A (very) partial survey of wireless technologies

 

Tom's Main Menu

Physical Computing Home

Intro to Physical Computing Syllabus

Networked Objects

Sustainable Practices

blog

Resources

code, circuits, & construction

my del.icio.us links

 

Wireless data is a big area. Really big. Big enough that this survey can't begin to cover it all. Instead, I'll touch on a few wireless technologies I'm acquainted with, to give a general sense of the wireless world, and a place to get started looking for more in-depth resources. This survey covers two applications of wireless technology that are applicable to this course: data transmission and location.

Short Range Data Transmission

One of the most common uses of wireless in physical interaction projects (and in general) is data transmission, i.e., getting information from one object to another. There a number of questions to ask when beginning a wireless data project that will help you determine the best approach:

How is the transmission structured:

  • Does the sender need to receive a reply from the receiver?
  • What happens if the receiver gets incomplete or garbled data?
  • How often does the data change?
  • How much data is being transmitted at any one time?
  • How fast does the data need to be received and processed?
  • Can the transmission be repeated?

How is the physical interaction structured:

  • How far apart are the sender and receiver?
  • Are the sender or receiver moving?
  • Is it critical that transmission be directed, or omnidirectional?

The techniques discussed here are for short range data transmission, i.e. between objects that are less than 50 meters apart. The two most common solutions to this problem are Radio Frequency transmission (RF) and infrared transmission (IR).

Radio Frequency (RF) transmission

RF transmission involves putting a small radio transmitter on the sending object and a radio receiver on the receiver. The receiver reads pulses of radio energy put out by the sender as data bits, and interprets them serially. If both objects will be talking and listening, a transceiver, capable of both, is used, and timing between talking and listening is managed by a microprocessor.

RF transmissions are omnidirectional, and can go through many nonmetallic objects, depending on how thick they are and how powerful the transceiver is. This is good in an environment where you know there are many objects, and you want to make sure that all objects within a particular radius of the sender get the message. However, if there are many senders and receivers within range of each other, RF can cause all kinds of confusion, as each receiver will be getting messages from several senders.

Infrared (IR) transmission

Infrared transmission involves putting an infrared light source, usually an LED, on the sender, and an infrared sensor, usually a phototransistor, on the sender. The phototransistor reads the pulses of light from the receiver as data bits, interpreting them serially. As with RF, if both objects need to both talk and listen, an IR transceiver is used.

IR signals cannot pass through objects that are opaque to light, so an infrared receiver must have a line of sight to the sender if it's going to get the message. For a situation where several objects are talking and listening in the same space, this can present a useful, if complex, advantage over RF. If the transceivers are placed on the objects such that two objects can only "see" each other when aligned correctly, it's possible to block out signals from other objects merely by blocking the line of sight. As with RF, if two objects need to both talk and listen, a microprocessor is commonly used to manage the timing of talking and listening.

Talking and listening using Time Division Multiplexing (TDMA)

You'll often see the term TDMA, or Time Division Multiplexing, used in conjunction with wireless systems. This means that the thing doing the talking or listening is set up to share time with other objects on the same frequency. If you're only transmitting a small amount of data and can space the updates apart in time, say, once every second, TDMA is an effective scheme. You program the transceiver to send your bytes once a second, and to listen for bytes the rest of the second. Unless there's a lot of data to be sent, this usually means that each object is actually listening most of the time, which is a good thing. If two objects are talking at the same time, neither one is listening, and you've got a data collision, meaning that my data collided with yours, and neither of us got the message. The whole point of TDMA is to allow us both to talk on the same frequency (or wavelength if we're using IR), and not have collisions.

Dealing with errors

There are numerous ways of dealing with errors in a wireless system. The simplest method would be to keep re-transmitting the data over and over, figuring that once in a certain number if times, the receiver will get it right. Slightly more complex is to use a handshaking scheme, in which the sender sends data, waits for the receiver to send back a confirmation message, then sends again. This latter method tens to provide stability, since the sender knows when the receiver got the message right, but it adds complexity to the system design. It means that the sender always has to wait for a reply, and it means that the receiver must be a sender too. In a handshaking scheme, some form of a checksum is used as well. A checksum is a method for checking the accuracy of a packet of data by checking the sum of its bits. In the simplest form of checksum, you add one bit to the end of the packet. If the sum of the other bits is even, you set this last bit to 1. Of the sum is odd, you set it to 0.

As you move up the spectrum of complexity in a wireless system, error checking gets more complex. Checksums may become more than one bit long, address header packets may be added, more complex handshaking and retransmit schemes are added, and so forth. Often, you can take advantage of such schemes without having to learn them in depth by using a pre-built module. For example, wireless ethernet radios exist that will allow the designer to take advantage of all the networking smarts of ethernet over a wireless link. Even most simple RF and IR transmitters take advantage of the error-checking protocols of RS-232 serial communication.

The more work your transmission unit does for you in terms of error-checking and networking, the more it will cost you in cash, power, size and flexibility. As a designer, you trade off these factors to get the best possible solution for your project.

 

  ASK, OOK, FSK???

You'll see these terms a lot when dealing with RF receivers. Basically, they're all kinds of modulating the signal to produce 0's and 1's. ASK is Amplitude Shift Keying, in which the signal amplitude is varied to produce data; OOK is On-Off Keying, meaning that literally, On = 1 and Off = 0. FSK is Frequency Shift Keying, meaning that the frequency of a transmitter is varied to produce 0's and 1's. Of the three, OOK is easiest for a manufacturer to implement, but the most error-prone. ASK is somewhat more work, but less error-prone. FSK is the least error prone, but takes the most bandwidth.This is relevant, because it means that data sent by a receiver that uses these methods is a digital signal, not an analog signal. If you plan to send a varying voltage via RF, you'll need to encode it as a digital signal first.

Location, Location, Location

Wireless data transmission means you're not tethered to a location, which means you can be mobile. If you've got a large number of moving objects in a system, often you want to keep track of where they are. For this reason, location technologies are a big business. Following is a survey of three location methods, Global Positioning Systems (GPS), E-911, Real-time Locating Systems (RTLS), and Radio Frequency Identification (RFID).

Here's an application example of serial RF transmission.

Robert Poor has a brief tutorial on sending little packets that includes links to several RF module manufacturers and some C code for the PIC.

 

For more on OOK, here's a decent engineering white paper.

 

Global Positioning Systems

There is plenty of information available on GPS and D-GPS available on the Web, so I won't repeat it here. In a nutshell:

GPS, or Global Positioning System, is made up of an array of 24 satellites owned and operated by the US military. Receivers on the ground track as many of the satellites as possible to triangulate and determine the receiver's position on earth.

GPS comes in two versions, Standard Positioning Service and Precise Positioning Service. The accuracy of the former is approximately 100 meters horizontally, 150 meters vertically. The accuracy of the latter is approximately 18 meters horizontally and 28 meters vertically.

Formerly, access to PPS was restricted to parties with US military clearance, and it was necessary to use D-GPS, or Differential GPS to calculate an accurate position. Under this scheme, an RF receiver was placed on the GPS receiver, and signals were broadcast from nearby radio stations with corrections to the errors programmed into the satellite signals that made SPS less accurate than PPS.

GPS is easy to deal with if you've got a receiver already. Most all receivers have a serial output. The data from that output will be in some version of the NMEA protocol for encoding location data. This includes things like latitude, longitude, course heading, and so forth. For more on it, see the NMEA FAQ. That serial data can be read directly by any computer or microprocessor that can take serial data in.

The main disadvantage to GPS as a locating technology is that it relies on satellite data, so the signal is very weak when it reaches the earth. Consequently, a very strong (i.e. power-sucking) receiver is needed, and a clear path to the satellite is needed. In places where the signal is blocked by mountains, thick forest cover, tall buildings, ceilings, and other obstructions, GPS is not useful.

GPs is also limited in its signal reception time. It takes between 30 seconds and 15 minutes to get a location fix, so for tracking objects that move faster than a turtle in real time, GPS is not too accurate.

 

For a good overview of GPS, Trimble has an excellent site, with lots of information on GPS technologies. For background on GPS communications protocols, the NMEA FAQ is indispensable.
 

E-911

The FCC in the US has mandated that Cellular service providers be able to locate any cell phone operating in their network with a 67% accuracy by 2002. This has launched the E-911 industry, for using cellular radio signals to triangulate location. The solution here is basic triangulation: a cell phone emits radio waves, which are received by one or more cellular antennas (three are needed for a proper triangulation). the distances to all receiving antennas is calculated based on signal strength and time delay between the reception at all receivers, and from this data, the position of the phone can be calculated.

The FCC implemented this to speed location of emergencies called in by cell phone, but the business has been profitable for companies tracking mobile assets such as trucks, taxicabs, escaped felons, and more.

 

Fall Creek Consultants have some decent data on E-911 on the web.
 

Real-time Locating Systems

Real-time Locating Systems use a system similar to GPs, but for smaller areas. In an RTLS setup, a number of antennas are spread over the area to be tracked. Objects to be tracked have a radio tag attached which emits a "chirp" of data at regular intervals. The chirp contains the tag's ID and any status data. The chirp is received by at least three antennas in the system and the tag's position is calculated from that data.

RTLS systems can generally read a tag within 200 ft., and can locate it down to 3 feet depending on the system parameters. The system's time accuracy depends on the number of tags involved. With a low number of tags, there are few collisions between chirps, and position updates are frequent. As the number of tags goes up, the number of collisions goes up, and tag location updates go down. In a fully saturated system (about 1000 tags per acre), location can take up to 30 seconds per tag. Given this, RTLS is good for applications such as vehicle location on parking lots, truck yards, air fields, and so forth, but not so good for locating people instantly in an unlimited area.

WhereNet has installed a system called ParkWatch in which visitors to an amusement park are given watches with RTLS tags so that, should a person get lost from the group he came with, he can go to a visitor station and locate the group immediately. Parents of small children are relieved, and teenagers are horrified by this system.

More information on RTLS can be found at The AIM website, and from manufacturers such as Pinpoint and WhereNet.
 

Radio Frequency ID Systems

RFID is a location system that uses passive tags (unpowered coils of wire with a capacitor) that reflect the incoming query signal from a locator beacon. RFID systems can usually read and locate a tag within a 20 ft. range. The tags are cheap, often used for inventory control. RFID is very good at identifying what tags are passing a given reader, but not very precise on locating the position of those tags, since they do not do triangulation or time-of-flight calculations on the returned signal.

More on RFID can be obtained from the AIM site mentioned above, and from the TIRIS site