Wireless Mesh Passes Test

Light Reading today published the most comprehensive test report on wireless mesh networking equipment to date, performed by the testing lab Iometrix Inc. The results indicate that mesh hardware is ready for prime time, and that multiple radios in a mesh access point vastly improve network performance. (See Wireless Mesh: Ready!) Dozens of companies were invited to participate in the test, which was run by Bob Mandeville, president and founder of Iometrix. Half a dozen responded to the invitation, but only two companies ultimately put their products through the ringer -- mesh startups Firetide Inc. and Strix Systems Inc.

In the test, Iometrix used Azimuth Systems Inc. 's 800W wireless test platform, which uses a highly-controlled RF-shielded environment that guards against "the sundry aleatory effects of noice, interference, and environmental elements," according to the report. Mandeville notes that the RF shielding added complexity to the test but makes the results very clean.

Service providers Light Reading spoke to about the test say the results are indeed impressive, but they also note that vendors will need to keep costs down to develop the market.

While Strix and Firetide serve different markets, both companies' products employ a dedicated 802.11a radio for backhaul connectivity; Strix actually uses several. (Market leader Tropos Networks Inc. , which declined to participate in the test, uses a single radio for both backhaul and client connections.)

The mesh test, located at a test bed in South San Francisco, was based on the Institute of Electrical and Electronics Engineers Inc. (IEEE) 802.11 Task Group T, which specializes in tests. It garnered some surprising results. For example, tests of the Strix system disproved the commonly held belief that throughput rates will decrease as the number of hops from node to node increases. Instead, the tests showed that a Strix Access/One OWS 2400 Hardened Outdoor Mesh Node maintained a constant maximum throughput of 35 Mbit/s over one hop or four.

"From a platform experience, obviously multiple radios offer more scaleability across multiple hops, without a doubt," says Frank McCarthy, CEO of CitiWiFi Networks, a Florida municipal WiFi service provider and a customer of Strix.

Strix access points sport up to six radios for backhaul and client connectivity, and Iometrix attributes the number of radios to the product's performance.

Strix's nodes can backhaul up to 36 high-quality voice calls, again, regardless of the number of hops, the test showed. That number drops to 23 when there is a 50/50 mix of voice and data.

Firetide submitted its HotPort 3103 indoor mesh node, which provides wireless backhaul connectivity only, meaning it needs to be attached to third-party access points in order to connect to clients. Tests showed that backhaul throughput dropped from 30 Mbit/s to 8 Mbit/s over four hops, but the test report notes that this isn't as big a deal for enterprise application as it is for a carrier-class municipal network.

"The overall conclusion to be drawn from testing Strix's and Firetide's equipment is that wireless mesh technology is ready for use in telecom and enterprise environments," the report says.

Again, while this was arguably the most comprehensive third-party wireless mesh test to date, most of the companies in the industry declined an invitation to participate. Notable decliners included Cisco Systems Inc. (Nasdaq: CSCO), Nortel Networks Ltd. , and Motorola Inc. (NYSE: MOT). Market leader Tropos Networks Inc. met with Iometrix, but decided not to submit its single radio solution, citing reservations about the methodology.

So why did so many people decline to be tested? Reasons ranged from a "lack of resources" to a disagreement about the methodology.

"We looked at the test, and to us it was a speeds-and-feeds indoor test that didn't have a lot to do with what really happens in the real world," says Saar Gillai, vice president of engineering at Tropos. "It is generally a fact that in a controlled environment, more spectrum is more capacity. In a real-world environment, that's only true if you use the spectrum effectively... We're able to do more with one radio than others do with multiple radios because we use the spectrum effectively. In a lab you're not going to see that."

CitiWifi's McCarthy says that Tropos has a point about interference issues. "Nobody's come out with a magic bullet to make 5.8GHz penetrate foliage," he says, referring to the spectrum range where 802.11a operates. To that end, he always performs outdoor tests of prospective products. "Indoor tests won't tell us anything.

"But a multi-radio node will take something from a point of presence and distribute it farther," he notes. "It's basic physics... If your network isn't designed to scale easily, you're going to miss out on revenue. That's where the multi-radio mesh node makes lots of sense." BelAir Networks Inc. also initially agreed to be part of the test. But a month into the preparatory phase of testing, the company pulled out, citing resource constraints. Officials are apologetic.

"We left him high and dry," says Sheila Burpee Duncan, a spokeswoman for Bel Air, who says that the company just didn't have enough product managers to oversee the testing and serve customers at the same time. "We ran into resource constraints on our side... A number of deals started to pay off at once, and priorities were immediately rejigged in favor of some key wins and strategic partner engagements."

So what's this all mean for the wireless mesh vendors -- and the market? McCarthy says the performance is nice, but cost per square mile has become the deciding factor in choosing mesh equipment. "With these municipal networks, a lot of the RFPs want the networks at no cost to the city, so the cost per square mile is going to be a major decision factor. The vendor that understands the cost per square mile is going to win, whether they have great throughput or not." — Carmen Nobel, Senior Editor, Light Reading

Be the first to post a comment regarding this story.
Sign In