Other Articles: WiMAX Subscriber Units, Motorola Canopy Prizm 2.0, WiMAX Networks Management, Carrier Class Triple Play, VoIP over WiMAX

Brian Jenkins,
VP Product Management
SkyPilot Networks
March 1, 2006

Wireless mesh networking has grown considerably in popularity based on its many advantages in both metropolitan and rural applications. But wireless mesh has an Achilles’ heel that is now becoming apparent to many network operators: most mesh networks fail to scale as promised. At a minimum, this discovery can be quite embarrassing. It can also be quite costly to remedy.

Ironically, the problem is rooted in what is normally an advantage with wireless: the ability to communicate with multiple neighboring nodes in an omnidirectional fashion. But as the number of neighbors increases—both near and far, yet still within range—the mesh inevitably turns into a mess as self-interference degrades throughput and undermines scalability.

Vendors are now attempting to address this serious issue in various ways. Using different frequencies for access and backhaul is a rather obvious requirement to mitigate interference. And some vendors have recently started utilizing multiple backhaul frequencies to overcome the half-duplex nature of wireless communications; two separate channels do indeed allow a single node to transmit or receive simultaneously. But even with these advances, the fundamental problem remains. Which means users of these so-called next-generation multi-radio mesh solutions continue to face a rather unpleasant tradeoff: improving scalability by sacrificing reliability and performance.

Why? Because the only way to minimize omnidirectional self-interference is to minimize the number of nodes within radio range of one another, in effect creating more of a hub and spoke topology. When every node is fully operational and suffers from no intermittent signal attenuation, this approach works well enough. But in the absence of these ideal conditions, there is also an absence of a robust, self-healing mesh topology to come to the rescue. Just a single node failing or going “deaf,” for example, causes one of two problems: the nearest neighbor is so far away that the link modulates at too slow of a data rate; or, worse yet, there is no neighbor available to backhaul the traffic.

To overcome the limitations inherent in wireless mesh networking, SkyPilot advanced the state-of-the-art in two related ways: antenna sectorization and mesh protocol synchronization. The first utilizes an 8-way antenna array that is capable of creating point-to-point links with neighboring nodes, allowing transmissions to be made at much higher power levels than are permitted with omnidirectional antennas. Higher power levels provide a number of advantages, including lower error rates based on an improved signal-to-noise ratio, better penetration through obstructions that attenuate signals, and a much longer range of up to 10 miles (16 kilometers) between nodes.

The second innovation involves use of a WiMAX-like Time-Division Duplex (TDD) protocol to synchronize all transmissions throughout the mesh topology. TDD is a bi-directional extension to the time-tested Time Division Multiple Access (TDMA) protocol employed in cellular phone networks to solve the very same problem. Mesh-wide synchronization requires a common timing source, which is provided by a Global Positioning System (GPS) receiver built into every SkyPilot infrastructure node. The accurate, common clock allows the TDD protocol to coordinate simultaneous transmissions on multiple links throughout the mesh (all being out of directional range of one another, of course) with great precision.

Together sectorized antennas and synchronization maximize throughput by minimizing self-interference to break through the performance and scalability barriers encountered in previous generations of wireless mesh solutions. And a growing number of industry analysts, users and vendors are now beginning to realize that it is, in fact, impossible to create a scalable, high-performance mesh topology without the spatial and spectral reuse afforded by these twin capabilities.

About the Author
Brian Jenkins is Vice President of Product Management at SkyPilot Networks. Brian has more than 15 years of experience in marketing and selling communication networking products. Brian has held various product management, product marketing, and sales positions with AirFlow Networks, Turnstone Systems, Ascend Communications, and Whitetree. He holds BS degrees in electrical engineering and computer science from Duke University and an MBA from the University of California at Berkeley.

Back to Newsroom