wireless machine-to-machine interfaces represent the third wave of computers. The first wave was business computers, expensive mainframe and supercomputers affordable only to the largest businesses. This phase appeared and grew during the 1960s and 1970s. The second wave of computers started in 1981 when IBM introduced the PC. This period flourished in the 1980s and 1990s. The new millennium has given rise to the third wave of computers, during which decreases in cost and advances in technology allow a machine as prosaic as a toaster to have an embedded processor as well as a wireless radio. These processors provide utility when they stand alone but add even more value when they can communicate among themselves without human intervention. With the advances in small, cheap computers, advances in wireless technology have also appeared.
The advent of wireless-cell-phone techniques has spearheaded these advances. These techniques include wireless networks for mobile phones and burgeoning Wi-Fi-hot-spot phenomena (Figure 1). Although M2M (machine to machine) is the current marketing buzz word, the precedents for wireless-M2M technology arose decades ago. One term that has fallen out of favour is “telemetry.”
The early space program used radio telemetry for sending data from a spacecraft back to NASA and control signals to the spacecraft—all without human intervention. The availability of military-frequency allocations eased NASA’s task.
High power levels ensured reliable communications. More recently, designers have applied the concept of space telemetry to more worldly vehicles, such as Formula 1 race cars. On-board computers can send data from the car to a trackside computer that then automatically adjusts the air-to-fuel ratio and other parameters to achieve the best performance.
Other wireless-networking applications include vending machines with cell phones inside that “call home” when the machine runs low or when a need for service arises. In the automotive world, the term “telematics” has replaced the old-fashioned word “telemetry.”
Telematics covers entertainment, navigation, and emergency uses. General Motors has pioneered this concept with its OnStar service. OnStar includes a powerful cell phone with a car’s entertainment and navigation systems. It can download real-time traffic reports, and it allows users to report emergencies or request assistance.
Because the cell phone receives its power from the car battery and because the antenna need not fit into a tiny handheld phone, the ability of the system to connect almost always exceeds the ability of a handheld cell phone to connect.
Because M2M wireless networks represent convergence among several emerging technologies, such as spread-spectrum wireless, embedded processors, and network-routing protocols, an abundance of hype surrounds the market. The hype touts the fact that a wireless network will allow communication between a light switch and a refrigerator.
However, that idea is the result of a grand vision that drives M2M. Internet pioneer Tim Berners-Lee states: “Machines become capable of analysing all the data on the Web—the content, links, and transactions between people and computers. A ‘Semantic Web,’ which should make this possible, has yet to emerge, but when it does, the day-to-day mechanisms of trade, bureaucracy, and our daily lives will be handled by machines talking to machines, leaving humans to provide the inspiration and intuition” (Reference 1). The scope and perception of this comment cement Berners-Lee’s reputation as a genius and big-picture thinker. The problem comes from the fact that no one knows what the killer application will be for wireless M2M networks. A lot of unsolved issues remain between the dream of wirelessly connected machines and the engineering to achieve this goal.
The futurists and pundits envision an M2M network with machinery that connects to the Internet. The problem with that scenario is that it requires the embedded system in the machine to have not only a radio, but also the protocol stack and hardware for TCP/IP (Transmission Control Protocol/Internet Protocol).
This situation brings up the problem of assigning IP addresses for all these gismos and then providing DNS (domain-name server) or some other means to find and use these devices. Panasonic builds its network cameras with a hard-IP address to a server that the company operates.
When you buy the camera, you can navigate to that site with your browser, and the server can then establish the routing between your camera and your browser. This approach is a bit clumsy at best. Some researchers have proposed just randomly assigning an IP address to every piece of equipment (Reference 1). They point out that IPv6 (IP Version 6) provides for 2128 addresses, enough to put 6.6×1023 devices on every square meter of the Earth.
A large swath of wireless M2M networks will exist on the Web as subnets, often without routing or IPs. To get data from the Web to this subnet, you need to install a router and a gateway. All these realities conspire against the hype that wireless-M2M-network nodes will cost less than US$2 each and will all connect to the Web.
Full-blown computers in their own right, routers and gateways will offset the low costs of any network node. Only a few years ago, people thought that Bluetooth was going to remove every cable from your car, desk, and benchtop. But, in reality, Bluetooth serves as a wireless-headset enabler with a range of two feet. The realities of wireless networking include the large cost of writing and standardising the high-level protocols for these devices to find and connect to one another. Once engineers achieved that goal, they realised that these devices needed security. Any wireless M2M system that claims to be ad hoc or self-arranging must address all theses issues.
These delightful laboratory curiosities are less useful in a world of teenagers intent on vandalising your data just for the sheer, destructive fun of it all.
Let’s examine what M2M wireless networks are and are not. M2M wireless devices currently use either the older cell-phone or the burgeoning ISM (industrial/scientific/medical) network, which uses the 800-MHz, 900-MHz, and 2.4-GHz bands.
In the near future, however, the WiMax (Worldwide Interoperability for Microwave Access) network, which runs licensed in the 10- to 20-GHz range and unlicensed in the 2- to 11-GHz range, will dominate. Both cell-phone companies and computer giants, such as Intel, are also looking longingly at the analogue-TV bandwidth of 50 to 200 MHz. The low frequencies of these bands allow them to achieve longer ranges with less power; further, rain and fog do not affect their reception. Despite the surge of WiMax, it is still a technology of the future. Today, the two predominant wireless technologies are mobile phones and IEEE 802-style ISM.
Cell-phone networks have the advantage of long reach and pervasive deployment. Many field-application engineers had difficulties trying to find wireless IEEE 802 hot spots until their companies gave them cards that connect to the Web through the cell-phone system.
They can now check e-mail from almost anywhere. M2M networks that use the cell-phone network will enjoy those same benefits. As a result, mobile and remote applications, such as OnStar and trucking fleets that monitor vehicles’ positions by tying a GPS (global-positioning-system) receiver to a wireless module, have gravitated to cell-based connectivity. This feature helps fleet owners analyse routes and also has the Big Brother ability to check on drivers’ behaviour. Similarly, a bridge structure may have sensors that monitor stress, traffic, and degradation.
These sensors can connect to maintenance and highway-control computer systems and provide emergency alerts when an earthquake or an accident, for example, compromises the structure.
The downsides of these cell-phone-based systems are cost and power consumption. The cost of the wireless modules is declining rapidly due to the ubiquitous consumer cell phone, but the cost of using the network is still relatively high due to the telecom companies’ predatory pricing models that charge for connections or minutes when an M2M system often needs to send only a few bytes of data. The other M2M networks, IEEE 802 types, operate in the familiar ISM ranges of 800 MHz in Europe, 900 MHz in the United States, and 2.4 GHz worldwide (Figure 2).
In addition, proprietary networks can operate in other frequency bands, such as 434 MHz, which garage-door and keyless-entry remotes use, as well as in medical bands for more reliable communications. The most familiar standard for this type of network is ZigBee.
It uses standardised protocols to allow small, battery-powered devices to communicate. Some ZigBee proponents claim that batteries using the technology have lifetimes approaching 10 years, but a five- or even a two-year lifetime is more realistic.
The biggest problems with these networks are interference and battery-life issues. Because the 2.4-GHz band is unlicensed, there are no restrictions on how many transmitters can reside in any one area (Figure 3). Some proponents claim that several 802-style networks can coexist, but the success of the networks is also their failure.
If the world becomes rife with 2.4-GHz transmitters, the effective radius of communication will likely decrease to a few feet, and, even then, this technology can severely affect data rates.
EDN Senior Technical Editor Brian Dipert noted this phenomenon in testing a wireless-speaker system (Reference 2). The use of the wireless speakers causes his 802.11 wireless Wi-Fi (wireless-fidelity) LAN either to stop working or to achieve the connection at 50% data rates.
Despite concerns regarding interference, some successful M2M applications use these ISM wireless protocols. In the United States, Verifone’s POS (point-of-sale) terminals use Connect One’s iChip IP-controller chips so wireless LANs can connect to a credit-card company to authorise a purchase.
The benefit is the speed of transaction. It takes many seconds for an embedded modem to dial a phone number, connect, establish the communications, establish the encryption, and get the authorisation for a 16-digit credit-card number.
A wireless system can more quickly perform these tasks and needs no phone lines or Ethernet connections to the cash register.
Because they use a network connection, all the cash registers in a large store can simultaneously access the credit-card-authorisation server rather than wait for an open phone line.
This technology is a good fit in areas in which fast payment is a real benefit, such as fast-food counters and subway-ticket kiosks. In these scenarios, the cash registers all have IP addresses and all hook up to the Internet.
Having the devices directly on the Internet is not always necessary or advisable, however. ZigBee proponents are looking to connect tens, hundreds, or even thousands of sensors to a central node, or coordinator. You can install a gateway if you need to send data to or receive data from the Internet.
Although the ZigBee network is not a conventional subnet, it does use packet routing and other sophisticated techniques to route the data among peer devices and to central coordinators. Classic ZigBee applications are HVAC (heating/ventilation/air conditioning) and lighting control in buildings and data collection in factories or fields. One ingenious application uses ZigBee nodes embedded in the reflector bumps on roads (Figure 4).
These nodes can monitor and report parking-space usage in real time and allow collection of data to verify whether people are feeding the parking meters (Reference 3). Although some industry participants include RFID schemes as wireless M2M networks, others see the technologies as distinct markets.
To better understand the features and drawbacks of wireless M2M networks, remember that analogue-design principles apply in two critical areas: the actual radio communications of a network and high-level-system design. In this regard, you cannot combine all the claims of all the marketing people and expect your system to perform at that level.
Advances in high-speed CMOS may make a US$2 radio feasible, but that radio is a ZigBee-style 802.15.4 radio, not a radio that can use cell-phone networks. Furthermore, if you want the wireless device to be on the Internet, you must pay for a processor big enough to hold a TCP/IP stack and provide for a way to assign and route IP addresses.
In the same vein, vendors often talk about long battery life. Wireless devices in a mesh network pass data from end devices to the periphery of the mesh. That ability impacts the battery life of devices more central to the mesh. In addition, an ad hoc network must spend a considerable amount of resources identifying and incorporating new devices into the net.
If a device cannot route along an established mesh path, then it must negotiate and establish a new path. All this work uses up battery resources. Worse yet, battery use need not be uniform across the net, meaning that some devices will need battery replacement sooner than others. Or, more likely, users will discard partially discharged batteries because system-maintenance procedures will dictate replacing all the batteries in the mesh at periodic intervals based on the worse device’s battery consumption.
Further, ZigBee-network devices operating at 2.4-GHz worldwide bands can transmit data at 240 kbps, those using the 915-MHz US ISM band communicate at only 40 kbps, and those using the 868-MHz European ISM band communicate at only 20 kbps. So, although you may want to move your devices out of the crowded 2.4-GHz band, the slower data rates may cost you in shorter battery life.
Cell-phone wireless networks may give you “everywhere” connectivity, but they don’t provide “always connected in real time” connectivity. A reliable connection may use proprietary networks and frequencies, meaning that you cannot ride the low-cost coattails of the ZigBee-design protocol.
Smart, self-healing devices that form ad hoc networks may not be the least expensive. And as always, factors including interference, network topology, and device protocols have an adverse effect on battery life.
The bad news
Spread-spectrum techniques do not result in infinite available bandwidth. These techniques let transmitters share bandwidth, but each additional transmitter reduces the data rate of the other transmitters, their range, or both—that is, if all the transmitters use the same protocol.
The 2.4-GHz ISM frequency band provides a striking example of how interference can make all the devices on the band useless (Reference 4). The license-free ISM bands by design contain interference sources. The developers of the unlicensed, 2.4-GHz band established it because microwave-oven magnetron tubes operate at this frequency.
These ovens have a small but measurable impact on wireless interference. More troubling, inductive heating and molten sulphur lighting provide even more non-communications-related interference in this band. These interference sources are of concern, but permitted uses of the 2.4-GHz band are so numerous that connections are becoming unreliable in some areas because the regulatory agencies allow many protocols.
These protocols include FHSS (frequency-hopping spread spectrum), which the Bluetooth protocol employs. Actress and communication-technology inventor Hedy Lamarr invented frequency-hopping radios as she played along to a player piano (Figure 5 and Reference 5).
During World War II, she figured out that secret radio communications would benefit the war effort. She conceptualised that the receiver could hop along the same pattern as the transmitter did as it hopped to different frequencies, just as her fingers could hit the same keys that the player-piano roll was hitting. This realisation led to the idea that radios could communicate with each other while preventing eavesdropping.
The Bluetooth protocol divides the 83-MHz-wide, 2.4-GHz ISM band into 79 1-MHz slices. The Bluetooth devices then hop among 32 of these frequencies at a maximum rate of 1600 hops/sec. Two collocated Bluetooth devices could interfere with each other only 1/79 of the time.
When this situation happens, the high-level protocols request that the system retransmit the lost packets. If the Bluetooth device hops into the frequency of your ZigBee or Wi-Fi LAN, it will also interfere with those devices. Consumers’ insatiable need for bandwidth drove the 802.11b standards that provide 11-Mbps speeds. These systems use the DSSS (digital-sequence-spread-spectrum) technique, in which the radio uses 22 MHz of the 83-MHz to 2.4-GHz ISM band. A PRBS (pseudorandom-binary-sequence) phase modulates the frequency across the band.
Unlike FHSS, DSSS continually shifts rather than hops the discrete frequencies. Cell-phone implementations of DSSS allow multiple transmitters to operate on the same band. Unfortunately, the 11-bit Barker code that Wi-Fi LANs use provides insufficient code gain to allow CDMA (code-division multiple access), although high-level protocols implement CSMA (carrier-sense multiple access).
The transmitter senses when another transmitter is waiting until the channel is quiet before it can transmit. The 802.11b’s bandwidth allows only three and four devices, respectively, to operate at once in those countries that the FCC governs and that European standards govern.
If a maximum number of devices are operating, then interference will occur with Bluetooth, WirelessUSB, cordless phones, and ZigBee.
Wireless USB can be a wideband radio at 3-GHz and higher frequencies, but Cypress Semiconductor also has developed a 2.4-GHz WirelessUSB standard. Like Bluetooth, this standard divides the 2.4-GHz band into 79 1-MHz-wide bands, but Cypress uses DSSS rather than FHSS to modulate the signal.
The connection does not hop around the 79 bands but rather sticks to one band. The pertinent thing about this implementation is that it is frequency-agile—that is, if it cannot establish or maintain a good connection in one frequency, it jumps to a different one. WirelessUSB’s developers targeted it at replacing cables; it has the low data rates of HID (human-interface devices).
The 2.4-GHz ZigBee protocol divides the band into 16 3-MHz-wide channels spaced 5 MHz apart. It uses DSSS to modulate the signal, does not hop among the 16 channels, and does not provide for frequency agility. Cordless phones and baby monitors also use the 2.4-GHz ISM band. Cordless phones may use FHSS or DSSS. They generally divide the 2.4-GHz band into 10 to 20 channels. The phones are rarely agile, but many allow users to select an operating channel to avoid noise.
Figure 6 shows all of these radios and protocols in the 2.4-GHz band. If the spectrum were an ecosystem, you could look at the Wi-Fi wireless LANs like lions at the top of the food chain. They take up a chunk of bandwidth and, when busy, wipe out other traffic in that chunk.
Bluetooth devices are like insects flitting about their 79 1-MHz frequency bands. They hop around and pop up at indeterminate times, depending on who is walking by with a headset. If Bluetooth devices are insects, then ZigBee is like a groundhog that pops its head up to see whether spring is near.
It takes a wider part of the band but uses it infrequently. Because the groundhog is not agile, its hole is always in the same frequency. Cypress WirelessUSB is like a hyena—an agile hunter that is always prowling around looking for a clear frequency to operate on.
Once it finds that frequency, it stays there and can continuously transmit low-data-rate information. The biggest problem in this ecosystem is the cordless phone. Cordless phones are like tigers that can carve through everything. They transmit a powerful signal that drowns out all the other animals in the jungle.
For this reason, several Wi-Fi LAN manufacturers recommend that customers do not use cordless phones. The unlicensed, 2.4-GHz band is not unregulated, but the FCC dictates only power levels. The mixture of DSSS- and FHSS-modulation schemes may cause problems for both types of devices.
Two mitigating factors bring light to all this doom: locality and the infrequent transmission of some wireless devices. Even a weak Bluetooth transmitter on your belt will overpower a wireless LAN that is 20 yards away. Engineers at Dust Networks are working to overcome these two drawbacks.
Dust doesn’t strictly conform to the ZigBee standard because the company provides frequency agility; Dust’s devices jump to a different ZigBee frequency to get a clear channel. Texas Instruments has made similar efforts. The company in 2005 bought ZigBee pioneer Chipcon. TI’s new ZigBee transmitter has better sensitivity and selectivity than the ZigBee specification to extend radio range and reject any interference. Another approach is simply to use a less populated band. All ZigBee vendors’ devices can operate on 800- and 900-MHz bands instead of the crowded 2.4-GHz band. The lower frequency improves range, as well. Zarlink provides the ZL70101 implantable radio chip that uses the 400- to 405-MHz MICS (Medical Implant Communication Service) band.
The device provides 800-kbps data rates along with MAC (media-access control) that includes Reed-Solomon-encoding FEC (forward-error-correction) and CRC (cyclic-redundancy-check) error detection and retransmission to achieve a reliable data link.
One innovative company has found a market for building lighting-control products that are simpler than the mesh networks that ZigBee proponents envision. The Lightning Switch by PulseSwitch Systems uses a piezoelectric-powered transmitter to send a code to a 500W ac-line controller using the 434-MHz frequencies that key fobs and garage-door remotes use.
The transmitters never need batteries because users supply the energy when they toggle the switch. “Although certain garage-door openers and some remote-car-lock systems are assigned the same frequency by the FCC, there isn’t any chance of our transmitters opening someone’s garage door or unlocking someone’s car,” said Jeff Rogers, director of engineering at PulseSwitch.
“We use a patented ID-code system formatted to a certain pattern, which is different from that used by car locks and garage-door openers.” With more than 268 million codes available, he said, you could have the same number of transmitter-receiver pairs in the same room, and they would not interfere with each other or with other devices working on other ISM frequencies.
Imagine an application in which the DMV (Department of Motor Vehicles) would communicate to your car when it is time for an emissions inspection. The car would then—without human intervention—collect and relay back pollution-performance data over real road loads and drive cycles.
You would never need to get another emissions inspection. Nevertheless, look at the long adoption path of Bluetooth; it is now a regular part of most people’s lives. As EDN Executive Editor Ron Wilson points out, “You can recognise pioneers from the arrows in their backs.” The ricochet mobile wireless network was an early wireless-mesh network that failed.
The reality of M2M will be neither tragic failure nor wild success but somewhere in the analogue middle. And when someone invents that killer application, we will all smack our foreheads and ask, “Why didn’t I think of that?”
References
1. Gershenfeld, Neil, and Danny Cohen, “Internet Ø: Interdevice Internetworking,” The MIT Center for Bits and Atoms, Sun Microsystems, Aug 14, 2006.
2. Dipert, Brian, “Rocketfish: Spectrum Shark,” EDN, May 6, 2007.
3. Van Horn, John, “System Data Show Half of Meter Income Goes Uncollected,” Parking Today, pg 24, March 2007.
4. Burns, John, Richard Rudd, and Zoran Spasojevic, “Compatibility between radio communication & ISM systems in the 2.4 GHz frequency band,” Aegis Inc, June 24, 1999.
5. Markey, Hedy, and George Antheil, Secret Communications System, Patent 2,292,387, August 1942.