01 June 2004
Wireless versus wired networks
A notable problem with wireless devices is that they still need a power source.
By Dick Caro
Most generally consider Wi-Fi to be wireless Ethernet, but it is far more than that.
Wired networks, such as Ethernet, are for communications between fixed locations. Wireless networks, such as Wi-Fi, are for communications between devices. The distinction is lost for fixed-location devices, but device mobility is the primary benefit of wireless.
The air is free, but to operate wireless networks one still needs a wired connection to a computer or the wired network, a source of power, and radios. Estimating the cost of a wired network is easy. It is the sum of the cost of the network cable, junctions, and connecting wires; the cable and junction installation; the network interfaces; and the long-term maintenance of the installed wiring plant.
The costs of wireless networks are more difficult to estimate. They include the cost of wiring to access points, access point equipment, wireless interfaces, and long-term wireless troubleshooting and maintenance.
The other notable problem of wireless devices is they still need a power source. Wired network nodes can draw power from the local alternating-current receptacle, but mobile wireless devices depend on batteries or some alternative power source. Of course, you can always plug the wireless device into a local power source, but then you lose the mobility advantage and incur the cost of installing power connections at the device. To some extent, the recent Power over Ethernet (PoE) standard IEEE 802.3af came into being to help resolve this problem by transporting electrical power on the wired Ethernet network so it is available to wireless access points. It is still too early to see much acceptance for this standard, but it is likely to be popular once products for it penetrate the market. However, PoE still does not address the issue of powering the wireless device itself.
Signal loss and fading
In the early twenty-first century, wireless networks still suffer from mysterious dead spots—areas where there is no reception. They are mysterious because even very careful planning cannot remove all dead spots, and sometimes live spots just move or, in the language of radio, fade. The spontaneous loss of communications for no apparent reason is probably one of the most irritating aspects of wireless networks. Often, the signal mysteriously returns even before one can investigate the cause of its loss. This occurs with cellular telephones, with Wi-Fi devices, and with all other wireless local area network (LAN) technologies.
Fading can be because of interference from other radio signals present in the same part of the spectrum as well as because of moving equipment. Sometimes, a live spot exists only as a result of a multipath effect when the signal reflects from some stationary object. Wi-Fi may fade in areas where microwave kitchen appliances are in use or where a cordless telephone is operating at 2.4 gigahertz.
Dead spots may occur within buildings depending on their materials of construction. In the line of sight between the access point and the wireless device, each time the radio wave passes through a solid the signal attenuates. Denser materials attenuate more than less dense materials. Metals, particularly steel, used in building construction may absorb or attenuate most of a radio signal, creating a dead spot in its radio shadow. Moving the access point or the device by a small amount, perhaps only a few millimeters, may eliminate the dead spot.
Finally, there are sunspots! The sun emits a broad spectrum of electromagnetic waves at all frequencies, which generally constitutes noise. Once in a while, the surface of the sun experiences flares or dark spots that emit very strong electromagnetic waves that interfere with radio transmissions. These things simply do not occur with wired communications.
Radio waves move from an omnidirectional antenna in all directions. When these radio waves strike a very dense object such as metal or stone, they reflect, much as light reflects from a mirror or other shiny surface. Even when there is a clear path between the transmitting and receiving antennas, some of the signal reflected from other paths will arrive at the receiving antenna. This phenomenon is multipath distortion, and it can affect the received signal, because the longer path will cause the signal to arrive out of phase with the signal from the direct path. The effect of multipath distortion can range from nothing to the cancellation of the signal, depending on the paths and the resulting delays. In some cases, the multipath effect can even boost the received signal. This occurs when both paths arrive at the same time—in phase—such as when multiple transmitting antennas are used.
One of the problems of radio is that the spectrum is limited, and clever folks are constantly finding new uses for it. The attempt to allocate certain frequency bands for specific uses is the responsibility of governmental agencies—the Federal Communications Commission (FCC) in the U.S. The frequency assignment process is highly political and only loosely relates to the technology. Furthermore, frequency assignment is highly dynamic and sensitive to economic conditions and the appearance of new technology solutions. For example, the FCC originally assigned eighty-two 6 megahertz frequency channels exclusively for broadcast television—an enormous segment of the spectrum for a single purpose. In most areas of the U.S., only a tiny fraction of that spectrum is actually in use, because commercial television was reluctant to use the higher ultrahigh frequencies (UHF) because of their limited distance reception range. Some of the unused UHF television channels have moved along to serve other uses, and more will likely do so in the future. Needless to say, television stations are highly reluctant to change frequency channels once they are in use.
The U.S. military is one of the most demanding users of radio frequencies and is very reluctant to give up any frequency previously assigned to it. This same attitude is reflected in the military establishments in most other countries as well, even when the service using that frequency has been abandoned. Another demanding public sector is amateur radio, which has allocated to it small frequency bands scattered throughout the spectrum. Amateur radio broadcasters are also reluctant to abandon any frequency band.
Nevertheless, the U.S. and most other governments have ordered that all allocated users share the radio spectrum unless the service cannot function when shared. By definition, the military frequency bands cannot be shared. Public radio and television and global positioning satellite frequencies also cannot be shared. Certain public safety and many business uses are licensed and do not share. The remainder can be multiuse, and they delineate into both licensed and unlicensed frequency bands. Generally, licensed bands allow users to broadcast at higher power ratings to reach longer distances, while unlicensed bands have to limit radiated power to minimize interference between users.
Users of shared radio frequencies demand some type of access control so they can avoid interference. Fortunately, as the demands on radio bands have increased, so has the ability to economically use higher frequencies. Expansion to higher frequencies has enabled higher rates of information exchange. But this often results in messages of shorter length, and usually requires sacrificing range or distance between sender and receiver. Higher frequencies are usually limited to line of sight between transmitter and receiver. Most of the new methods for sharing radio frequencies have depended upon packet radio technology that is suitable only for digital data transmissions. In one such packet radio technology, wireless LAN, many users may share the same frequency through the use of spread spectrum technology. The Global System for Mobile Communication (GSM) is a wireless telephony technology that serves in most of the world. In the U.S. it shares a pair of frequency bands with both time division and frequency division multiplexing. Code Division Multiple Access (CDMA) is the wireless telephone technology of the future, and it depends upon packet-switching technology to share the bandwidth.
Loss of privacy
Once a radio broadcast enters the air, or ether, anyone may receive the signal. Wired communications require a physical electrical connection, or at least an inductive coupling that is very close to the wire so as to intercept the signal. Governments have declared that intercepting a wired communication signal is illegal and may only take place with a court order. No such limitations exist for radio signals. If you broadcast, anyone can receive. However, the law has made listening to some radio broadcasts illegal.
Solutions exist for making radio signals more private. Though no way exists to provide exactly the level of privacy of an ordinary wired communication, many methods are available for making radio transmissions difficult to interpret, even if we cannot make them impossible to receive. One of the most common ways to achieve privacy is to use highly directional radio antennas in which interception would only be possible if one had exact knowledge of and access to the line of sight between sending and receiving antennas. Locating these line-of-sight antennas on towers and rooftops physically limits the potential for interception.
Using encryption can make even an intercepted signal difficult or impossible to interpret, hopefully to the equivalent degree of wired communications. Encryption is the science of scrambling the data using a method and a key. Decryption is the method of using a key to unscramble the data to restore it to its original form. The interceptor would need the encryption key to unlock the data and decrypt it. Simple encryption is sufficient to protect noncritical or nonvital data, but more complex encryption is required for data exchanges that may involve personal or financial data.
There are two types of encryption: secret or private key and public/private key. Secret key encryption uses a key or cipher consisting of several characters to process the original message so as to create an encrypted message. The same key works to decrypt the message after its reception. Many processes or algorithms exist for secret key encryption. The best-known algorithm is the Data Encryption Standard (DES). The National Institute of Science and Technology (NIST) developed it, and it is widely published. DES uses a 56-bit secret key. To make it more secure, Triple-DES is sometimes used, in which the same key processes three times, though the key length is the same. Advanced Encryption Standard (AES) is the latest NIST development for assuring maximum security of the secret key method. It uses 128-, 192-, and 256-bit keys.
One of the most interesting methods is the public/private key encryption method, which verifies signatures. A user receives a public key. When the sender signs a document, the digital signature undergoes encryption with the sender's private key. The encrypted signature and the sender's public key transmit to the recipient, who then uses the sender's public key to verify the signature of the original user. Document privacy comes about by encrypting the whole document using the recipient's public key. When received, the targeted recipient, and only that recipient, may decrypt the document using his or her own private key. While complicated, no method provides greater assurance of privacy than public/private key encryption. For public/private key systems to work effectively there must be an open repository for public keys, such as www.keyserver.net/ en/ or http://pgp.dtype.org/, both of which only support PGP (Pretty Good Privacy) encryption keys. Clearly, there should be only a single key server, but this has not yet happened.
There are two dominant public/private key encryption methods: RSA (Rivest-Shamir-Adleman) and PGP. RSA is a product of RSA Security, a company that specializes in security issues. PGP is an open algorithm supported by software from PGP Corpora-tion. Both methods work, but PGP is more often the choice to encrypt an entire message. Secure socket layer (SSL), which uses RSA encryption, is the leading security protocol on the Internet. When an SSL session initiates, the server sends its public key to the browser. The browser then uses the public key to send a randomly generated secret key back to the server to have a secret key exchange for that session. The problem is that the public key infrastructure requires too much computational logic to implement easily on very simple handheld devices. The use of encryption is usually limited to verifying digital signatures and to financial transactions such as a credit card or bank account number.
Membership in a wired network happens by establishing a physical connection to the wiring or to a network element such as a wiring hub or switch. Wireless units are neither connected nor disconnected from a network. To communicate, they must first seek to join the wireless network. As part of the protocol for joining the network, a network address is necessary.
Network membership is actually a function of the routing capability, which embeds into the network by using an IEEE 802.1d protocol implemented by network switches. The algorithm is a spanning tree bridge. In it, the network switch learns the Ethernet address of each connected station when a message transmits from that station, because the from address is located in the message header. In this way, messages not intended for the network members for that switch do not clutter the network. For a station to join the switch's membership list, it must only send a message.
Roaming is an essential property of wireless networks, although the need for roaming exists anytime a portable computer works on different network segments. Any wireless device may move so as to be in the range of different wireless networks. The ability to roam means that applications may continue to perform their network communications as the device moves from one wireless network (domain) to another. Network membership transfers transparently from one domain to another.
For a wireless telephony network, roaming takes place transparently as cell phones move from the range of one cell tower to the next. It is not that simple for mobile computers on a wireless LAN, however. Usually, all of the wireless LAN access points connect to ports on a single network switch, which performs the routing function. However, this results in a clutter of messages executing from all access points in hopes of finding the targeted station.
The newest solution for roaming wireless LAN stations is the so-called wireless switch. This is an access point that has the ability to perform advanced 802.1d spanning tree bridging logic. Just like a wired switch, it learns that a station is within its range when the station transmits a message. The problem is that the station may have previously been in the range of a different access point. Recent advances to the IEEE 802.1d standard provide the network management capability to move the registration of a station from one switch to another. The wireless switch uses this capability to move the registration of a station from one access point/ wireless switch to another. Using wireless switches reduces the broadcast network clutter. W
Behind the byline
Dick Caro is an ISA Life Fellow. This article comes from his new book Wireless Networks for Industrial Automation, ISA Press, 2004. Write him at firstname.lastname@example.org.
Can't tell the router without a scorecard
The newest and most revolutionary form of network is a mesh network.
In a mesh network each station is both an end device and a network forwarding element. Mesh networks are naturally self-healing and redundant—exactly the property needed for industrial automation networks.
In a mesh network, each station is responsible for forwarding a network transmission not intended for itself to other stations within its radio range. Those stations, in turn, send the transmission to at least one other station within its radio range. Therefore, the network becomes very redundant, fault tolerant, and extended in range.
The drawback is that each station must remove redundant messages. In effect, each mesh network station becomes a network router. Additionally, because multiple paths are involved, each receiving station must reject duplicate messages received from divergent paths.
Standardized mesh network protocols also include the capability to build and maintain routing tables so as to provide clues for forwarding messages. This prevents messages from looping in directions other than their intended destination, which results in greater network efficiency.
Routing tables dynamically constructed as messages pass through each routing node of the mesh network. Because mesh networks that work in industrial automation tend to have 256 or fewer nodes, routing tables can be small and the routing simple. Routing tables need to update when new nodes appear in the mesh or if for any reason fail to respond to forwarded messages.
Mesh networks are not new. The Internet itself is a very large wired mesh network with very complex routing algorithms. Because IP addresses do not imply anything about location, messages routed on the Internet "hop" from one node to another that is (hopefully) closer to the desired destination. Internet routing algorithms are typically efficient enough that few messages need more than fifteen hops to reach their desired destination.
Wireless mesh networks pose a problem that one does not encounter with wired mesh networks such as the Internet. With them, there is no way, other than by using a highly directional antenna, to prevent a message transmitted by one node of a wireless mesh network from also transmitting to other nodes. This leads to multipath routing, or message duplication. Typically, the message identification field of the IP frame identifies duplicate messages, which may then discard that message. Multipath routing can improve network reliability by providing redundant message paths.