Sorting through the mesh mess
In recent years, wireless mesh networks have become a hot topic throughout the communications industry, permeating every sector from consumer-driven services to mission-critical applications. Promises of flexible, survivable wireless networks justifiably have attracted the attention of network designers, but the term “mesh” is being attached to myriad products with notably different characteristics.
Of course, all mesh architectures are based on the notion of transmitting a signal from its source to its destination point through a network of routers. But the way this is achieved in the wireless arena can vary substantially.
This reality can be a source of confusion and frustration to communications professionals, particularly as they begin the evaluation for making a purchasing decision. The good news is that the wide choice of mesh solutions on the market means at least one of them should fit the price/performance needs of users — particularly for those entities willing to be creative in funding the project (see story on page 88).
Most wireless mesh solutions involve fixed infrastructure, typically nodes or access points — nodes with direct access to high-capacity backhaul — located on buildings or utility poles. While receiving signals from mobile devices, the nodes remain stationary and act as a router, forwarding the signal elsewhere in the network so that it reaches the backhaul pipe as quickly as possible.
Crucial to the performance of the network is how quickly the signal can reach the backhaul pipe. Although nodes acting only as wireless routers can increase a network’s resiliency and coverage, the trade-off can be significantly reduced throughput in a single-radio system, said Rick Rotondo, director of marketing for Motorola’s mesh-networks product group. A single-radio network with a 16 MB/s data rate at an access point can see a 70% degradation after one “hop” through a wireless router and a 90% throughput loss after three hops.
Indeed, throughput limitations exist for shared radio systems, particularly where multiple users are sending and receiving significant information while accessing the same node, said Stephen Rayment, chief technology officer for BelAir Networks.
“As you add more nodes to those shared architectures, the throughput decreases,” he said. “But they’re lower cost and a little easier to deploy.”
With this in mind, many vendors use wireless nodes that include more than one radio. In this architecture, one radio — typically operating at 2.4 GHz, where client devices are least expensive — can be dedicated to transmitting signals while another is dedicated to handling backhaul duties.
However, Tropos Networks, one of the most successful vendors at deploying wireless broadband networks, disputes the notion that multiple radios are needed in each node.
“We see a lot of competitors who are throwing radios at the problem,” said Bert Williams, Tropos’ senior director of marketing. “We have concentrated more on the software. … We’re really the only people who have designed a routing algorithm.”
Echoing this opinion is Ellen Kirk, Tropos’ vice president of marketing, who cited the company’s software investments and noted, “It’s really not about how many radios you have, it’s how smart your radios are.”
And a properly designed network also limits the number of hops needed to any given backhaul access point, Williams said. Network designers trying to avoid multiple hops in the network do not need a 2:1 or 3:1 ratio of nodes to access points — a 10:1 ratio in a correctly designed network typically can prevent the need for more than one hop to the backhaul point, he said.
However, to optimize performance, many vendors have opted to pursue the multiple-radio route, noting that using separate radios for client communication and backhaul functions reduces the chances of a bottleneck occurring, such as when a single-radio simultaneously tries to send and receive significant amounts of information.
Rayment said his company uses multiple antennas at each node to sectorize cells, which enhances capacity and produces a more focused signal. Instead of a shared-radio architecture, BelAir employs a switched, point-to-point, architecture that lets the network perform better when multiple hops are needed (see figure on page 50). In addition, a 5 GHz backhaul link is used to complement the 2.4 GHz connection to client devices, he said.
There are other ways to achieve a similar result. For example, Alvarion does not provide mesh access directly but often provides the backhaul to mesh networks, said Patrick Leary, Alvarion’s assistant vice president of marketing. Leary noted that the 4.9 GHz spectrum dedicated to public safety is especially useful for this function because public safety doesn’t have to share the band with commercial operators, and it is the only licensed band for public-safety multipoint systems.
He added that the company’s most cost-effective solution utilizes the 4.9 GHz band for backhaul and an Alvarion 900 MHz radio — operating in the unlicensed ISM band — at access points to deliver traffic to a mobile radio typically mounted in an emergency vehicle. The vehicle then is used as a Wi-Fi hotspot, enabling personnel around the vehicle to utilize plentiful and relatively inexpensive Wi-Fi devices, he said.
“We can do those kind of projects for about one-tenth — or better — than the cost of doing mesh,” Leary said. “At the same time, we give all the benefits of commercial, off-the-shelf abilities without all the interference risks of having a metro-scale Wi-Fi [network].”
A less common architecture involves ad hoc meshing, which lets devices route traffic among themselves, whether they eventually connect to a network or not. To date, ad hoc — or peer-to-peer — meshes have been used almost exclusively by military and emergency-response users, which are more likely to require a communications path when traditional infrastructure is unavailable.
For those customers wanting ad-hoc meshing capability, BelAir has partnered with PacketHop, which provides a software-based ad-hoc meshing solution that adds only milliseconds of delay as a signal is routed through each device on the mesh, said PacketHop CEO Michael Howse.
“When you do multihop, you do introduce a millisecond per node, but with data, it’s completely imperceptible,” Howse said. “Where it does become perceptible is with voice over IP.”
CoCo Communications, another software-based ad-hoc mesh company, has announced contracts with the U.S. Coast Guard, Dallas Love Field airport and schools in Seattle and Virginia that operate over existing infrastructure. Like PacketHop, the CoCo solution is frequency-agnostic. By using the company’s radio gateway and software, a CoCo network with normal backhaul can be used as a VoIP interoperability solution, said Beth McRae, CoCo’s vice president of channel sales and marketing.
Perhaps the most-deployed mesh-networking product is Motorola’s MotoMesh. Featuring two 2.4 GHz and two 4.9 GHz radios in each node — with one radio at each frequency supporting the company’s proprietary multiservice enterprise access ad-hoc radio solution — MotoMesh provides both fixed and ad hoc meshing capabilities in a single package, said Motorola’s Rotondo.
“Right now, we’re the only network that does infrastructure meshing and ad-hoc meshing,” Rotondo said. “The hard part is not doing one thing or the other. The hard part is doing both.”