01 July 2003
To deploy wireless Ethernet
In a typical installation, the cost of running the cable will exceed the costs of the rest of the equipment.
By Tim Cutler
Until recently, factory floor communications has been the domain of several standard buses: Modbus, Profibus, DeviceNet, and others. Early attempts to use Ethernet were unsuccessful due to the nondeterministic nature of Ethernet.
With the advent of higher speed Ethernet and the decreasing cost of Ethernet switches and hubs, the obstacles to using Ethernet have eroded.
The advantages of Ethernet are many:
- Ethernet components are commodity items and thus are low cost
- Ethernet performs as a transport medium regardless of the programmable logic controller communications protocol by encapsulating the Modbus or Profibus protocol within the TCP/IP packet
- the multidrop nature of Ethernet provides a straightforward expansion path
- the destination devices where the information is sent are already on Ethernet networks
However, Ethernet still shares one disadvantage with the other buses. It needs to run on a cable.
In a typical installation, the cost of running the cable exceeds the costs of the rest of the equipment. And when equipment must change locations, one incurs additional cabling expenses.
There is a solution to this problem—wireless Ethernet.
Wireless Ethernet provides the benefits of Ethernet without the need to run cables. Wireless Ethernet allows for fast, easy deployment and redeployment and can cover miles of range.
Wireless technologies in office and factory environments have a bit of a checkered past. Some wireless vendors have been guilty of overpromising and underdelivering on their wireless products. However, when properly understood and deployed, wireless devices provide reliable, robust communications.
Advantage licensed radios
Wireless technologies employed in factory and industrial applications are both licensed and unlicensed. Licensed bands offer a reserved set of frequencies without concerns about other users occupying the same band.
In 1987, the Federal Communications Commission (FCC) passed the spread spectrum rules allowing unlicensed operation in specific bands. Although no license is required, there is the possibility of interference from another user occupying the same band.
The spread spectrum rules came out to allow multiple users to simultaneously occupy the same band with limited interference. In indoor factory settings, there is not much concern about interference from other users, as the facility structure will provide reasonable isolation from outdoor users.
However, the use of multiple networks at a facility can pose problems unless one plans for this situation in advance.
Outdoor industrial facilities, particularly in supervisory control and data acquisition applications, have a long history of licensed wireless usage. However, since the spread spectrum rules passed, unlicensed radios have been making steady inroads into these applications.
An advantage that licensed radios have—because they do not need to share the band—is longer range due to higher transmit power. The disadvantage, of course, is the need to get permission to use the spectrum from a frequency coordinator.
The overwhelming majority of wireless Ethernet devices operate in unlicensed frequency bands. There are three main frequency bands of unlicensed spectrum that are suitable for reasonably sophisticated data transmission.
Best effort direct technique
Spread spectrum radio technology, as the name suggests, spreads the information signal over several frequencies. By so doing, interference on a single frequency does not block the signal.
There are two types of spread spectrum technology: direct sequence and frequency hopping. Recently, the FCC allowed the use of orthogonal frequency division multiplexing (OFDM) in the industrial, scientific, and medical (ISM) bands.
Direct sequence (DS) spreads the signal by multiplying the data stream by a pseudorandom noise signal of higher frequency than the data stream. This causes the resulting signal to spread over a bandwidth equal to the frequency of the pseudorandom noise signal.
The ratio of the spread signal to the unspread signal is the processing gain. Its units are decibel (dB), and it reflects the amount of signal impairment that can occur without a loss of information.
For the latest 802.11b systems, the information signal spreads over 22 megahertz (MHz) of bandwidth. Using some more complex modulation techniques, they are able to maintain 10 dB of processing gain.
By using the wider bandwidth, DS systems offer higher potential data rates, although at the expense of lower receive sensitivity. Because of this tradeoff, 802.11 systems use a best effort approach where they transmit data at the highest rate allowed by the environment.
For example, an 802.11b device will transmit at 11 megabits per second (Mbps) when it has sufficient signal strength to do so. If there is not sufficient signal strength, or if too many retransmissions are required, an 802.11 device will reduce the transmission rate to 5.5 Mbps, 2 Mbps, or even 1 Mbps.
Hopping in pseudorandom
Frequency hopping (FH), on the other hand, continuously varies the carrier frequency of the information signal. It transmits a small burst of data on one carrier frequency, then changes to another carrier frequency and transmits another burst.
By hopping to frequencies in a pseudorandom pattern, frequency-hopping systems have a high degree of jamming immunity as well as immunity to multipath fading. Until 2002, frequency-hopping systems in the U.S. were required to hop over at least 75 frequencies with a bandwidth of no more than 1 MHz.
Then the FCC issued new rules that require hopping over just 25 frequencies with 5 MHz of bandwidth. These new rules are consistent with European Telecommu-nications Standards Institute rules that govern European radio operation.
Because of the narrower bandwidth allowable, FH radios typically have lower data rates.
OFDM is a multichannel technique similar to frequency hopping except that it transmits over multiple channels or subcarriers simultaneously. All of the subcarriers are orthogonal, which means that each subcarrier has a null at the center frequency of all other subcarriers.
This avoids any intercarrier interference. The radio frequency (RF) spectrum of an OFDM signal looks similar to a direct-sequence spectrum, but the similarity ends there. The theory is that while one or multi-path fading or an interferer may affect more subcarriers, others will get through.
OFDM also uses training channels to characterize the RF paths. OFDM interleaves the data over multiple channels to improve the ability to reconstruct the data when subcarriers are impacted.
802.11a requires one to use forward error correction. OFDM will be similar in co-located networks to 802.11b due to the similar spectrum occupation.
Jamming interference immunity
To decide which technology is best suited for an application, it is helpful to understand the factors that affect radio transmission. In indoor applications, including factories, two primary factors are multipath fading and interference.
Multipath fading occurs when multiple copies of the signal arrive at a radio at the same time but with varying phase. This causes the multiple signals to cancel each other to some degree, which results in a "faded" or reduced strength signal.
Interference occurs when another RF source generates a signal at a frequency of interest that is of higher field strength than the intended signal. The interfering device does not have to be another radio.
In the 2.4 gigahertz (GHz) frequency band, microwave ovens and welding equipment can be sources of interference. While interference reduces throughput by requiring retransmissions of data, multipath reduces range.
In the past there was much debate over whether direct sequence or frequency hopping offered better immunity to jamming and interference. Currently, 802.11b is the predominant direct-sequence product used.
802.11b uses minimal spreading to limit the occupied bandwidth of its signal. The spreading employed is the minimum amount of spreading necessary to meet FCC rules.
As a result, experts agree that frequency-hopping technology provides superior jamming and interference immunity. This is especially true in rising noise floor environments such as factory and industrial settings.
Refer to wired Ethernet
802.11abg are wireless Ethernet standards that promote interoperability between vendors' office local area network (LAN) products with the goal being to foster competition resulting in lower costs.
The standards define not only the media access control (MAC) layer but also the physical (PHY) layer for 802.11 radios. The PHY layer in this case refers to the over-the-air protocol used by the radios. Although the goals of this standard are certainly worthy, the progress in achieving interoperability has unfortunately been limited.
802.3 is the standard that defines 10BaseT Ethernet (802.3u defines the 100BaseT standard). These refer to wired Ethernet. Wireless devices that are 802.11 compliant are actually 802.3 compliant in the connection to the network and 802.11 compliant in the over-the-air protocol.
The important point here is that while a wireless Ethernet device must be 802.3 compliant to connect to a wired Ethernet network, it does not need to be 802.11 compliant.
Factory data tens of kilobytes
Usually, more is better than less. In the case of wireless communications and over-the-air data rates, more is not necessarily better. This is because in wireless communications, higher data rates also mean a reduction in receive sensitivity.
Receive sensitivity refers to the signal strength necessary to receive data at a given bit error rate. For every doubling of the data rate, the receive sensitivity is reduced by 3 dB, other things being equal.
802.11-based products were originally for office LAN applications. As a result, they are trying to get as close as possible to wired Ethernet speeds—between 10 Mbps and 100 Mbps. This is important because typical office LANs transfer Windows-based application files between computers or between a computer and a printer.
This approach makes sense in an office environment where it is easy and inexpensive to place access points wherever need dictates. 802.11 systems vary the over-the-air data rate as conditions dictate.
In factory and industrial environments, due to the ranges and RF-noisy floors involved, it is common for 802.11 systems to drop down to the 1 Mbps data rate.
In factory and industrial environments, the amount of data transferred tends to be tens of kilobytes rather than the hundreds of kilobytes transferred on office LANs. However, the ability, ease, and expense of adding access points are very different.
These difficulties have given rise to the need for wireless Ethernet. Thus the data requirements must be crystal clear to avoid paying for unnecessary bandwidth.
Latencies 25–50 milliseconds
The first step in deploying wireless Ethernet is to determine how much data one needs to transmit and how quickly it must transmit. Also, decide the amount of latency that can incorporate into the process without adversely affecting it.
The first piece will determine the amount of throughput needed. The latency will determine how much data can transmit at one time. Throughput and latency are inversely proportional.
Regardless of how robust a radio link is, there will be times when a transmission is unsuccessful the first time and must transmit again. This retransmission is automatic and is transparent to the network, but it does introduce additional latency. Thus although a longer data transmission increases throughput, when retransmission is required, a longer data transmission increases the latency.
Because some data will have to transmit twice, latency will vary in wireless systems. This is not to say that latency has no bounds. It is reasonable to expect latency of 25 milliseconds to 50 milliseconds.
Thus, wireless Ethernet cannot work for applications where latencies of this magnitude are not acceptable.
One should determine the architecture of the wireless Ethernet network next. How many remote devices will connect wirelessly will figure in the decision of whether to operate one point-to-multipoint network, multiple point-to-point links, or some combination of the two.
This will impact the required throughput per device. If multiple links are a part of the network, each link will need less throughput. If multiple links are in a single location, the ability of the wireless devices to operate in the presence of other devices must be part of the planning procedure.
For example, typical 802.11b devices have just three nonoverlapping channels. This means that only three separate links can be operational in the same location. By contrast, frequency-hopping systems that support multiple hopping patterns can support a larger number of co-located networks.
The location of the wireless Ethernet devices deserves some attention as well. The location of the factory device will determine the general location of the remote wireless Ethernet device. The nearest point of the wired network will control the base wireless device.
As a rule of thumb, the antennas for all the devices should be placed as high as possible without placing them behind an obstruction. Although 2.4 GHz is a line-of-sight frequency band, indoors—particularly in factories—there are sufficient surfaces to reflect the signals to provide communications between two devices without line of sight.
Of particular importance in industrial settings where one needs a range of several miles is the ability to easily locate the antenna where line of sight can be obtained.
It is expensive and difficult to install long RF cable runs. In addition, RF cable attenuates the signal, reducing range or necessitating amplifiers. Products are available on the market that allow one to locate the radio near the antenna and away from the Ethernet connection. This provides line of sight without substantial cost or reduction in performance.
Another consideration in deploying wireless Ethernet is that the device must be suitable for the environment. For example, it is not practical to place a device designed for an office environment in a factory or industrial setting.
Along the same lines, what is the operating temperature range of the device? An industrial environment needs an industrial temperature range.
Ethernet, indeed, is ready and willing, and it is making its way onto the factory floor. W
Behind the byline
Tim Cutler holds two patents for microprocessor-based design. He is a vice president at Cirronet Incorporated in Norcross, Georgia. Write him at firstname.lastname@example.org.