Is multi-radio Wi-Fi mesh worthwhile?
When deploying a mesh network, a system integrator can choose between single-radio or multi-radio mesh. Which one is better? Making that decision is not as simple as you might think. When deploying Wi-Fi mesh networks, system integrators can choose among single-radio or multi-radio mesh node equipment. The single-radio models, such as what Tropos offers, uses a solitary 2.4GHz radio for connecting user client devices to the mesh network and interfacing mesh nodes together. This requires both users and node-to-node traffic to share a single radio. With single-radio systems, the radios in neighboring mesh nodes are set to the same RF channel; otherwise, the mesh nodes can’t communicate with each other. With a common channel, a user device transmitting data to a mesh node precludes the mesh node from transmitting data to an adjacent mesh node (and vice-versa).
Other vendors, such as Cisco and Strix, offer multi-radio mesh nodes. In the multi-radio case, one radio in the mesh node interfaces users to the mesh network, and an additional radio, operating at a different frequency, provides communications between mesh nodes. This allows traffic between a user and the mesh node and between the mesh node and neighboring mesh nodes to occur simultaneously.
Which approach is better? One would think that more radios means better performance, and that’s generally true. Keep in mind, however, that the single-radio hardware is significantly less expensive as compared to multi-radio versions. This should encourage you to consider the tradeoff between performance and price. If the potentially lower performance of the single-radio system is sufficient to meet requirements, then the single-radio system is the one that you should strongly consider. If requirements dictate higher capacity, then look closely at whether a multi-radio system is really necessary and cost-effective. This just offers some broad direction. Of course you should also think about other parameters, such as security, operational support, and support for newer technologies.
- - - - -
Loading ...
You should look further into your claims of cost difference for single vs. dual radio solutions. Proxim’s AP-4000 mesh solution for instance costs 30%-50% less than the Tropos single radio solution.
Certainly Bel-Air, Cisco, and Strix are more costly solutions, but there are some great dual radio solutions out there by other companies that easily rival the cost and obviously performance of single radio mesh solutions on the market today.
Actually buffalo has a dual radio product WHR-HP-AG108 which retails for $70 and has been ported to DD-WRT. Adds peanuts to cost of a network.
Why didn’t Mr Geier simply provide List prices on a One Radio Tropos Node, a 2 Radio Cisco Node and a 4 Radio Strix Node. One would find very quickly that the List prices on the 3 are very similar.
I just love the Spin these vendors (supported by supposed experts)place on their deployments emphasizing the fact that the more Nodes/AP one deploys in an network the better the service or throughput. This appears logical but very misleading.
What we need is simple Cost/Meg/Node deployed figure.
What is better: A Tropos or Cisco network that requires a Gateway (injection of bandwidth) every 3 or 4th Node with effective throughput of 3-6 Mbps on the 3rd Node or a Strix Network of a Gateway every 6-8 Nodes and an effective throughput of 10-14 Mbps after the 5th Node.
What these expert analyst can do to help the Muni and Service Providers researching which vendor to select for their Metro Mesh Nets is come up with some simple guidelines and performance targets the Muni can incorporate in their RFP with a Performance guarantee (Bond)tied to delivery.
Jacomo
There is much confusion on what multi-radio really means. A more accurate terminology would be first, second and third generation mesh products.
There is a vast difference in performance between the three generation of products, please see http://www.meshdynamics.com/third_generation.html for more.
Regarding costs, overall costs are the issue, not the cost of the nodes. The running costs of more Ethernet backhauls needed with first and second generation solutions adds up as the deployment sizes increase. Further, if latency is an issue with moving Voice or Video, especially over multiple hops. So more Ethernet links are needed to reduce the number of hops.
These are some of the reasons why our customers are replacing competing first and second generation products with our third generation 3 and 4 radio boxes. Our customers get 18-22 Mbps TCP/IP over 10 hops. Previous generation products (that were replaced) were barely able to provide 1-2 Mbps over 3 hops.
Francis daCosta
Single radios also = more backhauls if you want the same amount of bandwidth. More backhauls = more network entry points = more equipment = more workload/power/software/warranty/upgrade/replacement charges. Multiply all the mores times the number of square miles you want covered and by the guaranteed uptime and the multi-radios are a steal. Yes I have spreadsheets that have haunted my dreams at times.
As far as design, from what I have seen, BelAir seems to have put the most engineering and forethought into their hardware - but I have not looked at any hardware in a while now so things could have changed.
There is a reason Tropos started making multi-radios too! Sidestepping answering the single radio questions at the conferences got to be a pain.
There are only three reasons to prefer multi-radio mesh to older single radio mesh technology: better performance; lower total cost of ownership; and better support for future applications. Other than that, single-radio mesh is fine.
Independent tests have shown 50X-1000X performance gains with third-generation WiFi multi-radio wireless mesh when compared to single-radio mesh over the multiple hops typical of a large deployment. Many of MeshDynamics’ customers come to us for a multi-radio solution after purchasing and discarding a one-radio or two-radio solution. The problems in Mountain View, CA and elsewhere show the limitations of single-radio mesh.
Mr. Geier is incorrect on his assumption of unit costs. For example, MeshDynamics three-radio mesh nodes have a list price lower than some popular single-radio nodes. And total cost of ownership is less because fewer costly wired drops are needed for a mesh network built on well-designed multi-radio nodes.
Finally, emerging voice and video applications demand not only high bandwidth, but lower jitter and latency. Multi-radio mesh nodes substantially reduce the contention found in single-radio mesh networks. Reduced delay and jitter and makes voice and video work better, providing application “headroom” to support future needs — avoiding costly upgrades.
Even the industry leader in single-radio mesh has acknowledged its shortcomings by adding a two-radio product — so why are we still having this conversation? Customers have begun the migration to third generation multi-radio mesh and it will only accelerate from here.