Stretching The Wireless Connection

Recent trends indicate that enterprise wireless LANs are becoming more pervasive and less tactical. Many vendors are certainly headed in that direction. For example, Aruba recently announced its grid-like architecture for ubiquitous enterprise wireless access points.

However, as WLANs become more pervasive, one need becomes increasingly clear: Enterprises and consumers are starting to clamor for their WLANs to have greater range.

In the enterprise, part of the desire for greater range is financial; every additional access point requires another Ethernet and electrical run, which can be difficult to accomplish and expensive. On the home front, the return rate for SOHO wireless equipment can be as high as 30 percent, much of which can be attributed to poor wireless coverage, according to some industry observers.

Ultimately, the IEEE 802.11n standard will satisfy the need for greater coverage. But that standard is at least a year, and more likely two years, away from ratification.

That's why it shouldn't be surprising that some technologies to increase range are starting to appear. Examples include Belkin's so-called pre-N access points based on Airgo Networks' MIMO technology and Vivato's new and expensive panel-sized access point.This article will discuss why it's so difficult to increase range and look at a couple of WLAN products that claim to do so.

The Ins And Outs Of Stretching Wireless

Traditional antennas provide some gain -- the amount an amplifier increases the strength of a signal -- by focusing the wireless energy in a certain direction. The dipole (or rubber duck) antennas as seen on most access points are rated around 2 dBm and generate a donut-shaped coverage pattern. Patch antennas can increase that to around 6 dB by using a more focused signal.

There are two more ways to accomplish better coverage. The first is to increase the transmitter's power, although FCC regulations limit transmit power for point to multi-point coverage in the 2.4 GHz spectrum. Using an omni-directional antenna with a gain of 6 dBi or less, the maximum EIRP (effective isotropic radiated power, or power from the antenna) is 4 watts. That's much higher than the more powerful access points that can transmit at 100 mW with 2.2 dBi antennas, or about 166 mW EIRP.

So why not just crank up the power for WLAN equipment? For starters, more power output would drain laptop batteries much more quickly. Also, if the transmitter power is high it will easily overpower neighboring access points and client cards. Finally, European Union standards generally limit transmission to 100 mW EIRP.The second way to achieve better coverage is to focus on the receive sensitivity, or how well the device listens. The physics of electromagnetic transmission is that the power decreases by the square of a given distance. So at two feet the power is four times weaker than at the antenna and at five feet its power is 25 times weaker.

That means if the device transmits at 100 mW (which is equal to 20 dBm), that at five feet the receiving power will only be 4 mW, or 6 dBm. Extend this 100 feet, which is not uncommon, and then it becomes only 0.01 mW, or -20 dBm. That's a loss of 40 dB in free-space, but when walls, objects, people and interference or noise are introduced, 100 feet usually translates into a much larger loss in signal.

Most client cards are good enough to listen to about -85 dB, which means if the access point is transmitting at 20 dBm (100 mW) then 105 dB of loss can be introduced and the card can still pick out the signal. A good rule of thumb is that another 6 dB in signal is approximately twice the coverage area.

Two Attempts At Extending Range

We examined products from two vendors that make long distance or high power PCMCIA cards and access points: ParkerVision and SMC. Products from a fourth vendor, Belkin, will be covered in a separate Mobile Pipeline review in the near future.

ParkerVision's line of wireless products includes a USB wireless adapter (USB1500), a wireless access point (WR3000) and a PC Card (WLAN1500). SMC provided their SMC2555W-AG tri-mode AP and their SMC SMC2532W-B card. We compared them to the Cisco 350 client card and the Cisco 1200 access point.Overall, the results were disappointing. Although the ParkerVision and SMC products claimed to have much better coverage, their advantage over the Cisco client card and access point was minimal. When compared to a Linksys combination (WAP11 access point and WPC55AG client card), the products we tested provided only slightly better indoor coverage.

In fairness, all the advertisements and marketing blurbs from the vendors refer to outside coverage, because indoor environmental factors such as walls and other large objects will quickly attenuate any strong signal.

We performed two types of tests. In the first round we configured each vendor's client card on the Azimuth 800W test system with the same vendor's access point in the mini-test head. We ran the traditional rate-versus-range test to find out the speeds and the amount of loss these clients could sustain as well as the access points' antennas, but their absence will vary their total link budget by just a few decibels and the access point antennas all had 2 or 2.2 dB gain.

The next test was a site survey in the building that encompasses Syracuse University's RealWorld labs. Every combination of card and access point was tested for a total of 9 site surveys. We also tested a Linksys access point with all the cards. The results between the two types of tests were quite consistent.

ParkerVision The ParkerVision product, originally marketed under the HORIZONS Wireless moniker, is one of the few 802.11 wireless products wholly developed and manufactured in the United States. Their technology is based on "RF Energy Sampling," which instead of carefully handling the carrier wave during signal processing, 'crushes' it to extract the data. The company says this technology is the source of better performance.

The ParkerVision access point ($130 on their website) is really a SOHO gateway, although company representatives claim that they have received a lot of interest from the corporate market because of the device's wireless coverage.

The client card ($100 on their Web site) is larger than most, protruding about 2 inches from the edge of the laptop and has signal strength of 21.2 dB (132 mW). The software utility, which allows basic site surveying and profile configuration and setup, appropriates much of the screen.

One problem is that the driver doesn't integrate or cooperate with Windows XP's wireless Zero-Config utility. Our Azimuth testing showed that the NDIS driver does not communicate rates lower than 11 Mbps, even though the ParkerVision client interface does show it. Even at the fringes of our test area the signal strength stayed above 30%, and then the connection would drop.

The ParkerVision combo won out for range, with an absolute loss of 114 dB on the downstream, and an even higher value of 118 dB on the upstream tests. It was able to sustain an error-free 11 Mbps connection until there was an absolute path loss of 104 dB. Between 104 and 107 dB the actual data throughput dropped because of errors, such that it actually dropped to 5.5 Mbps connection at 107 dB. Under optimal conditions the top downstream rates using a specific Chariot script were only 4.75 Mbps. Upstream speeds fared slightly better at about 5 Mbps.However, the results of the ParkerVision devices site survey were mixed. ParkerVision's access point had the least range, even compared to a Linksys access point, but outside its coverage area it managed to provide more intermittent connectivity than the other products outside their respective fringe areas.

The client card, on the other hand, had the best coverage against all the access points, performing the best with the Cisco access point. Ping times were also somewhat erratic near the far edges, anywhere from 3 to 5 milliseconds. But the extra coverage gained can be counted in feet, not yards. If you need to get the most coverage possible and you can't choose the access point, the ParkerVision wireless card is your best bet.

Cisco

We tested with Cisco's 350 client card ($169 list) and the Cisco 1200 access point ($899 list) with integrated 802.11b. Both are rated to transmit up to 100 mW. It has a high price tag, but it has traditionally set the bar in performance, both in coverage and speed, and we use it in our labs as the baseline for coverage and performance.

Cisco had an upstream and downstream range of 111 dB absolute path loss, three decibels fewer than the ParkerVision product, and sustaining an error-free 11 Mbps rate for both the up and downstream tests at 100 dB. Between 100 and 103 dB the actual data rate dropped because of errors, such that it actually dropped to 5.5 Mbps at 103 dB. Downstream rates were measured at 5.6 Mbps, while upstream fared slightly better at 5.9 Mbps.Our site survey showed that the Cisco 1200 access point had the best coverage of all the access points we reviewed. The Cisco 350 client card consistently did better than the SMC but worse than the ParkerVision card. One thing we noticed was the Cisco card's lower latency denoted by pings around 1 to 2 milliseconds, as compared to variable 2 to 4 milliseconds of the other cards.

SMC

Unlike ParkerVision, SMC doesn't claim any advancement in signal processing technology to its advantage other than increasing the transmit power of their SMC2532W-B card ($69.99 MSRP) from a common industry peak of 100 mW (20 dBm) to 200 mW (23 dBm). The access point that they supplied, the SMC2555W-AG ($799.99 MSRP) supports for 802.11a, b or g. They also have a lower-cost access points that provide just 802.11b service, also at 100 mW.

The SMC access point and client card pair eked out a total upstream and downstream range of 109 dB, and it maintained an error free 11 Mbps downstream rate until it had an absolute path loss of 99 dB. It dropped to 5.5 Mbps at 101 dB. The upstream statistics were much better. An error-free 11 Mbps upstream rate lasted until 102 dB, but it quickly dropped off thereafter. In optimal conditions, upstream and downstream data throughput rates were almost identical, about 5.8 Mbps.

Surprisingly, our tests showed that the SMC access point had better coverage than the ParkerVision's access points, even though its power output was rated 4 dB less. The SMC client card, though, had the least range, but only by a few feet. Although its performance was fair, the driver and/or PCMCIA interface was such that we had to invariably reboot the laptop after inserting the card; otherwise it would constantly disconnect and reconnect. A company representative claimed that this was not a known issue and supplied us a newer driver not available from their Web site that we did not have a chance to test.Bottom Line

If you thought there was a magical solution to increasing coverage, they are not to be found in these products.

If you are looking for the best possible coverage at the lowest price, something that concerns most consumers, use the ParkerVision 1500 series of products. Cisco's pair will set you back three times as much as ParkerVision's and won't provide any better coverage. The SMC solution only provides slightly better coverage than the Linksys combo that we tested.

In the end, though, while the high cost of metro-wide and outside deployment may motivate you to find product with better coverage, denser deployment of access points may well be easier.