Some of the first products based on 802.11n show the wireless standards performance potential—and problems.
During eWEEK Labs tests, Linksys products based on Version 1.0 of the 802.11n draft standard were indeed fast—faster than anything weve tested to date—but issues with range and interference with legacy wireless networks show room for improvement.
The IEEE 802.11n standard promises massive increases in aggregate wireless throughput and range through the use of MIMO (multiple input, multiple output) spatial multiplexing techniques. The standard calls for link rates of 150M and 300M bps, which could translate to as much as 150M bps of real throughput capacity.
The current draft of the 802.11n standard was approved for letter ballot in March; the full standard is expected to be ratified by the second quarter of 2007.
We tested Linksys new WRT300N Wireless-N Broadband Router ($149) and the WPC300N Wireless-N Note-book Adapter ($119), both of which became available April 24. Both the router and notebook adapter feature three integrated antennae and support 40MHz channels for increased throughput performance.
The Linksys products we tested are based on Broadcoms Intensifi chip set. However, Linksys officials said they typically move so many wireless products each month that chip-set availability is a problem, so they will also be releasing draft 802.11n products based on chip sets from Atheros and Marvell.
Mike Hurlston, vice president and general manager for Broadcoms Wireless LAN Division, claims that the Intensifi chips will support the 5GHz band, but none of the first-generation draft 802.11n products will operate in that frequency. Hurlston said a desire to keep costs down is the primary reason why Broadcoms hardware partners have not yet added support for that spectrum.
Early indications are that draft 802.11n performance will be highly sensitive to environmental conditions. Initial test runs in our San Francisco offices failed to elicit performance above 40M bps (well below expectations) because the draft 802.11n products quickly backed off from full performance when legacy wireless products were detected in the same vicinity. Therefore, we moved testing to the relatively clean airspace of a home office.
“We have a mode in our chip set that is able to detect whether adjacent networks are operating and shrink the 40MHz channel back down to the normal 20MHz, so we can actively modulate between 20MHz and 40MHz, depending on traffic that is seen,” said Broadcoms Hurlston. “[802.11n] affords that opportunity, and we implemented it as a standard feature in our product.”
When viewing the WRT300Ns performance with a spectrum analyzer, we noted that the 40MHz channels cut a wide swath of noise in the 2.4GHz spectrum. We expect that the way wireless administrators allocate channels in the 2.4GHz band will need to change dramatically once 802.11n products gain wider acceptance.
We also found that Linksys draft 802.11n router caused performance issues with legacy 802.11g networks. With the WRT300N router beaconing nearby (with no clients associated it to it) in the same part of the spectrum as our production 802.11g access point, we periodically lost association with the 802.11g access point and needed to manually reconnect—something wed never experienced before with the network. Once we reprogrammed the WRT300N to the other end of the spectrum, the dropouts stopped.
The 802.11n task group is aware of the current drafts issues with legacy wireless LAN devices (specifically with how 802.11n shares bandwidth with attached legacy clients), and representatives from Cisco and Motorola broke off to look into the issues before the next meeting of the draft subcommittee, which is scheduled for May.
Expectations vary widely, depending on whom you talk to. In previous conversations with Dave Borison, Airgos director of product marketing, we leaned that Airgo is not making chip sets based on the draft standard because the company thinks the issue of legacy interoperability is significant enough to necessitate small modifications to the silicon.
Broadcoms Hurlston, meanwhile, doesnt think the changes will be significant enough to warrant hardware modifications. As with 802.11g, Hurlston anticipates that a software upgrade will address the problem and that Broadcoms current line of chip sets will be fully upgradable to the true standard, whenever it gets ratified.
During discussions with eWEEK Labs, representatives from the various companies producing products based on the 802.11n draft standard stopped short of guaranteeing that such an upgrade will be possible. So, buyer beware—because there wont be a refund if it isnt.
With this uncertainty in mind, it is not advisable to invest in these products lock, stock and barrel. Enterprise-grade WLAN manufacturers continue to wait for the standard to fully bake, and enterprise customers should do the same. Like the Airgo MIMO-based products weve tested during the last year and a half, draft 11n products should be considered only for highly specific needs requiring a fast wireless connection. And buyers should not yet expect the products to support the standard down the road.
2
During tests, we found Linksys draft 802.11n gear to be the fastest wireless equipment at short distances that weve seen to date, besting even a pair of products based on Airgos Gen 3 True MIMO chip set. With a maximum theoretical link rate of 270M bps, Linksys WRT300N router and WPC300N notebook adapter combined to top out at 112.17M bps of real full-duplex traffic.
In comparison, our previous wireless throughput champion, the NetGear RangeMax 240 (WPNT834)—which is based on Airgos Gen 3 True MIMO chip set—could only get up to 104.53M bps in the same test. We also found that not all products based on the Airgo technology are created equal, as the ASUS 240 MIMO Series Router couldnt come close to the performance we saw with the RangeMax.
To measure performance, we used iPerf 1.7.0, a free bandwidth measurement tool from the Distributed Application Support Team of the National Laboratory for Applied Network Research. We used this tool to measure total bandwidth, half-duplex performance for uploads and downloads, and bandwidth performance at various distances.
In our upload/download tests, we noted that Linksys draft 802.11n products excelled particularly with wireless clients uploading data to the wired side of the network, but the unidirectional traffic numbers were ultimately bottlenecked by the routers 10/100M-bps Fast Ethernet switch, which cant move as much half-duplex traffic as the WRT300Ns wireless networking components. We found that a single WPC300N-based wireless client could upload an impressive 91.6M bps (in comparison, we could move 94.5M bps of half-duplex traffic between wired clients), but it could download only 81.57M bps.
In our experience, wireless-enabled systems are more likely to need to download large amounts of data than to upload them, so wed prefer to see those numbers reversed.
In our range tests, Linksys products did not fare nearly as well, however. Starting at 50 feet (which included 20 feet of elevation and a few walls), the draft 802.11n performance lagged considerably when compared with the Airgo products we tested against.
With the WRT300N, Linksys introduces a new form factor for its router family. Its about an inch less wide than the older WRV54G. It also has three antennas, like Linksys WRT54GX4 MIMO router, but the WRT300Ns middle antenna is a flat paddle. (As we confirmed when we pried it apart, the WPC300N notebook adapter has three integrated antennas.)
We tested the Linksys router running the latest firmware, v0.92.4, which presents the same familiar configuration pages weve seen with Linksys products during the last couple of years.
However, we did notice a few screens of interest.
Under the Wireless tab, we configured the router to use wide channels—Linksys way of describing 40MHz channels—for the wireless link. We could select only from among channels 3 through 9 for the wide channel, further limiting our options in the 2.4GHz band.
We could set a 20MHz channel within our wide channel for use with legacy clients. For instance, when we selected channel 9 for our wide channel, we could select either channel 7 or channel 11 for the 20MHz channel.
The WRT300N offers up-to-date wireless security, supporting both the enterprise and personal versions of WPA (WiFi Protected Access) and WPA2. For some mysterious reason, however, Linksys chooses to use the terms “PSK” and “PSK2” (as in PSK2 Enterprise and PSK2 Personal), which could certainly lead to confusion in a class of product already rife with acronyms.
We also noted that the routers help system needs an update, as the various pages have not yet been updated with information specific to dealing with the new radio parameters.
3
Netgear was first out of the gate with products based on the 802.11n draft, but we have been unable to acquire one of its products. In early March, Netgear announced the WNR854T RangeMax Next Wireless Router Gigabit Edition. Based on a chip set from Marvell, the WNR854T has a Gigabit Ethernet switch to remove the throughput bottleneck on the wired side of things.
It appears that Netgear has released only a limited number of these devices so far. Early units of the WNR854T were briefly available from TigerDirect, but they are not available at this time.
Netgear also announced a 10/100 model based on the Broadcoms Intensifi chip set. This product, the WNR834B RangeMax Next Generation Wireless Router, is expected to be available this week.
In the coming weeks, we expect to receive more products based on the 802.11n draft standard, including Netgears. As we receive more devices, we will conduct tests to evaluate interoperability among different hardware manufacturers (and chip-set vendors) products. We also will further explore the claims of the new standards interference with legacy wireless clients.
Technical Analyst Andrew Garcia can be reached at [email protected].
Check out eWEEK.coms for the latest news, reviews and analysis on mobile and wireless computing.