Unlicensed Wireless Data Communications, Part II: Specifying RF Parameters

Tim Cutler

After defining requirements, understanding the key parameters of a wireless data communications system is the next step in designing a successful wireless application.

Illustration by TAISHA PAYTON

The first part of this article discusses how to determine the right frequency band and protocol for a wireless data communications application. This part focuses on the three main parameters that define a radio-frequency (RF) data system: over-the-air data rate, transmit power, and receive sensitivity. Although range is an important criterion for any wireless application, it is not a parameter to be specified; range is the result of these three RF parameters and can be used to define them.

Range

If transmission range is insufficient, the application may simply not work or may require repeaters or additional access points. Range is also the most difficult of criteria to ensure; it is easier to predict outdoors than indoors due to the relatively larger amount of multipath observed in indoor environments. In simple terms, range is determined by transmit power, receive sensitivity, antenna gain, and transmission medium. The free-space loss (FSL) equation is typically used as an estimate for path loss for line-of-sight range. For the 2.4 GHz band, the FSL equation can be expressed as:

For example, a range of 1 km would have a path loss of 100 dB. Good radio practice allows 10 dB of fade margin, which allows for differences in FSL, multipath, and other non-ideal realities. For the 915 MHz band, the path loss can be estimated as:

For the 5.7 GHz band, the path loss can be estimated as:

To illustrate estimating range, consider an RF system operating at 2.4 GHz. The transceivers, which transmit at 18 dBm, have receive sensitivities of –93 dBm. They are fitted with 9-dBi antennas. The link budget would be as shown in the following equation:

Allowing a 10-dB fade margin yields 119 dB of usable link budget. Using the FSL equation and solving for range results in an estimated line-of-sight range of just less than 10 km.

Although a larger link budget, in general, yields a longer range in indoor environments, other factors impact indoor range. A very important factor is indoor location geometry. In indoor applications, multipath signals are often the only signals that reach the receiver. So the ability of the signal to bounce off obstructions to reach the entire area is needed to avoid shaded regions where there is simply no signal present.

To ensure reliable system operation in a variety of indoor locations, there is no substitute for a site survey to identify shaded areas and to determine the obtained range. These data should be used to plan RF equipment deployment.

Over-the-Air Data Rate

The over-the-air data rate is determined by data throughput requirements and protocol overhead. In RF communications, some transmissions will not be received correctly, requiring retransmission, and this must be factored into the speed calculation. In a well-designed and properly installed system, more than 90% of transmissions should succeed on first attempt. Allowing 10% retries provides a comfortable margin for retransmission, which should rarely take more than two attempts.

In point-to-point systems, the speed calculation is straightforward and can be expressed as:

where D_pl is the payload data in bits, D_ao is application overhead in bits per payload transmission, D_ro is radio overhead in bits per payload transmission, and rt is the retry percentage.

For example, assume a remote unit needs to send 1000 bytes of payload data in response to a 2-byte access point command every 75 milliseconds. The application uses a packet format consisting of a start byte, an address byte, a data-length byte, the payload data, and a checksum byte. The D_ao would be 4 bytes, or 32 bits. In this example, the D_ro is 80 bits per transmission. Because there are both access-point-to-remote and remote-to-access-point communications, the total amount of data for both transmissions must occur in 75 milliseconds. Therefore, the required data rate would be expressed as:

RF data rate	=	[((16 b + 32 b + 80 b) + (8000 b + 32 b +
		80 b)) x 1.1] / 0.075 sec
	=	9064 b / 0.075 sec
	=	120.853 Kb/sec

In multipoint applications, unless a polling scheme or a time-division multiple access (TDMA) scheme is used, the calculation is more complicated. In a polling scheme, no remote transmits without first being requested to do so by a master device. This process can be thought of as multiple point-to-point networks separated by time. However, the 75 milliseconds must be divided by the number of remotes. In a TDMA scheme, time is reserved for each remote to transmit, even if no data are present.

The receive sensitivity of a radio indicates the level of signal strength that must be present to correctly receive data at a specified bit-error rate. Note that receive sensitivity is different from signal-to-noise ratio (SNR). Although receive sensitivity can indicate radio quality, it is primarily a function of the over-the-air data rate. The theoretical receive sensitivity can be calculated using:

where N_t is the thermal noise floor, N_s is the system noise figure, BW is the symbol rate, and SNR_min is the minimum SNR required for a given bit-error rate.

The formula shows that, all things being equal, every doubling of the over-the-air data rate reduces receive sensitivity by 3 dB. This shows the tradeoff between data rate and range, because receive sensitivity is the largest component of the link budget.

Note the role of modulation technique in receive sensitivity. Various modulation techniques may be necessary to achieve data rates within specified bandwidth allocations, but the higher-order modulation techniques require higher SNRs. Offsetting this effect is the narrower bandwidth required for the same data rate at higher modulation compared with lower modulation. Table I shows various minimum receive sensitivities versus data rates for 802.11a devices. From the table, the trade-off between data rate and receive sensitivity is readily apparent.

Transmit power is another component of the link budget. The determination of transmit power is usually driven by regulatory and power-consumption considerations. For example, FCC allows up to 1 W of transmit power in the United States in the 2.4 GHz band, but the European Telecommunication Standards Institute (ETSI) allows just 100 mW effective isotropic radiated power (EIRP). Most other countries follow either the U.S. or ETSI rules for the 2.4 GHz band. Therefore, to create a product that can be used in this band worldwide, most manufacturers limit the transmit power to 100 mW, or 20 dBm. (Note that although FCC allows 6 dB of antenna gain for a 1-W transmitter, ETSI allows 20 dBm including the antenna gain, meaning the transmit power before the antenna must be no more than 20 dBm minus the antenna gain.)

When selecting a transmit power level, remember that even the most efficient power amplifiers are typically only slightly better than 50% efficient. Therefore, transmitting 100 mW requires a minimum of 200 mW of power—just for the power amplifier. Offsetting this power consumption is the duty cycle of transmitting. Except in very heavy traffic point-to-point links, any one radio will rarely transmit more than 25% of the time.

Another power-saving technique is to turn off the radio when there are no data to transmit, typically referred to as sleep mode. In applications in which remotes send data whenever they have them and there is little or no access by remote traffic, the remote host can sleep the radio until either data are ready or at a preset wake-up interval. Because it takes time to wake up the radio, an increase in data latency is likely.

With other components of the link budget determined, antennas can be selected (and possibly make up for any shortfall in link budget). Antenna selection is usually guided by antenna size and cost, followed by desired coverage area and gain. For example, it is impractical to use large antennas on handheld devices or to use directional antennas at central locations with remotes spread over 360° areas.

In indoor applications, directional antennas can provide better performance than omnidirectional antennas. This is not due to the gain increase typically associated with directional antennas, but rather to backside and off-axis rejection that can reduce multipath cancellation. Not all directional antennas (e.g., Yagi antennas) have much backside rejection.

Antenna gain results from focusing transmitted energy into a smaller cross-sectional area. Ideal radiators, called isotropic radiators, radiate energy in all directions from a point source at equal intensity. Limiting the radiated energy to a portion of this ideal sphere increases signal intensity in the focal area. This is referred to as the EIRP. Therefore, the smaller the coverage area, the higher the antenna gain. The more narrowly focused the coverage area, the more difficult it is to properly aim antennas over long ranges.

Regulatory rules are another consideration in antenna selection. In a 2.4 GHz ISM band multipoint application in the United States, transmit power is limited to 36 dBm EIRP. This can be obtained by adding a 6-dBi antenna to a 30-dBm transmitter or by using an 18-dBi antenna with an 18-dBm transmitter. Also in the United States, antennas used with ISM radios must be tested with the radios and submitted for FCC approval. In Europe, 2.4 GHz radios are limited to 20 dBm EIRP.

Modulation technique relates less to receive sensitivity than to fitting a given data rate into a specified bandwidth. The two primary modulation methods are frequency-shift key (FSK) and phase-shift key (PSK). Each technique has several levels of bits per symbol.

Binary FSK is perhaps the simplest technique. Data are transmitted by modulating the carrier frequency with one of two frequencies—one frequency represents a bit value of 1, and the other frequency represents a bit value of 0. This approach offers simple implementation and low cost, but with relatively low bits/hertz. Higher-order FSK techniques can increase bandwidth efficiency, but at the cost of wider occupied bandwidth or lower receive sensitivity. Some radio designs employ a four-level FSK, but few go to an eight-level FSK.

When higher data rates or bandwidth efficiencies are needed, most designers turn to PSK methods. The popular Quadrature PSK provides 2 bits per symbol. For 802.11b, radios employ multiple levels of PSK to achieve various data rates. Ranging from BPSK to CCK, 802.11b radios offer a range of data rates along with corresponding receive sensitivities. PSK techniques require digital signal processing (DSP) to decode the data, which has traditionally meant higher cost. There is currently very little additional cost associated with PSK due to the reduction in DSPs.

Receive sensitivity and data-rate performance should be the same in two radios with identical transmit power, but there are other considerations. RF energy, either in-band or out-of-band, is typically present. Without sufficient front-end filtering, a radio's performance can be substantially degraded by out-of-band noise. Because spread-spectrum radios do not use all the spectrum at any one time, in-band frequency selectivity should be considered. Such selectivity is often specified as adjacent-channel rejection and can be an indication of immunity to jamming.

In addition, although the SNRs of the various stages in a radio are reflected in the receive sensitivity, it is important to accurately measure the SNRs. Other radio performance factors to consider are baseband-to-RF conversion (and vice versa) and whether single- or double-conversion stages are used. Quantifying the effects of these factors is difficult, but each affects cost. And although that is not to say that a more expensive radio is better, you do get what you pay for.

Tim Cutler is vice president of sales and marketing for Cirronet Inc. (Norcross, GA). He can be reached at tcutler@cirronet.com.