Wireless chip vendor Engim is releasing test results that show dramatic improvements in WLAN throughput using Engim's chipset.
The tests compared the performance of third-party access points with conventional WLAN chipsets against a prototype access point using Engim's chipset and an Intel network processor. Depending on the mix of WLAN clients, throughput for the Engim-based access point was 10 to 20 times greater than the others, according to Scott Lindsay, Engim's vice president of marketing.
The Acton, Mass., company paid The Tolly Group, an independent testing firm, to run the tests and evaluate the results.
Engim's silicon and algorithms are designed to minimize what's called cross-channel interference. Conventional WLAN chips transmit on only one channel. By contrast, an Engim chip is designed to transmit data simultaneously on several channels at once, increasing the access point's total throughput.
For an 11b/g WLAN, the 2.4-MHz frequency band is divided up into 11 channels. But only three of these - channels 1, 6, and 11 - are "far enough apart" in radio terms to avoid overlapping signals and a resultant loss in throughput. Conventional access points are assigned one of these channels for client connections, ignoring the others. (But chipsets from established rivals Atheros and Broadcom offer proprietary techniques for doubling the 11g data rate to just over 100M bit/sec.)
Engim's technology embraces all three of these non-overlapping channels and then filters out and cancels out the interference to clean up the channels and make all of them useable in a single access point, at the same time.
The Tolly Group tests compared how a mix of 11g and 11b wireless clients performed when connecting to each of three third-party access points, and to the prototype with Engim's EN-3000 11g chipset. Results were measured using the Ixia Chariot WLAN performance analyzer software.
The clients connected to the access points at different distances: the further the distance, the lower the data rate because the 802.11 standard causes the link to throttle back to maintain a reliable connection. When both 11g and 11b clients are accessing the same access point, the picture becomes more complicated: the presence of 11b clients causes an 11g access point, rated at 54M bits/sec, to limit itself to 11M bits/sec, so that 11b clients can connect.
In the tests, with two 11g clients, at 54M bit/sec, and two 11b clients - one at 11M bit/sec, the other at 1M bits/sec - the capacity of the third-party WLANs was in the range of 1M bit/sec to 3M bit/sec for TCP traffic. But for the Engim access point, capacity for the same group of clients was just under 50M bit/sec.
When the two 11b clients both ran at 11M bit/sec, the overall capacity of the conventional access points jumped to 5M bit/sec to 7M bit/sec; the Engim device's capacity was just over 50M bit/sec.
Engim's goal is to supplant Atheros, Broadcom and other WLAN chipmakers as the chip of choice for access point and component manufacturers. Accton Technology is one of the companies that has been testing Engim's chips. A new version of the chipset, which is actually manufactured by IBM, is due out by May.
The Tolly Group tests make a useful marketing tool for Engim, but they won't be decisive for manufacturers making WLAN silicon decisions. They'll have to make some complex decisions about how the price of the Engim chipsets will affect the overall manufacturing costs of their products, and how much of that if any can be passed on to buyers. And both Atheros and Broadcom, and several others, are well established in this market.