|Wireless Networking - An Overview|
|Originally published June, 1997|
|¿ 1997, 2005 Carlo Kopp|
|The wireless networking
paradigm is beginning to emerge in the marketplace, as an increasing
number of vendors adopt this new approach to moving data between
machines. At this time the principal limitation of wireless networking
technology is bandwidth, and given the available technology and
spectral bandwidth, this is unlikely to change in the forseeable
In the most general sense, a wireless network is any network in which data is transmitted between machines using a medium other than dedicated cables, regardless of whether these are copper or optical. More specifically, wireless networks use either radio links for indoor and outdoor connectivity, or infrared optical links for close range indoor connectivity.
A special category of wireless networks are those which distribute the networking signals over a carrier wave on the mains wiring (this is interesting, in that a wire is still used to carry the data, but it is not a dedicated LAN wiring scheme). Will wireless networks replace the existing cabled network infrastucture ?
In the forseeable future, the answer is no, because the bandwidth limitations of wireless technology make it simply uncompetitive against fast Ethernets running over fibre or copper cabling. However, the wireless network can be an extremely useful supplement to the fixed network infrastructure. For one, it does not require any cabling support, other than to a wireless transceiver.
A wireless network can provide network access to portable machines such as laptops used in building areas away from the existing network, or provide a practical means of hooking laptops into the existing network infrastructure with a minimum of fuss. An excellent example would be a laptop used by a sales type, who carries it off site to impress customers. Once he/she returns to the central office, data can be uploaded or downloaded without the messiness of having to plug the machine into a desktop docking unit.
Similarly, areas such as meeting rooms, demonstration rooms, and other common use facilities may be provided with network access without the cost and messiness of cabling. Organisations which provide migratory teaching or training activity, need not haul a box of network cables to a client's site. Another example would be the situation where a temporary network is required, and the cost of a wired network cannot be justified for the short time it is needed. Extending this model further, use of radio or laser network repeaters provides connectivity between buildings in situations where leased lines are too expensive, and dedicated inter-building cabling connection is not feasible.
A good example would be a situation where an organisation is split between two buildings which are within line-of-sight of one another, but separated by roads or other people's real estate. Where other people wish to charge a fortune for the privilege of running a cable across their property, a wireless repeater link may be a viable alternative. Up to this point we have looked at the issue of wireless LANs. However, this is not the only possibility which exists.
Another is the wireless Wide Area Network (WAN) or Metropolitan Area Network (MAN), using either satellite, microwave, packet radio, or GSM telephone services as a transmission medium. The wireless MAN/WAN is at this time still suffering bandwidth limitations, in comparison with wireless LANs, but it offers some interesting possibilities. Personnel who need access to a central server while off site may be provided with network access, without the need to hunt for uncommitted telephone wall sockets on a client's site.
Moreover, network access may be provided to truly mobile platforms such as company vehicles. An example situation is where service or support personnel can look up a database, while on a client's site. The old method of hauling many kilos of paperwork out to site simply disappears. The wireless WAN/MAN is likely to be of significant usefulness to any organisation which provides a service on a client's site, an excellent example is healthcare, where a worker can look up diagnostic data or case studies at arbitrary patient's homes.
This is without doubt a technology which has tremendous potential, particularly for support and service oriented industries. To better appreciate the strengths and limitations of the various alternatives available, we will take a closer look at the basic techniques used for this purpose.
Optical Wireless Networking
Optical free space links have been commercially available in various forms since the eighties, their proliferation resulting from the availability of low cost semiconductor lasers, high power Light Emitting Diodes and PIN or avalanche diode detectors. It is of some interest that these supporting technologies grew out of the massive development effort supporting the initial deployment of optical fibre networking in the early eighties. The use of these semiconductor devices for industrial purposes, and free space links, is a very good example of technological spin-off from basic technological research.
The simplest application of optical networking is the indoor wireless optical network. In this arrangement, transceivers similar in concept to our familiar TV or VCR remote controller, are used to carry traffic between machines in an office environment. The conceptually simplest scheme in use is termed the
Multiple machines with such transceivers all pointed at the same patch of the ceiling can then receive traffic from, and transmit traffic to their peers. This arrangement is analogous in many respects to the Ethernet collision detection shared channel model. The reflective ceiling assumes the role of the cable in the classical copper Ethernet. Understandably, dummies who drop their notes or manuals on to the transceiver will cut their machine off from the rest of the network. A more complex and less frequently used strategy is that of the
In this arrangement, separate optical receiver and transmitter heads are employed, using typically 802.5 Token Ring protocol. The infrared beam from one host's transmitter to another's receiver supplants the Token Ring cable. Needless to say an important limitation of this techniques is that a dummy standing in the beam will interrupt the ring and take it down. The typical application fro this technology is in large and continuously shuffled open plan offices, where cabling causes endless troubles. Bandwidth in existing products runs to several Mbit/s, and given the availability of sufficiently fast and powerful LEDs or Laser diodes, tens of Mbit/s are in theory achievable.
Security can be excellent, particularly if Infrared opaque window coatings are used. If not, it is feasible that a third party with a custom built Infrared receiver and a large dish reflector could feed a sniffer from an adjacent building, receiving radiation leaking out of windows. Subject to design, direct sunlight in a work area can degrade performance. Should traffic need to be carried between buildings, optical point to point datalinks can be employed.
Many commercial products, with either CCITT (ITU) or Ethernet line interfaces, are available in the marketplace. A typical product in this class employs a laser transmitter, avalanche or PIN diode receiver, and an embedded Ethernet bridge or regenerative repeater. Ranges of about a kilometre can be achieved, with bit rates in excess of 10 Mbit/s (some years ago the author did a paper design for a 32 Mbit/s product on this class, with about 1 km range). A limitation of such designs can however be sensitivity to haze, fog, and heavy rain, which scatter the laser beam and degrade signal strength to the point of partial or complete dropout.
Radio Frequency (RF) Wireless Networking
Radio frequency ( ie microwave ) technology promises to significantly improve the flexibility of LANs and MANs. Unlike infrared optical LANs, which are limited to direct line od sight in an uncluttered office environment, RF technology provides a respectable capability to penetrate through building walls, windows and floors/ceilings. As a result, RF LAN/MAN technology has the potential for both indoor and outdoor use.
It is of some historical interest that the use of collision detection radio techniques for packet data protocol transfers predates the copper Ethernet we love and know so well. The first use of these techniques dates back to the Aloha network built for research and teaching usage purposes by the University of Hawaii, using narrowband radio links to carry computer traffic between campuses spread across the Hawaian Isles.
Since it was too expensive to lease submarine cables, in true University fashion clever improvisation was applied to solve the problem. The creators of the Ethernet, Metcalfe and Boggs, subsequently adopted this revolutionary idea and applied it to a copper network using cable TV components such as taps, connectors and cables. So it is interesting to contemplate that the technology has travelled a full circle, returning to its RF origins.
In the last issue we discussed spread spectrum communications techniques, which are the foundation of this new technology. Spread spectrum techniques employ pseudo-noise (PN) coding methods to provide a signal with excellent immunity to interference, as well as a respectable measure of robustness against eavesdropping. Because of the process gain produced by using PN coding of the baseband signal, spread spectrum techniques also provide the ability to operate robustly with relatively weak signal strengths, and thus accommodate the arguably pathological transmission environment typical of urban areas.
The enabling circumstance for RF LANs and MANs, other than maturing communications hardware technology, was the essentially international agreement by radio spectrum management authorities to make the Industrial, Scientific and Medical (ISM) frequency bands available for use without licence, providing that equipment meets specified (low) transmit power levels. The ISM bands occupy 28 MHz, 83.5 MHz and 125 MHz slices in the 900 MHz, 2.4 GHz and 5-6 GHz bands respectively. Not a huge amount of bandwidth, but enough to provide several Megabits/s channel throughput subject to PN coding type used. Importantly, the transmit power limits essentially constrain range to less than a kilometre, in an uncluttered outdoor situation, and much less in heavily cluttered environments.
The only real issue with the availability of these bands is potential for narrowband interference, indeed harmonics from the magnetron power oscillators in microwave ovens fall into exactly this band. In the lucrative business of selling spectrum allocations to users, a "freebee" band by default is likely to end up with that which nobody else would pay for. Many jokes exist in the RF networking community, about LANs with run very well until the office microwave oven is turned on.
The basic transmitter and receiver technology is common to adaptor cards for hosts, repeaters (hubs) and bridges, as well as providing, with suitable directional antennas, a means of interconnecting buildings within direct line of sight. At this time several manufacturers are supplying adaptor cards, repeater/bridge hardware, and point-to-point link hardware, the most notable players being Proxim, Lucent, Xircom and Windata. With the booming growth in this technology we can expect others to crop up in the next several years. The principal obstacle to the rapid mass use of RF LANs and MANs has been standardisation of protocols, and the IEEE 802.11 Wireless Ethernet standard has been slow to mature.
This should not come as a big surprise, given the typical jockeying for position which characterises such committees, typically comprising representatives from most of the prominent players in that industry segment. At this time much of the available hardware complies with the draft IEEE spec, but is sufficiently different to be marginally if at all interoperable between different manufacturers.
Once the standard matures, we can expect all vendors to fall into line with the standard, providing the transparent interoperability we see today with the basic Ethernet standards. Until this occurs, any hardware you acquire will most likely be orphaned once standard is available. A future article in this series will examine the IEEE 802.11 standard in much more detail. In terms of networking topology, the RF LAN will in most instances employ the well proven model used in existing 10-Base-T environments, however in RF LAN terminology the basic workgroup is termed a
In this instance, a host will listen for incoming RF traffic encoded with the appropriate PN code, and should it wish to transmit, it will assert its carrier and if no collision occurs, send a packet. The simplest arrangement which exists is the single cell RF LAN, in which an essentially isolated workgroup of hosts communicates via adaptor cards, analogous to a workgroup connected by thinwire Ethernet.
Each adaptor has an omnidirectional antenna, fed via a short coaxial cable from the adaptor card backpanel. The most common adaptors at this time are based upon ISA, PCI and PCMCIA bus standards, reflecting the dominance of PCs as low cost desktop platforms. All hosts broadcast to the cell, and the recipient host responds to the broadcast packet, in the same fashion as an Ethernet works. Should a collision occur, the host backs off and retries. Because a cell uses a unique PN code, interference between cells is avoided.
To connect multiple RF LAN cells, the typical method is to use an RF wireless repeater or bridge, usually termed a wireless access point (AP). Such devices are directly analogous to their Ethernet-Ethernet cousins, but instead map Ethernet protocol messages into the RF LAN protocol, and vice versa. Unlike pure Ethernet devices which can be easily built in hardware alone, a wireless RF AP must have adequate buffering to accommodate the disparity between a 10 or 100 Mbits/s Ethernet on one side, and the hundreds of kbit/s to 2 Mbits/s typical of wireless RF protocols.
The wireless AP thus allows multiple wireless cells to communicate across distances beyond direct radio range, and also allow access to other parts of the wider network. A typical building installation would have wireless RF APs strategically positioned in offices, corridors, warehouses and open plan work areas where network connectivity is useful but cabling installations impractical or simply too expensive.
The APs would be connected to the building Ethernet backbone. It is also quite feasible to use RF wireless techniques to interconnect multiple wireless LAN cells, with the caveat that range requirements must be satisfied. A feature which will be of considerable interest to users on larger sites is the
The limitation of existing roaming schemes is that they are wholly tied to the datalink/physical layer, and thus can only be used in a single IP subnet, thus limiting roaming to a single site, building or office, subject to size and IP subnetting used. Without doubt wireless networking techniques promise to significantly improve the flexibility of networking, particularly by providing connectivity in situations where it would otherwise marginally profitable.
The limiting factor in the technology at this time is immaturity, but this is changing very rapidly and within two years we can expect a significant improvement in interoperability and choice of manufacturers. It will be an interesting addition to the existing array of tools at our disposal.
|$Revision: 1.1 $|
|Last Updated: Sun Apr 24 11:22:45 GMT 2005|
|Artwork and text ¿ 2005 Carlo Kopp|