. Also, the sizes of the networks vary from a few feet to several miles.
The Wireless Sensing Workshop at Sensors Expo in Chicago on June 4, 2001, tried to determine the interest and requirements for wireless interfaces for sensor-based networks, especially in the industrial community. An informal survey taken at the workshop showed that most industrial users needed 32 or fewer nodes per network and typically <300 bps per node, or an aggregate data rate of <10 kbps for the network. At the same time, the industrial users generally wanted network ranges of a few kilometers, not just tens of meters. While this sampling population was too small to be statistically accurate, the results showed that at least one of the physical layer options must offer long-range operation, but perhaps less throughput would be acceptable (see the sidebar, “Wireless Sensing Workshop in Philadelphia”).
Workshop panelists also discussed reliability. Some asserted that potential users of wireless sensors must sacrifice some reliability to go wireless; others wanted a network protocol that optimized the reliability of the link. One way or another, physical layers chosen for a wireless IEEE 1451 system must address data transmission reliability. Within this context, the concept of reliability has several possible definitions, depending on the application. These include the probability that a message will get through; the probability that a message will get through within a given amount of time; or the probability that errors in messages will be detected.
Requirements of Sensor Networks
The main purpose of a sensor network is to distribute sensor data. For the data to be verified, certified, and coordinated, the user must be able to access the sensor manufacturer’s calibration, correction, and device-identification data. Therefore, any viable sensor network must include mechanisms to identify each sensor and to access associated databases, whether embedded in the sensor (local TEDS) or stored in another location available to the user (virtual TEDS).
It could also be argued that synchronous sampling of sensors (or the determination of when samples were taken) is also a necessary component of a sensor-based network. In general, a sensor network must have a more tightly controlled, more deterministic time-base than random-accessed general data networks can typically provide. Any physical layer suitable for a IEEE 1451–type standard smart transducer network must incorporate a robust mechanism to synchronize various components of the system (e.g., the individual transducer-to-bus interface modules [TBIMs] and the system bus controller [TBC] in the proposed IEEE P1451.3 standard).
Time resolutions required for more precise applications may be in the microsecond range, although many scenarios will be far less demanding. Any new proposed standard will have to at least provide a means of achieving this higher resolution. Two approaches proposed for system synchronization in IEEE P1451.3 include a separate sinusoidal sync signal on the cable and the simultaneous provision for transporting formatted timestamp data from the TBC to the remote TBIMs over the interface. IEEE P1451.3 is designed to use cable as the physical link for communications.
One issue highlighted at the workshop was sensor node power drain. Many participants expressed the need to deploy large numbers of sensors that could be powered by inexpensive batteries for one to five years. Because most of these units would report their readings only a few times a day, the need for a power-efficient wireless transmission scheme coupled with low unit quiescent drain is paramount. The consensus of the group was that an appropriately low-power format still does not exist in the marketplace.
How Do We Develop a Wireless Interface Standard Quickly?
The medium access and physical layer definitions from the IEEE 802 family address some of the needs of a wireless IEEE 1451 network. Adopting parts of these definitions could greatly reduce the time required to develop an IEEE 1451 wireless standard. In addition, the general structure of the IEEE 802 family can also be used as a basis for the structure of the IEEE 1451 family. The IEEE 802 family encompasses several combinations of medium access and physical layers while maintaining consistent network management functions (see Figure 2). The overview, architecture, and management functions are similar to that of IEEE 1451.1.
Leveraging existing products broadens acceptance in the marketplace and has a greater impact on the market. If appropriate product lines can be identified, they should be considered as candidates for adoption. Some products are discussed in the last section of this article.
Like the IEEE 802 family, the IEEE 1451 family will have to accommodate a dynamic market by retaining portions of the network standard that are reusable and by encompassing new technologies, especially at the media access and physical layer levels. Whether one or more physical layers are adopted in the first implementation of a wireless 1451 standard, future technologies and bandwidth needs will require a standard that can grow without throwing out the infrastructure of the rest of the standard.
The most fundamental choice to be made in the deployment of wireless sensor networks is the selection of the frequency of transmission. The options include licensed bands in the VHF and UHF regions, all of which require expensive and time-consuming application procedures through the FCC; leased transmission facilities and/or services from independent providers, such as private industrial radios, shared trunked systems, œellular digital modems (e.g., charge-coupled photodiode or photoconductive decay units), or satellite-based systems (e.g., GOES and Argos); and unlicensed systems restricted to specific ISM and similar bands, including ones centered at
Although the first option offers straightforward use and relative freedom from interference, the overall costs and implementation delays are often prohibitive for wireless sensors unless the networks can be shared with other services, such as voice communications. The second option usually incurs high ongoing costs because of periodic provider charges, but it may be useful in applications in which there are widely spaced sensors in remote locations or at a few monitoring points. In general, these systýms are well suited for low-data-rate, low-read-rate applications. The third option affords the greatest implementation flexibility and will usually be the least expensive to operate, although a basic concern of ISM-band systems is the finite probability of experiencing interference from unlicensed and licensed users in those bands. As a result, spread-spectrum technology is generally needed to achieve satisfactory transmission reliability in most wireless sensing applications.
Coexistence with Other Networks/Interferers. As more and more wireless systems are fielded, coexistence becomes a bigger issue. It’s not legal to build a system that overrides or interferes with other licensed systems. In short, it should have minimal impact on the performance of other systems. ©enerally speaking, these requirements favor the use of spread-spectrum technology, although conventional narrowband RF transmissions may suffice for some applications.
FHSS vs. DSSS vs. OFDM. Table 1 compares a few of the physical layers in the IEEE 802 family, as well as those in a couple other standards.
The familiar frequency-hopping and direct-sequence spread-spectrum formats have been fully covered in earlier issues of Sensors, as well as in many other publications. OFDM has also been discussed, although not as extensively as the first two. Therefore, we’ll highlight only the issues of these three that bear on the planning of a wireless IEEE 1451 standard.
As shown in Figure 3, DSSS (on the left) produces a more noise-like spectrum, and FHSS (on the right) produces narrow-band peaks.
The image on the right can also be used to illustrate the spectrum of OFDM, which uses a group of n closely spaced (minimally spaced, orthogonal) carrier frequencies, each of which carries 1/n of the total bits in a message. In FHSS, the carrier hops to each frequency one at a time, while in OFDM, all carriers are used simultaneously. OFDM is not, strictly speaking, a spread-spectrum technique, although it’s sometimes referred to as such.
The better OFDM systems adapt to the channel by avoiding frequencies that exhibit high bit-error rates. Because this method uses all the frequencies (with minimum spacing) all the time, it seems to have the best spectral efficiency (bits/s/Hz) of the three methods, particularly when compared with typical implementations of the other two.
This is true in the case of adaptively modulated OFDM formats (like those used in ADSL wired links, the new IEEE 802.11a protocols, and several wireless versions proposed for the developing IEEE 802.16 Metropolitan Area Networking standard). Here, each OFDM carrier can be individually modulated via binary phase-shift keying (BPSK) or multilevel quadrature amplitude modulation (n-QAM) constellations (usually with n = 4, 16, 64, or 256), yielding even higher spectral efficiency when the link has a high SNR.
In general, direct sequence has the second best spectral efficiency, and frequency hopping is third on this list. A final overall selection consideration is that under current ISM regulations, the FCC doesn’t permit OFDM in the 915 MHz and 2.45 GHz bands, although a possible change in this aspect of the ISM-band rules is under consideration.
Table 2 shows generalizations of the three technologies. Although controversial, these observations are generally true (especially if adjacent-cell reuse is not considered).
OFDM is more optimal for a few nodes streaming lots of data, and DSSS is better suited for lots of nodes handling less data per node. FHSS is often preferred over DSSS by the military, mostly because DSSS requires careful management of the transmit power of individual nodes to overcome the near/far problem and FHSS does not.
The ranges are all approximately line of sight because they typically use carrier frequencies at VHF or above. However, their respective ability to adapt to the RF channel and interferers will determine their effective range.
The point of this table is to give general guidance. Tradeoffs between the technologies are application and product dependent.
The Next Step
Furthermore, the standard should allow the sensor network to maintain connectivity when nodes or links fail. This feature will be necessary to extend the range and reliability of networks. Examples of implementations that offer this type of connectivity include Scatternet topologies, as described by Bluetooth, and Mobile Ad Hoc Networking (see “Mobile Ad Hoc Networking” in the January 2001 issue of Sensors).
To show the feasibility of pushing forward with a wireless sensor standard, a few product lines have been identified in this article. The inclusion of these products shouldn’t be considered an endorsement. Rather, the availability of this hardware demonstrates that the standard could be immediately useful.
As shown in Photo 1, there are a wide variety of products available from various Bluetooth manufacturers.
These units can be used in a network containing one master and up to seven slaves in a configuration called a Scatternet. The slaved units can be placed in various lower power modes to optimize power consumption while retaining connectivity to the network.
Bluetooth uses FHSS and TDMA to accommodate several sensor nodes located close to each other. It transmits ~721 Kbps of data and was originally targeted at a range of ~20 ft. Higher RF power options have been added to increase the range. Frequency hopping is accomplished using 79 carriers in the 2.45 GHz ISM band. Also, the Bluetooth units use <100 mW in standby mode.
The popularity of Bluetooth can be used to the advantage of a developing standard, assuming it fits the requirements of users. It is already in the process of being adopted as a physical layer option by IEEE P802.15.
One example is the AX550. According to the company’s literature, the AX550 “is a credit-card sized (2.2 by 3.4 by 0.9 in.) transceiver and has four external digital inputs (5 VDC), one digital output (5 VDC, 15 mA sink, and 1 mA source max.), one open collector output (100 mA sink), and an RS-232 (5 V) or an optional RS-485 communication interface. The transceiver operates over a temperature range of –40°C to 85°C in the frequency band 902–928 MHz (selectable in eight 3 MHz steps). It has a sensitivity of –110 dBm (average power) at 19.2 Kbps. The transmitter has a power output of 100 mW, using BPSK spread modulation, OOK data modulation, and an effective data rate of 19.2 Kbps, consuming a little more than 1 W (165 mA; ~7 V).
The Axonn transmitter, using the same RF communication features as the AX550, can achieve a battery life performance that is limited by the battery shelf-life (7+ years using 1.3 AH, 2/3 A size) for transmissions every one minute. The Axonn devices operate over a line-of-sight range of up to 1 mi. Axonn currently has more than 5 million nodes in operation.
Components such as the AX550 can be used in the system architecture shown in Figure 5, which gives an overview of a generic wireless sensor network.
In general, these networks will include a heterogeneous mixture of components. They may include transmit-only units, receive-only units, transceivers, network gateways, and repeaters. There also exist point-to-point and multi-point-to-point network topologies. Physical layer options of a future wireless IEEE 1451 protocol should address all these types of components and topologies.
Michael R. Moore and Stephen F. Smith are RF Designers, Oak Ridge National Laboratory, PO Box 2008, MS 6006, Oak Ridge, TN 37831-6006; they can be reached at firstname.lastname@example.org and email@example.com, respectively.
Kang Lee is a group leader of Sensor Development and Application Group, the National Institute of Standards and Technology. He can be reached at firstname.lastname@example.org.