Nobody who watches a video or listens to music appreciates gaps or skips in the programming. But there is a real potential for these problems as it becomes practical to beam audio or streaming video over wireless networks.
Until recently, the idea of sending such material via wireless LAN was only theoretical. Existing wireless networks were too slow to handle data-intensive video. Moreover, there was little need for video on the business enterprise networks where wireless local-area networks (LANs) first took hold.
This situation has changed as wireless nets have moved to 2.4 and 5-GHz transmission bands. These support data rates high enough to handle full-color video. In addition, industry observers see wireless networks migrating into the home. They think interactive gaming and other graphic-intensive uses will foster a need for high-bandwidth wireless.
Trouble is, gaps and skips can crop up when transmitting video or audio over a LAN due to effects that generally don't affect ordinary files. The problem arises out of techniques in the 802.11 protocol that govern how nodes access the network to send information. Devices that request use of the LAN back off and repeatedly try again if they discover the network is already occupied. Devices waiting a long time eventually get a higher priority. This lets a new data-sender temporarily interrupt transmitting devices, putting them back in the queue to get network access.
This behavior causes few problems when transmitted data are just ordinary files. But it is obvious when a source sending real-time video or audio gets kicked off the network. And juggling network resources becomes even more problematical when numerous sources of audio and video are involved.
Fortunately there are enhancements to the 802.11 spec that can let networks accommodate video and audio streams without interruptions. These come under the moniker Quality of Service (QoS) which, in a nutshell, minimize the chance that audio and video streams will be interrupted once they are on the network. The technique used is to assign a priority to all network processes, with audio and video getting the highest priority. This keeps less time-critical tasks from interrupting them.
There are other difficulties with audio/video transmissions that take place on wireless networks operating in the 2.4-GHz band. Networks in this band share the spectrum with common household devices that include microwave ovens and wireless phones. Such appliances can interfere with network signals and thereby garble video or audio streams. Standards committees are now devising extensions to the 802.11 protocol that permit operation even in the midst of these sources.
Interestingly enough, work devoted to better video streaming may have an impact on wireless networks deployed in factories and businesses. The reason is that QoS features can give any sort of network transmission a high priority if need be. This principle can ensure that important messages beamed between industrial machines, say, can get through in real time. Similarly, measures devised to make home network communications more reliable in the 2.4-GHz band may benefit wireless nets in electrically noisy factories.
GIVE ME FIVE — GIGAHERTZ
An example of QoS facilities soon will be found in a 5-GHz chipset recently developed by Intersil Corp., Irvine, Calif. Its set of Prism Indigo ICs costs about $35. Intersil figures that WLAN adapter boards built around these devices can cost about $60. Firmware written for Prism Indigo incorporates QoS enhancements for video streaming as well as for Voice over Internet Protocol (VoIP) and multimedia-related services.
The QoS features for video will only become available this summer, says Intersil Senior Manager of Strategic Marketing Bruce Kraemer, because the standards themselves have yet to be finalized. Intersil sees the chips, which implement 802.11a protocol, going into wireless gateways built as part of DSL or cable boxes in the home, as well as in cardbus and miniPC cards used in business enterprises.
The 5-GHz band offers some inherent protection against radiated interference, Kraemer points out, partly because the sources that are in that band transmit infrequently. "The (5-GHz) band has been allocated to satellites, radar, and video location services," he explains. "If you look hard enough you can find interference from those sources but it's weak and not geographically stable. Earth exploration satellites, for example, pass over a given point on the planet's surface once every two weeks and radiate it with a few nanowatts of energy."
In factories and business enterprises, broadband noise sources such as arcwelders and high-power relays are more likely to garble network transmissions. But the RF transmission technique used for 5-GHz WLANs offers some inherent protection from even these potential interferers. "The orthogonal frequency division multiplexing (OFDM) used in 5-GHz networks happens to be relatively immune to impulse noise," says Kraemer. "If a wideband source desensed your radio, you would just need a higher carrier-to-noise ratio to detect the network signal. It might cause a problem at the fringes of your coverage area, but the receiver would not mistake the noise source as being another network node."
OFDM transmission divides the data stream into multiple parallel bit streams. The bit streams have a lower bit rate than the original material. They each modulate several subcarriers. In 802.11a systems at 5 GHz, a 20 MHz-wide channel gets divided into 52 such subcarriers. The subcarriers are spaced apart in frequency such that, at the receiver, the demodulator for one doesn't "see" the modulation of other subcarriers. This property ensures there is no cross talk between subcarriers, though their spectra overlap (hence the orthogonality in the name).
In addition, OFDM systems can cancel out signal echoes (as arise, for example, from rain) by judicious choice of timing relationships in modulation. All in all, echoes present in the signal that have a delay not exceeding some predetermined amount won't cause interference.
It also turns out to be relatively easy to modulate and demodulate the numerous subcarriers involved in OFDM. The reason is that the necessary tasks are equivalent to Discrete Fourier Transform operations, for which there are efficient Fast Fourier Transform (FFT) algorithms. These algorithms lend themselves to implementation in either software or ICs, making feasible mass-produced OFDM transceivers that are affordable.
WI-FI AT 2.4 GHZ
There is a proposal to implement OFDM in 2.4-GHz 802.11b networks as well as at 5 GHz. But 2.4-GHz networks deployed so far either used frequency hopping or direct sequence spread spectrum transmission. Wi-Fi nets installed in factories tend to use frequency hopping. Frequency-hopping transmissions modulate a data signal onto a narrow-band carrier that hops in a random but predictable sequence as a function of time. A hopping code determines the transmission frequencies. A receiver uses the same hopping code as the transmitter to listen for carriers at the right time and frequency.
Vendors say frequency-hopping transmission tends to work more reliably than DSSS in electrically noisy industrial environments. The reason is that narrowband interference only affects the signal if both happen to be at the same frequency at the same time.
DSSS systems artificially spread the frequency spectra of the carrier through modulation with pseudo-random digital codes. These codes exceed the data rate of the transmitted signal by a large degree. At the receiver, a locally generated replica of the pseudo noise code gets correlated against the demodulated signal. The correlator output is the original signal of interest.
One advantage of DSSS systems is that ordinary narrowband emissions (as produced by radio stations and most other analog transmitters) generally don't interfere with DSSS signals unless they are super strong. The reason is that during demodulation, a correlator spreads out any existing narrowband interference as it reconstructs the original transmitted data. This spreading action effectively cancels out interference below a relatively high threshold.
As wireless networks move into homes, one fear is that appliances such as microwave ovens and wireless phones may put out signals powerful enough to cause problems. The reason is that these appliances radiate at frequencies near 2.4 GHz where experts expect home networks to operate at least until the cost of 5-GHz versions come down.
Such sources of interference are being addressed, however, in new versions of 802.11 standards. One recent development in this area is a protocol nicknamed Whitecap by its creators at ShareWave Inc. (which has been acquired by Cirrus Logic, Austin, Tex.). Whitecap chipsets now implement 802.11b but are considered among the first incarnations of QoS enhancements to be included in the 802.11e spec for multimedia applications. Some developers of Whitecap, in fact, are on the IEEE committee devising the 802.11e protocols.
Whitecap protects streaming video content from interruptions and interference by borrowing a few techniques from digital satellite operators. One of these techniques is forward error correction coding. Basically the idea is to add a few extra bits to each transmitted frame so that a decoder can attempt to correct-on-the-fly errors that crop up during transmission. This contrasts with the normal method of simply dropping received data that are corrupted and asking for a retransmission (OK for files; not OK when the dropped data are a few video frames).
Whitecap2 (the latest version of the protocol) also builds in a means of avoiding sources of in-band interference: One node in the network serves as a coordinator, monitoring packet rates communicated from devices on the network. If it notices a problem with channel quality, it tells all nodes in the wireless LAN to switch channels. And if something goes wrong with the coordinator node, other network nodes decide among themselves which should take over the role of coordinator.
Similarly, devices on the network can communicate directly with each other rather than route messages through a central device or server, as is the case in standard 802.11 networks. This peer-to-peer topology also works well with enhancements planned for the future such as multicasting. This is the transmission of media traffic only once to multiple clients on the network (rather than a separate transmission to each receiver individually).
Several home networking devices have begun incorporating Whitecap technology, which is available in the form of PCI, PCMCIA, 10BT, MiniPCI, and USB-format chipsets. Uses include network interface cards, bridge/access points, and links between set-top cable boxes and interactive TV screens that double as terminals for surfing the Internet.
WIRELESS NODE HITS 200 MPH
The Our Gang Racing Team has set up what might be the "fastest" wireless LAN in the world. One node on the net sits in a Pontiac Firebird that competes in the International Hot Rod Association's Top Sportsman class. A 1,380-hp, 700-in.3 engine propels the dragster through the quarter-mile at speeds over 200 mph.
Team owner John McCaully also runs Industrial Automation and Control, a Pennsylvania-based consulting firm and integrator of I/O systems from Opto 22 in Temecula, Calif. In the car is an 802.11b wireless I/O node that operates at 2.4 GHz and employs frequency-hopping transmission, one of the RF schemes the standard supports. The Opto 22 SNAP Ethernet gear lets the dragster pull into the proximity of the team's laptop for a quick data download after a race. This eliminates the need to hook up data cables, a process that had been eating up several precious minutes between runs.
WIRELESS NETWORKS FOR CRAMPED QUARTERS
The home network scenario is one where numerous wireless networks put out signals that overlap and potentially interfere with each other. This is a distinct possibility in closely located homes or apartments, or even in offices and production facilities.
Features planned for the 802.11e protocol take just such scenarios into account. The idea is to first let individual networks switch channels automatically until they find space to transmit at full bandwidth. But if there are more overlapping subnets than available channels, colocation features kick in. Two different subnets can share the same channel through negotiations between nodes that function as coordinators.
Coordinators on different subnets report bandwidth requirements to each other and monitor network transmissions to avoid interference. If coordinators are too far away to "see" each other, some other node on the network jumps in as a proxy coordinator to forward the necessary traffic to the coordinator of the other subnet.