The ever-decreasing cost and increasing performance of silicon radio-frequency ICs (RFICs) make it practical to manufacture wireless communication capabilities into everyday goods. Many more applications become possible, however, when multiple devices are networked to communicate with each other and with the Internet.
With the addition of appropriate control software, simple RFICs can become store-and-forward nodes in device networks, giving rise to a new category of wireless embedded networking to add sensing and control to everyday devices.
Wireless embedded networks make it possible to embed low-cost, low-power links into common manufactured goods, enhancing those devices with bidirectional communication to each other and to larger networks. By using a mesh architecture with redundant paths, it is possible for developers to build highly reliable networks even in the presence of attenuation, interference and multipath effects.
In any network-wired or wireless-the reliability of communication between any pair of nodes is degraded by three phenomena: attenuation, interference and multipath. In wired networks, the degradation is generally so minor that errors are extremely rare. By contrast, in wireless networks, the data travels over an uncontrolled and unpredictable medium-the airwaves-and signal degradation becomes an important consideration in the design of the network.
Attenuation: Radio waves weaken with distance. In free space, this attenuation follows an inverse-square law-doubling the distance results in a factor of four reduction in amplitude-but in physical spaces with lots of clutter, path loss exponents of three or four are common. Radio waves are further attenuated when passing through solid materials such as doors or walls.
Interference: Interference can come from intentional radiators, such as other radio transmitters, or from unintentional radiators, such as microwave ovens. Spread-spectrum techniques, whether frequency hopping or direct sequence, offer some degree of immunity from interferers.
Multipath: When a transmitted radio signal arrives at a receiver by means of two or more paths (for example, due to reflections off nearby objects), the multiple signals combine at the receiver's antenna. The wavelength for the 2.4-GHz industrial, scientific and medical band is approximately 4.8 inches, so when direct and reflected paths differ by multiples of 4.8 inches plus 2.4 inches, the signals cancel one another. Nearly everyone has experienced this in a car while listening to an FM broadcast: when you stop at an intersection and the radio signal becomes fuzzy, you can often creep ahead a few inches to restore the signal. This is a classic example of multipath interference.
In many RF communication systems, it is common practice to increase the transmit power in an attempt to increase reliability. Increasing the transmit power will help overcome attenuation and interference, but it does not generally help overcome the effects of multipath: increasing the transmit power increases the amplitude of the direct signal and all of the delayed components equally. Furthermore, in the case of dense wireless networks, one radio's signal is another radio's interference, so increasing the transmit power of any one transmitter has the undesirable effect of causing more congestion.
Not your parents' network
Wireless embedded networks have several characteristics that set them apart from conventional wireless local-area networks, such as IEEE 802.11. A node in a wireless embedded network has these attributes:
Low power: Many applications must run unattended for years using battery power. Commercially available wireless embedded network nodes consume about 50 milliwatts when active and less than a microwatt in sleep mode. With a battery source of three watt-hours, such systems will last 2.5 days if active 100 percent of the time, but with a duty cycle of 0.1 percent, the same system will last over six years without a battery change.
Relatively low data rates: Your average light switch communicates approximately four bits per day, or 46 microbaud. In many applications, high data rates are not a requirement.
The IEEE 802.15.4 specification defines a communication rate of 250 kbits per second, which is more than adequate for low-rate sensing and control applications.
Low costs: Wireless embedded networks are engineered to be built into a broad spectrum of products, many of which are cost-sensitive. It is now possible to purchase RFICs for less than $3, making them suitable for many commercial and consumer applications.
Naturally, Moore's Law holds for RFICs as well as for other silicon devices, so the price of RFICs will continue to fall over time and thus will enable ever-broader applications.
Reliable, autonomous operation: Since there are typically many more nodes in a wireless embedded network than there are human attendants, nodes must be easy to deploy, operate reliably without human intervention and be self-maintaining. A failure of a single node should not interrupt the operation of the entire network.
Taken in toto, these attributes argue against a conventional star architecture, typical of 802.11 and Bluetooth networks, favoring instead a self-organizing mesh architecture (see figure).
The benefits gained by the use of a self-organizing, self-healing mesh architecture include:
Reduced transmitter power: Since nodes only need to communicate with their nearest neighbors rather than reach an access point in a single hop, transmitter power can be reduced. Assuming inverse-square path loss, halving the link distance reduces the required transmitter power by a factor of four.
Simplified deployment: In a star network, if a node is out of range from its access point, there is no recourse other than to move the node or the access point, and often even that's not an option. In a mesh network, extra nodes can be quickly deployed to serve as repeaters to fill in any holes in the network.
Increased reliability: In real-world environments, environmental factors can cause frequent interruptions to communication between any given pair of nodes. A mesh architecture offers multiple links among nodes, so messages automatically travel along any available link, greatly enhancing the reliability of the overall network.
Wireless mesh architectures offer significant protection against attenuation, interference and multipath effects. By placing receivers and transmitters closer together, both attenuation effects and interference effects are reduced. And since the physical path lengths between transmitters and receivers are reduced, the reflected paths are proportionally shorter, which decreases the destructive effects of multipath.
In addition to creating more reliable links among transmitters and receivers, a mesh network with redundant paths offers a significant boost to overall reliability.
Robert Poor (email@example.com) is founder and chief technology officer of Ember Corp. (Boston, Mass.).
See related chart