Mesh Networks: Disruptive Technology?
January 25, 2002
Disruptive doesn't imply bad in this case. Mesh networking is an adjunct technology that would only serve to increase the robustness of 802.11. Take a look at how this cool technology is evolving.
MeshNetworks Inc. (www.meshnetworks.com) aims to turn the wireless world on its ear.
In its vision of WLANs and wireless access networks of the very near future, MeshNetworks sees every client device also becoming a relay point or router for network traffic.
One immediate benefit is that such networks can in effect see around corners. Even line-of-sight network technologies like 802.11 can become non-line-of-sight - almost overnight if MeshNetworks can deliver what it's promising.
And the next-generation networks the company is building will also power mobile broadband services.
The Maitland FL-based network equipment maker is so sure it's going to turn things topsy-turvy that it actually has a director of disruptive technologies, Rick Rotondo.
Rotondo's title, he explains, is derived from Clayton M. Christensen's 1997 business book The Innovator's Dilemma, which tried to show established companies how to withstand the impact of revolutionary technology from upstarts like, well, MeshNetworks.
The definition of a disruptive technology?
"In real-world terms," Rotondo says, "it has to meet at least two of three criteria: be ten times cheaper than any alternative, have ten times higher performance, and ten times higher functionality. All three is best."
There are two parts to MeshNetworks' supposedly disruptive technology.
One is QDMA (quad-division multiple access), a proprietary radio technology developed for and currently used by the military.
MeshNetworks has enhanced the technology and is now commercializing it under an exclusive license from its original developer, White Plains NY-based ITT Industries (www.itt.com). ITT is also an investor.
QDMA's most notable characteristics are that it's IP from end to end and supports high-speed mobile broadband access and infrastructure-free "ad hoc peer-to-peer networking."
The company claims it can deliver up to 6 Mbps to each user in a QDMA wireless network.
The technology also has built-in GPS (Global Positioning System) capabilities and, in MeshNetworks' implementation, QoS (quality of service) for IP voice and video.
The ad hoc peer-to-peer part means that nearby clients in a QDMA network can talk directly to each other over the air rather than going through network access points and routers. (This is a capability also possible with both Bluetooth and 802.11.)
QDMA could work in any sub-10GHz frequency band. MeshNetworks has implemented it first in 2.4GHz and has already built prototype routers, relays and PDA-size client devices. Next it's developing equipment for licensed MMDS (2.5GHz) operators.
The company is using the prototype 2.4GHz gear in a five-square-mile QDMA-based test network around its headquarters in Maitland, a suburb of Orlando.
It now has an FCC experimental license to build a much bigger nationwide 4,000-node test network. This time it will use the next generation of QDMA client devices which come in a PCMCIA form factor. The company hopes to start building the new network in cooperation with service provider partners later this year.
The second part of MeshNetworks' disruptive technology, and the core intellectual property the company brings to the party, is the set of tricks and techniques for mesh networking.
This is the part about turning every client in a network into a relay point or router. It's the part MeshNetowrks can and is extracting and applying to other non-QDMA network technologies, including, first up, 802.11.
The company claims that by the end of this year it will start shipping a software-only overlay solution that lets 802.11b clients in existing networks work in mesh-mode. Each client device would run the MeshNetworks software.
It hasn't worked out pricing for this offering but it will likely be no more than the volume pricing for the chipset version it is also developing, which is less than $30.
So. Is this truly disruptive technology?
"We definitely meet the ten-times cost and ten-times performance advantages," Rotondo contends.
On the third criteria, it's a question of whether you interpret it to mean ten times the value or ten times the number of features. Rotondo argues that the extras in the company's QDMA-based technology - built in QoS and GPS and support for infrastructure-free peer-to-peer networking - are ten times more valuable than any alternative."I wouldn't bet my pay check on the last one," he concedes, "but I would the first two."
Whether its technology is "disruptive" according to a definition laid out in some book most of us haven't even read is ultimately beside the point, of course. The real question is, what can MeshNetworks do for 802.11 and 802.11 users?
The software solution for 802.11 networks will do a couple of things, says MeshNetworks CTO Peter Stanforth.
While it will not add any mobile broadband capabilities beyond what 802.11b can already support, it will extend the range and link robustness of existing Wi-Fi networks by allowing mesh-style multi-hopping.
Instead of a network client communicating with the nearest access point or base station, it finds a nearby client and hops through it - or possibly multiple clients - to get to the nearest POP.
One major benefit is that mesh networks provide non-line-of-sight linking which dramatically increases network coverage. The initiating client may not have line of sight to a POP, but the next one it hops to will, or the one after that.
"There are fixed wireless providers we've spoken to who tell us that without a non-line-of-sight technology they will miss anywhere from 40 to 60 per cent of their market," Stanforth says.
Mesh networks also reduce the distance of each link. This is important because in any wireless network there is always a trade-off between packet throughput and link range.
All other things being equal, as data throughput requirements increase - as in 54-Mbps 802.11a networks - range is reduced.
Mesh networks can facilitate higher throughput without sacrificing effective range - or greater range without loss of throughput.
Furthermore, because link distances are shorter in a mesh network, there is also less inteference between clients. And that makes for much more efficient re-use of frequency.
Stanforth says the primary advantage MeshNetworks has over other technologies, especially others targeting operators using MMDS and PCS spectrum, is that because of its non-hierarchical topology, it offers far superior frequency re-use.
It comes down to how many bits per hertz per square mile you can squeeze out of the network. "We will get orders of magnitude better bits per hertz per square mile than hierarchical network systems like cellular," Rotondo says.
This also equates to the number of simultaneous users at a given throughput rate for a cell sector or POP. MeshNetworks claims it will outperform hierarchical systems on this measure by a factor of ten to 30 times.
We did think of a few possible downsides to such a network architecture, but Rotondo and Stanforth had answers.
No, they say, individual clients in a retrofitted 802.11b network will not take a performance hit because of overhead associated with establishing and maintaining the mesh topology.
There is certainly some overhead, but the intelligence in MeshNetworks' technology more than compensates for it, says Rotondo. "When you take it on balance and say, 'What's my real packet throughput?' in general we're as good as or better than [a native 802.11b network]."
Nor, they claim, is security a problem - at least from a technological point of view - even though packets destined for one client pass through another to get there.
Security features in the company's QDMA-based systems include a hardware firewall on a chip that makes it impossible for a client to access somebody else's packets.
That feature will not be available in retrofitted 802.11 networks using the software overlay technology, but there are plenty of other security tricks MeshNetworks uses, including route diversity.
In a well-populated mesh network it will be possible to send packets by different routes so that any one client-relay point will see only every second or third packet in a transmission hopping through it.
And the hardware firewall will be a feature of the mesh-enabled 802.11 gear the company hopes to get existing manufacturers to make under license.
"Security is clearly a big issue," says Stanforth. "We've spent a lot of time working through it - to the point that we do not believe it will be a significant issue [in selling the technology], beyond the fact that it's not easy to give a two-second answer as to why it shouldn't be."
Longer explanations and white papers will convince network operators, he believes. But in the case of wireless ISPs, we wonder how easy it will be for them to convince subscribers that using their devices as relay points is such a great idea.
This is a quibble, though. We think any 802.11 network operator - whether access service provider or enterprise - that doesn't watch very closely what MeshNetworks does in the next few months really ought to read that book Rotondo and Stanforth are so big on.