Considerations on Converting a Wired Product to Wireless

By: By Bruce Ulrich, Texas Instruments Inc.
( 1 Sep 2007 )

As products prove out their market potential, and customers require more functionality, wired communication links naturally tend to progress to wireless. Many products are not initially wireless, and many of the implementing engineers are not necessarily experienced in RF designs. As a result, the end product probably should be reengineered from the ground up to incorporate the necessary changes. However, this is not reasonable for most companies in terms of time-to-market or cost.

Therefore, let's consider some paradigm shifts in wired vs. wireless designs, and thoughts about transitioning a wired design to a wireless design. There are other benefits to implementing a wireless design in terms of flexibility, cost, and the ability to be reconfigured. If an engineer has the opportunity to architect a new system, it may be preferable to adapt a standard embedded systems network, such as ZigBee which incorporates all of the below considerations and more.

POWER
Wired links have a physical tether, and are most likely line-powered. Wireless nodes tend to enable freedom and can be both line- and battery-powered. A transceiver should be optimized for power considerations. Some systems already are designed to operate 99 percent of the time in sleep mode, which makes a transition to wireless easier.

There are three commonly implemented methods for reducing power:

• Periodic Sampling: periodically wake, transmit and/or receive, and then sleep

• Event Driven: wake on interrupt, process, and then sleep

• Store-and-Forward: can incorporate periodic sampling and/or event driven sampling for non-real time reporting and processing

For a transmit function, the node could wake up on an event interrupt (such as a key stroke) or at a pre-scheduled time, and follow a simple algorithm such as:

• Check status registers

• Handle Interrupt event

• Load data for DMA transfer

• Initiate transmit

• Reset registers and go back to sleep

Given the short awake/transmit time, average power consumption is minimal. Battery life depends upon the frequency of wake-ups, transmit current, and sleep current.

With a Store-and-Forward algorithm, a sensor or controller collects this data over time, and forwards it after a specified time or cumulative data. This type of algorithm reduces the high current transmit periods, and enables a much lower average current. For small data packets, a transceiver may be transmitting only for a small amount of time as compared to the wakeup, preamble, and synchronization given relatively high data rates. When this occurs, and the system can afford non-real time processing, the Store-and- Forward algorithm provides the most efficient power profile.

For the receive side, the power profile is not as optimal since events can be random. If the receiver is line-powered, then the receiver can stay on fulltime waiting for a transmitted event. Products such as the CC2500 and CC1100 from Texas Instruments Inc. (TI) incorporate a feature called WakeonRF which implements a periodic wake-up for power efficient reception.

Other power-saving strategies can be incorporated from a ground up implementation, but our focus is on converting an existing wired design to wireless.

SECURITY
Security in communication is a general term describing the privacy, validity and confidence of a transmission. The ISO has a document ISO/ITU 13594 that provides guidance on the implementation of security in the lower layers. As most links can be monitored, whether point-to-point or networked, security typically is addressed in most systems. However, in some point-to-point proprietary systems, security may be limited only to Layer 1 type security for verification of packet integrity.

Generally, point-to-point wired communications are a closed/controlled environment. As a result, they are inherently secure. Wireless communications are capable of being monitored, jammed, repeated, or duped. This risk suggests consideration for a security plan. Converting this simple point-to-point connection to a wireless link requires considerations for multiple layers of security since the remote transmitter may be a hacker acting as an external node, either malicious or benign. These security aspects are loosely categorized as:

• Integrity validation: i.e., cyclical redundancy check

• Authentication: valid signal from a known remote transmitter

• Spread spectrum/frequency agility: to avoid jamming and increase transmit power

• Encryption: for secure communication

• Rolling/variable encryption key

When implementing energy conscious designs, the remote transceiver may have limited computational/encryption ability. Some RF system-on-chip products, such as the CC1110 or CC2510 from TI, have embedded encryption engines which enable the remote transceiver to employ encryption using AES (Advanced Encryption Standard US FIPS 197). This enables a system to implement encryption without an added punitive cost or power effect on the remote nodes.

Some systems use frequency hopping as their method of security. Frequency hopping spread spectrum lowers the potential of jamming, enables a higher power transmission, and provides some protection against monitoring. However, this does not prevent a malicious hacker from capturing a transmission and repeating it. If the transmission was a remote trigger for access control, then a hacker can retransmit a known-good packet and gain access, without ever decoding the packet. Thus, an encryption system and a rolling key code are necessary for adequate protection. Note that an encrypted signal alone still can be captured and retransmitted, if the key is not dynamically changing.

A typical wireless protocol has a throughput of approximately 40 percent to 95 percent (of the packet size), depending upon the size of the data portion of the packet and the size of the overhead. Proprietary embedded industrial protocols usually try to limit the overhead to 5 bytes to 15 bytes (excluding the preamble) for efficiency.

As an example, a typical packet for a point-to-point RF link comprises the following elements:

• Preamble: Allows the RF receiver to recognize when data is arriving, as compared to the noise floor

• Start/Synch: A sequence for the receiver to know when the data packet starts. Enables the receiver to sync the clock between the TX and RX

• Length: Data that indicates the length of the packet

• Time Stamp: This feature is optional. In data acquisition systems, enables the receiver to organize received data. In continuous real-time sampling systems, it can be replaced with sequence in case retries are necessary for lost packets

• Key: A rolling key in case the transmitter is changing the encryption key

• Time Sync: Some systems use a time sync to identify the next communication so the receiver can stay asleep until the next transmission

• Command: Data indicating the reason for the transmission

• Data: Data communicating across link

• CRC: Cyclic Redundancy Check

REGULATION/CERTIFICATION
RF systems are subject to governmental regulation and certification requirements. For example, frequency of operation, output power, channel bandwidth, and channel spacing often are particular to the region of operation as specified by the governmental agencies. Since each region specifies these parameters based upon their local constraints, these tend to vary by region and the designer must consider these in each design. An application note addresses these specifications for the 2.4GHz band.

In addition to specifications for operation, many regions also require testing, certification, and registration to ensure compliance with regional regulations. Note that the 2.4GHz band is very popular, and it is possible to design this system to meet most worldwide requirements.

BANDWIDTH, OUTPUT POWER, PATH LOSS AND DISTANCE
Throughput expectations in wired and wireless systems are based on different parameters. In a wired system, the throughput is heavily impacted by the capacitive and resistive effects of the line; thus the longer the line, the lower the throughput potential. Wired networked systems can experience retries, or delayed transmissions, due to collisions during high traffic periods.

Other parameters that can impact throughput are environmental noise, type of transmission (differential/single-ended), simplex/duplex, etc. However, once a system is established, characteristically it is stable unless one of the environmental elements changes significantly. You can also increase bandwidth by adding multiple communications lines in parallel, effectively multiplying throughput.

For wireless systems, throughput is dependant upon multiple parameters such as carrier frequency, channel spacing, frequency deviation, type of modulation, transmitted power, range, and the list continues. Environmental elements such as noise or crowded spectrum also can cause numerous retries and diminish throughput. To increase throughput, the existing design may not be expandable. A new design will likely have to be started, including considerations for channel allocation and spacing, transmission frequency, output power, repeaters, focused antennas, and so on. It is not straight forward to increase throughput since a second RF link (at the same frequency) will cause interference and may provide only a marginal improvement in throughput.

Wired systems can have bandwidths in tens of Mbps. Industrial wireless systems usually have lower bandwidths, i.e., a few Mbps down to a few kbps. Both of these types of systems can be point-to-point, or point-to-multipoint links. To understand the impact of frequency selection on range, start with the Friis equation. The power at the receiver's input is given by:

PRx = PTx {(GTx • GRx • λ²)/(16π² • d² • L)}

where:
GTx = transmitter antenna gain
GRx = receiver antenna gain
λ = wavelength (same units as d)
d = distance separating Tx and Rx antennas
L = system loss factor (≥1)

This sets transmit power and antenna gains to unity (since the objective is to correlate range and frequency for a given output and antenna).

Converting to dBm we can simplify to:
PRx (dBm) = 20log (4π/3) + 20log d(m) + 20log f(MHz)

This equation approximates the power loss due to transmission, depending upon the frequency and range. Plotting this equation, for key transmission frequencies, against range, yields.

Note that as the equation would suggest, Free Space Losses increase with frequency at approximately 6dB per doubling of frequency. This graph also enables another useful observation—the range is approximately doubled for a 6dB output difference. Combining these two observations suggests the same system and output power can double the range by halving the frequency.

As an example, note that 433MHz at 100m experiences the same Free Space Losses at 868MHz at half the distance.

Assuming this to be true, why would anyone implement a higher frequency solution when a lower frequency has a larger effective range for the same output power? This is because regulations often limit the transmit power, unless some spread spectrum solution is implemented. Thus, even though 2.4GHz has a 40dB greater loss than 27MHz, 27MHz is limited to 10mW (or less) in most regions (without spread spectrum). Whereas, 2.4GHz can increase the output power to 1W using spread spectrum techniques. The available channel bandwidth at 27MHz makes spread spectrum impractical.

Another consideration for choosing a higher frequency also may be the required channel bandwidth for higher data rates. An additional consideration on frequency selection is the size of the antenna. Since the antenna size is related to the wavelength, which is inversely proportional to the carrier frequency, a 27MHz antenna will be much larger than a 2.4GHz antenna, or the receiver will suffer great losses.

We also can demonstrate the impact of data rate upon the receive sensitivity, and effectively the range. The receive sensitivity is given by:

Receive Sensitivity = N_THERMAL + NF_S + 10log(BW) + SNR_MIN

where N_THERMAL is thermal noise; NF_S is the system receiver's noise factor; BW is the receiver's noise bandwidth; and SNR_MIN is the minimum signal-to-noise ratio (SNR) of the receiver's demodulator to detect the signal with a given bit-error-rate (SNR_MIN is dependent on the modulation format).

The receive sensitivity increases (meaning: becomes less sensitive) with increasing BW. For most low power and commonly used modulation formats (e.g., 2-FSK) in commercially available radio transceivers, an increase in data rate requires an increase in the bandwidth of the transmitted signal. Thereby, requiring at least a similar increase in the receiver's noise bandwidth (e.g., for 2-FSK).

A 4X increase in data rate, while keeping everything the same, typically translates to a 6dB required increase in receive sensitivity:

10log (4X) = 10log (4) + 10log X = 6.02 + 10log X (in dBm)

Since, as we saw earlier, an output power increase of 6dB approximately doubles the range, then lowering the data rate (for, e.g., 2-FSK) by 4X also doubles the range, provided the receiver's noise bandwidth can be reduced accordingly. Other factors such as the type of modulation, receiver/demodulator architecture (i.e., implementation specific effects), and type of spread spectrum all can impact the rate of change that data rate has on range. However, unless highly advanced demodulation techniques are used, the general rule is that an increase in data rate shortens the range.

ADDITIONAL CONSIDERATIONS
There are many other items to consider when implementing a wireless design, such as noise and interference, type and orientation of antennas, proximity to RF shielding objects, channel crowding, and so on. Assuming the system has an adequate link budget for the range and data rate, the environmental elements described could impact the packet error rate, forcing retries. Thus, the environment should be understood in order to estimate the required data rate to account for retries.

CONCLUSIONS AND IDEAS
Wireless communications has many considerations that need to be factored in over wired communications. Initially, these may seem challenging. But they are well understood and accounted for by experienced RF engineers. However, the transition to wireless may also incorporate new considerations for power and security, too. Fortunately, these additional concerns are well-modeled and can be easily incorporated.

A customer's transition from wired to wireless can be eased by initially embedding an RF transceiver modem as a daughter card. This affords an opportunity to understand the dynamics of the implementation, while not missing the market opportunity. There also are tremendous benefits to adapting an industry standard, such as ZigBee since it has proven networking functionality and system optimization.

REFERENCES
[1] A New Virtual Backbone for Wireless Ad-Hoc Sensor Networks with Connected Dominating Set, Reza Azarderakhsh, Amir H. Jahangir and Manijeh Keshtgary, The Third Annual Conference on Wireless on Demand Network Systems and Services, Jan 2006.

[2] Federal Information Standards Publication 197, Advanced Encryption Standard, Nov. 26, 2001.

[3] Texas Instruments, Low Power Wireless, Application Note AN032 SRD regulations for license-free transceiver operation in the 2.4GHz band, Morten Engjom.

[4] S. Rhee, D. Seetharam, S. Liu, "Techniques for Minimizing Power Consumption in Low Data-Rate Wireless Sensor Networks," IEEE Wireless Communications and Networking Conference (WCNC), March 2004.


About the Author
Bruce Ulrich is the manager of the Low-Power Wireless Third Party program at Texas Instruments. He has held positions in design, program management, sales and marketing over that last 20 years. He has a BSEE and MBA.
Click here for Illustrations:

Figure 1

Figure 2