Is Mesh a Problem for MuniFi?

15 Comments

The CEO of Israeli WiFi software equipment company InspiAir, Tamir Galili, says the inherent limitations of WiFi mesh technology are behind some of the slow MuniFi rollouts in U.S. cities — a lot of the test pilots that these cities are doing are failing, and mesh is just not suitable, he says.

Is he just trying to sell InspiAir’s alternative WiFi software and hardware solutions, or is there any truth at all in his assertion? The poor quality of some MuniFi networks is something which is starting to be discussed, but we always thought it had to do with the limits on WiFi and not the mesh architecture. If there are issues with the mesh architecture itself, that could end up being a startling upset for companies like Earthlink and Tropos, as well as cities that have already committed to these companies. That is if there is any truth in this.

Though, Galili is pretty vague on the details of how his technology is better. He says that the company’s proprietary software for WiFi equipment provides an optimized signal that beats out products like Tropos’ on range, and a better use of voice and video. When I asked for more of an explanation of the technology he pointed me to the company’s white paper — thanks, more jargon.

Galili says the company has WiFi networks mostly in Asia and Europe, including the city of Helsinki and a hotspot in Manhattan. When I asked him about the limitations of mesh vs WiFi, he said “the “Mesh people” can blame the WiFi, however, as I told you, those “WiFi limitations” do not exists for InspiAir users.”

Glenn Fleishmann at WiFiNetNews has been pretty skeptical of the company, and questions the physics behind their claims. This story in Techworld calls their technology “WiFi Black Magic,” though I’m not sure if that is good or bad. We asked WiFi analyst Craig Settles who specializes in Municipal wireless networking what he’d heard about the limitations of mesh itself. He says:

WiFi mesh is indeed limited in certain respects. It’s not great for indoor coverage, it is susceptible to interference by devices as basic as microwave ovens, trees and buildings, which can block the signals. The couple of companies that have supposedly better products that overcome these shortcomings are not ones that have made headway in the marketplace.”

We called Tropos’ Director of Marketing Bert Williams, looking for the counter argument. He said, “Yeah, I’ve heard they’ve (InspiAir) been dissing mesh. WiFi, whether it be regular or mesh, has the same limitations that the Internet has on sharing bandwidth. You have to layer a quality of service over it.” Basically, any limitations are not mesh related, he says.

That doesn’t hide the fact that many publications have written about the potential limitations of Tropos’ equipment recently, and some have complained about spotty reception that marks Google’s Mountain View municipal wireless network. Google says they are happy with the Tropos product. Williams won’t comment on the Mountain View network but says there are always a certain amount of complaints about communications services.

Galili said the U.S. is the most advanced in the world when it comes to MuniFi deployments, so I asked him why then is everyone in the U.S. betting on mesh. He says mesh is the incumbent technology and InspiAir is new to the market. That is one strong statement — can he really back it up? Or is he just an easily quotable executive with a good PR team?

15 Comments

¥Ñ¥Ö¥ê¥·¥Æ¥£ÂåÍýŹ¤Î´ë²è¥Þ¥ó¥Ö¥í¥°

Éð»Î¤Î°ìʬ¡¡£±£´ºÐ¤ÎÊì¡©£±£´ºÍ¤ÎÊì¡©…

Éð»Î¤Î°ìʬ¡¢ÏÃÂê¥é¥Ã¥·¥å¤Ç¤¹¤Í¤§¡£¾ù¤é¤Ê¤¤¿´¡¢¾ù¤ì¤Ê¤¤°¦¡£¡Ö¤Ê¡¼¤Ë¸À¤Ã¤Æ¤ó¤À¤«¤Ã¡×¤Ã¤Æ¸À¤ï¤ì¤Á¤ã¤¤¤½¤¦¤Ê¥­¥ã¥Ã¥Á¥³¥Ô¡¼¤Ç¤¹¤¬¡¢…

¥Ñ¥Ö¥ê¥·¥Æ¥£ÂåÍýŹ¤Î´ë²è¥Þ¥ó¥Ö¥í¥°

¤Û¤ó¤¸¤ã¤Ë¡¡¼«Å¾¼Ö¾¯Ç¯µ­É¬¸«¡ª¡©´ØÀ¾¥¸¥ã¥Ë¡çÃËÁ°¤ä¤Í¤§¡ª…

¤ä¤Ã¤È½ÐÄ¥¤«¤éµ¢¤Ã¤Æ¤­¤Þ¤·¤¿¡£¥Ñ¥Ö¥ê¥·¥Æ¥£¤â»þº¹¤¬¤¢¤ê¤Þ¤¹¤¬¡¢ÆüËܤαüÃϤؽÐÄ¥¹Ô¤¯¤ÈÅÔ²ñ¤È¤Î»þº¹¤¬¤Ç¤­¤ë¤Î¤Ç¡¢¥Ö¥í¥°¤â¤«¤±¤Þ¤»…

serhiy

have you, guys, checkout meraki.net (a spin-off the MIT’s rooftnet project) ?

it looks very promising, though to commercial installations yet.

PhoneBoy

InspiAir claims they can cover a wider area with less access points and better speeds. Even if the antennas are more expensive than the competition, a smaller network with fewer nodes has to be easier to manage and maintain than a network with significantly more nodes. That will reduce overall operating costs, which is a good thing in any business.

Even if “mesh” is a problem or not for MuniFi, the solution provided by InspiAir certainly is impressive and bears investigation.

While there may be some marketing hype there, I’ve seen InspiAir’s solution work with my own eyes.

Read more on my blog:

http://www.phoneboy.com/node/1095

Tom Sharples

Qorvus’ two largest mesh installations each have over 500 permanent paying subscribers (many equipped with inexpensive indoor CPE), and they’re getting excellent performance. But as Glen Fleishman points out, these large installation are actually hybrid deployments using the built-in meshed or point-to-multipoint backhaul on e.g. 900 Mhz or 5 Ghz, with single-radio meshing taking place only at the end-points of the network, and usually with just a few nodes. This kind of deployment plays well to mesh technology’s greatest strength- the ability to create a free-form layout to cover the last-mile with high RF density, while still supplying acceptable bandwidth to end-users. There’s no free lunch with mesh, and Tropos’ legacy installations of many nodes with insufficient backhaul may be fine for itinerant users or tourists, but I doubt they’d be suitable for long-term subscribers.

Stu Browne

WiMax is what municipalities will be deploying in the near future if the FCC frees up some of the 700 mhz UHF TV bands for non-licensed use. WiMax offers true QoS where Wi-Fi(point to point or Mesh) does not. WiMax in a point multipoint configuration can obtain larger coverage at higher data rates than any Wi-Fi network — Mesh or otherwise. Wi-Fi is an indoor, limited distance technology which was never designed as a WAN. WiMax was designed from the ground up as a WAN technology.
Municipalities deploying WiFi are throwing money away.

David Young

When you get past InspiAir’s marketing hype, they claim to improve WiFi throughput and range with a software-only solution that is embedded in the access points, which are interconnected in some way. It works with the WiFi notebook or PDA you have today. This is not implausible on its face. People can (and do) improve WiFi performance with innovative architectures and software-only tweaks.

I would caution potential investors that there is nothing new under this sun, and even if InspiAir’s technology can live up to their claims, it may not be new and patentable. Their vague technology whitepaper faintly reminds me of techniques that you can read about in “the literature,” such as frame combining, multiple-radio diversity, and virtual access points. What I am getting at is that InspiAir may use well-known techniques that are not merely “unmagical”, but they may be easily reproduced by a more established competitor. (I am thinking here of Tropos, SkyPilot, but also CUWiN.net.)

That said, InspiAir sounds like hot air to me!

Chris

At the end of the day, this is fundamentally an RF carrier class engineering problem to have reliable service. As Glenn points out, “BelAir, Strix, and SkyPilot, to name three all use point-to-multipoint or dedicated directional point-to-point backlinks on every node. The nodes are each autonomous. (BelAir does offer a single-radio mesh node; I don’t know what the uptake is on that; it was a later addition to their architecture.)” This is the only way to make it work. Check out what Wireless Evolutions has done in San Diego for a project that is working. This project recently won one of InfoWorld’s Top 100 Award fro 2006.
http://www.infoworld.com/article/06/11/13/46FE06iw100telecommunications_1.html

Tony Li is rumored to have left Tropos this week, maybe it was clear to him that this is not a problem that routing black magic can solve.
http://gigaom.com/2005/09/16/tony-li-has-left-the-cisco-building/

Mike

Whatever the definition, the data carrying capacity of the backhaul has to be at least the same than the mesh can provide to the endpoints, right? So, there is no point having a huge mesh with only one backhaul that has limited capacity, or making the mesh so large that each hop adds latency up to an unbearable point.

WiFi is not a technology designed to penetrate buildings, and as such, the problem will not be the size of the cloud, but its ability to reach the people that are being asked to use it. In my experience, the network in San Jose built with Tropos equipment did not work. We got an IP address but that was it. Where the problem was – I do not know, maybe it wasn’t even Tropos’ problem, but the reality is that the system did not work.

Whatever this company claims, there is no way to improve the physics of a radio signal using software. Even more so if you are using COTS hardware. Are they using a leaf-sync algorithm to make the RF pass through foliage? Maybe something to do with subspace fields? Every time I hear of a new, revolutionary way to break the laws of physics, my first thought is: ‘next, they will be asking for advance fees in exchange for territorial agreements with potential partners and distributors’. As it is happening with xMax, it will probably happen with InspiAir. Will they deliver? History usually repeats itself.

Glenn Fleishman

This mesh discussion is tricky because most “mesh” companies aren’t really using mesh. Tropos recommends very small clusters of mesh nodes; I’ve been hearing 3 to 8, with 4 to 6 being more typical. Each cluster operates as an autonomous unit with separate backhaul (injection points). Cisco also offers actual mesh, as I understand it, using Redline’s 5 GHz “WiMax” (uncertified, but apparently conforms to profile) for injection.

BelAir, Strix, and SkyPilot, to name three all use point-to-multipoint or dedicated directional point-to-point backlinks on every node. The nodes are each autonomous. (BelAir does offer a single-radio mesh node; I don’t know what the uptake is on that; it was a later addition to their architecture.)

My definition of mesh may be too tight. I see a mesh network as a set of nodes that use the same frequencies and a single radio to communicate among each other, whether it’s true 802.11 for their intra-node communication or not using primarily omnidirectional or sectorized antennas for reaching clients on the network.

Once you have an injection point that feeds into something that’s like a switched backhaul, it’s a different architecture because you gain enormous frequency reuse over space and time based on the switching and routing algorithms for those backhaul radios.

Rick

Green Packet has a mesh solution called SON (self-organizing network). The company is out of Malaysia. I don’t know much about their technology, but it’s supposed to have good QoS.

Marty Hahnfeld

Please… all this from a company with exactly ZERO relavant experience in the space.

The companies leading in this space (including, my company SkyPilot… ding) have invested heavily to deliver solutions which are truly capable of quality outdoor border-to-border Wi-Fi service delivery. When somebody essentially says “those mesh guys are wrong”, they’re taking the opposite side of $100’s of millions in investment and 100’s of very friggin’ smart people.

This is not easy stuff, and it would be no easier for InspirAir than it is for the rest of us. The complexities far exceed marketing claims, and the leaders in the space are delivering REAL SOLUTIONS for these networks.

The answer to all “black magic whiteboard solutions” like this should be… go raise some money, win some deals, build some networks, and prove that you’re right… else, spare us. The scrap pile is filled with companies and pitches like this.

Comments are closed.