Connectivity by Rawn Shah

Wireless LANs finally make their way to standardization

Backed by IEEE 802.11 standardization, wireless LAN products are gaining acceptance. What is 802.11? How does it work? And what problems still need to be tackled?

SunWorld
May  1998
[Next story]
[Table of Contents]
[Search]
Sun's Site
Abstract
Now that vendors of wireless LAN products have the IEEE 802.11 to help provide interoperability between their products, the need for and importance of wireless LANs is growing. Rawn explains what the IEEE 802.11 standard covers and what must still be addressed. (2,300 words)


Mail this
article to
a friend

Wireless local area network (WLAN) products were once a specialty, available from only a few vendors and built according to their respective proprietary specifications. The Institute for Electrical and Electronic Engineers (IEEE) recently formalized a standard that will allow such products to interoperate. With wireless applications in the industrial and even intra-office mobility areas growing, this standard is the first step toward making wireless networking more acceptable.

Wireless LANs usually have a range of less than five hundred feet in open air or a few hundred in enclosed offices. They most commonly use the industrial, scientific, and manufacturing (ISM in North America) frequency band, between 2.4 gigahertz (GHz) and 2.4835 GHz. Although there have been a variety of proprietary solutions, the IEEE finalized in June 1997 an approved standard, 802.11, that organizes this technology. This standard actually approves three different physical transmission systems:

All three provide transmission speeds of 1 megabit per second (Mbps) and 2 Mbps, depending upon implementation and encoding system used. The first two use the ISM band, but in different ways. Frequency hopping breaks up the band into 79 separate channels, and individual nodes communicate randomly across these channels as they become available. This method allows you to deal with the problem of multiple overlapping wireless networks within the same area. It can handle up to 26 collocated networks and ensures a certain amount of fairness in communications. Direct sequence systems are simpler in design and require less power. The infrared physical system has not been fully defined but uses a diffuse band of frequencies in the infrared range between 850 and 950 nanometers.

Advertisements

How 802.11 works
IEEE 802.11 and Ethernet use the same data transmittal technique, carrier sense multiple access (CSMA), but 802.11 employs collision avoidance (CA) rather than the collision detection (CD) used by Ethernet. CSMA/CA is a complicated way of saying: Check if the medium is free (carrier sense) from other transmitting nodes (multiple access); if so, back off for a short while as specified by an algorithm (collision avoidance) and then send the message. The difference between this and the Ethernet CSMA/CD system is that the nodes back off before encountering a collision. Because air time is relatively more expensive, this can save on the number of collisions at the expense of slightly longer inherent delays.

There is one more step. In communicating between two nodes only (unicasting), when the sender sends a data packet, the receiver immediately responds with an acknowledgment (ACK) packet if the data packet is received intact. When a node is broadcasting to the entire network this ACK step isn't required.

Fragmentation of data is a fact of life in shared media. Even with the collision avoidance algorithms, there is still a chance for a collision between simultaneous access by multiple nodes. Furthermore, the wireless media is more likely to encounter external interference from other electromagnetic sources like TVs, microwaves, cell phones, power equipment, etc. If the ACK isn't received by the sender in a unicast situation, the sender considers its previous transmission to be fragmented and resends the message after backing off according to a pseudorandom algorithm.

The IEEE sees two configurations of how this can operate: The independent model, where a group of end nodes all communicate with each other directly in an ad hoc fashion; and the infrastructure model, where a base station access point handles the communications between nodes and an external distribution system (i.e., an Ethernet network).

The obvious difference between the two is the access point (AP). Without it, the independent model has a much smaller coverage area, as all nodes have to be within range of each other. With the AP, each node can be relatively further away from other nodes as long as it's within range of the AP.

A basic service set (BSS) is a group of nodes within one service coverage area. If there are several APs and coverage areas, this is known as an enhanced service set (ESS). The problems associated with the ESS are complex, especially with the issue of roaming. The ESS usually requires a secondary distribution system like an Ethernet to connect multiple APs. The traffic management between separate BSSs through this distribution system is outside the realm of the 802.11 standard; it may be routed, or it may be bridged. The standard does, however, define certain service calls that must be supported by the distribution system. This comes into effect in node addresses when the nodes are allowed to roam between BSSs.

Roaming is the ability to take a node from one service area to another without manually re-establishing the node's identity or network services. This wonderful feature of automation makes life complicated for the protocol engineer. The basic method is as follows: When the node finds that its link with the AP is getting poorer, it starts openly scanning for an AP with a better link. Upon finding a suitable AP, the node sends a reassociation request. If the new AP accepts, the node is reassigned to the new AP and BSS, the distribution system is notified, and the old AP is disassociated from that node.

Problems with the media
There are several common problems associated with wireless LAN environments. First, you have to deal with background noise and unauthorized or unwanted interference from other devices working in your frequency band. Second, you must take into account that multiple nodes may be competing for the same frequencies. Third, wireless networks are strictly regulated throughout most of the world by national and international organizations. Fourth, wireless networks are often used by mobile devices that have problems in terms of location-dependent link reliability and battery and power usage. And finally, wireless networks suffer from not having the security of any physical boundaries to delimit authorized members of a network.

Background noise comes from all kinds of sources, but the most prevalent form comes from the ordinary microwave oven. These ovens emit a range of microwave frequencies across the ISM band in alternating four-millisecond intervals. You can see this quite visibly if you have a television close to a microwave. In wireless LANs this kind of interference causes packet fragmentation and data corruption. There's really no predicting or avoiding this fact of life, which is why the fragmentation recovery mechanism is implemented.

In Ethernet, when multiple nodes try to use the media at the same time, they all receive the attempt (a collision) and back off according to an exponential algorithm. Basic Ethernet hubs do nothing more than connect the signals between the different wires. In wireless networks, nodes may not be able to hear each other, as they're out of the reach of other nodes, and thus may not know to back off. Wireless nodes only need to worry about being in reach of the AP. This is further complicated by the fact that the timing may differ according to the distance between nodes. This is solved by adding communication via proxy through the AP and delay or busy indicators to the connection process. To prevent fragmentation, the AP sends a net allocation vector (NAV) that describes how long to wait for an ACK to compensate for timing differences of a proxy packet.

Regulation of the frequency spectrum is maintained by the Federal Communications Commission in the U.S. Across national borders, the International Telecommunications Union, Radiotelecommunications (ITU-R) handles similar responsibility. According to the ITU-R, the world is organized by three regions for radio regulation. Region 1 centers around Europe but also contains the Middle East, Africa, and the Soviet Commonwealth. Region 2 consists of North and South America, Greenland, and the Pacific east of the international date line. Region 3 consists of the rest of Asia, Australia, and the Pacific west of the date line. The problem lies in the fact that the frequency bands in these regions were assigned at different times by different regulatory bodies. This means that services are not global, and even within regions, may not be transferable between countries. One thing that almost everyone can agree upon is in the 2.4- to 2.5-GHz band and smaller area. In all these areas, this band is available for data communications. Therefore, the IEEE working group focuses on this.

Idle nodes on a network may go into a sleep mode but first notify their AP. The AP stores the state of the nodes and buffers data packets intended for any sleeping nodes. The AP broadcasts a traffic indication map through a beacon indicating available packets for specific nodes. A sleeping node wakes periodically to see if a beacon is active. It then resynchronizes with the beacon before fully starting up and receiving the data. This kind of buffering is also done for all broadcast and multicast messages intended for groups of nodes.

Finally, the goal of 802.11 is to provide wired equivalent privacy (WEP). The optional system allows packet-level encryption using the internationally exportable RC4 algorithm that uses a 40-bit key. If the WEP flag in the packet header is set, the data is encrypted with the key information stored at the end of the packet. This method is secure from node to node. It is not decrypted en route by the AP or distribution system. However, because it is decrypted as part of the protocol stack of the node, it is limited to the security of the operating system of both nodes. Further, 40-bit keys are breakable, so a really persistent cracker will still be able to break through. Some vendors are considering higher level security mechanisms for non-exportable versions.

Vendor acceptance
This standard is still fairly new but has been adopted by many of the leaders in the wireless LAN area. Many simply had to make some tweaks to get their products in spec. Others may wait until a future standard comes into line with their products. Table 1.1 shows a few vendors and their products for wireless LANs.

Wireless LAN products
Vendor Product Type Range Indoors Frequencies Bandwidth 802.11 compliant?
Lucent WaveLAN WLAN 200-800 ft. 902-928 MHz, 902-928 MHz, 2.4-2.4835 GHz 1-2 Mbps Yes
Netwave AirSurfer WLAN 500 ft. 2.4-2.4835 GHz 1 Mbps Yes
Aironet 1200, 2200 WLAN 500 ft 902-928 MHz, 2.4-2.4835 GHz 1-2 Mbps Not Yet
Proxim RangeLAN2 WLAN 500 ft. 2.4-2.4835 GHz 1.6 Mbps No
RadioLAN CardLINK WLAN 120-300 ft 5.8-5.9 GHz 10 Mbps No
Windata FreePort WLAN 260 ft. 5.725-5.8 GHz 5.7 Mbps No
BreezeCom Breezenet WLAN 500ft 2.4-2.4835 GHz 1-2 Mbps Yes

The P802.11 working group will meet again in June 1998 to discuss expanding the standard to include other frequency bands and ratifying small elements of the current standard. The most interesting upcoming developments are the 5-GHz band and wireless personal area networks (WPANs). The higher 5-GHz band (actually 5.725 GHz to 5.875 GHz) may expand the data bandwidth available for wireless networks from 5 Mbps up to 50 Mbps. That's a 1,000 to 50,000 percent increase in overall bandwidth from the current system. The problem is that this band is available in the U.S. but not necessarily anywhere else.

WPANs have a much shorter range of less than 10 meters and significantly lower power consumption and data transmission rates. Yet, WPANs will bring about a new revolution in wearable computing devices. The market for such devices range from manufacturing and distribution to healthcare and service organizations. Although the complexity of such devices will not be too great, their utility has the potential to really speed up many ordinary repetitive tasks that we do by hand today. The days of Star Trek-like insignia communicators aren't so far off.

Wireless LANs have finally reached standardization and are becoming available in PCI and PC card formats. Some vendors are sticking to the traditional Windows 95 market while a few are even considering the Windows CE handheld market. Other vendors like Telxon, which focuses strictly on industrial applications, are even making WLANs for JavaOS systems. The need for WLANs is obvious, especially with the growing complexity of electronic devices and the expansion of network computing.

Resources

About the author
Rawn Shah is chief analyst for Razor Research Group covering WAN and MAN networking technology and network-centric computing. He has expertise in a wide range of technologies including ATM, DSL, PC-to-Unix connectivity, PC network programming, Unix software development, and systems integration. He helped found NC World magazine in December 1996, and has led the charge to the deployment of network-centric computing in the corporate world. Reach Rawn at rawn.shah@sunworld.com.

What did you think of this article?
-Very worth reading
-Worth reading
-Not worth reading
 -Too long
-Just right
-Too short
 -Too technical
-Just right
-Not technical enough
 
 
 
    

SunWorld
[Table of Contents]
Sun's Site 		[Search]
Feedback 		[Next story]
Sun's Site

[(c) Copyright Web Publishing Inc., and IDG Communication company]

If you have technical problems with this magazine, contact webmaster@sunworld.com

URL: http://www.sunworld.com/swol-05-1998/swol-05-connectivity.html
Last modified: