Which Mesh is Best for wVoIP?
Strix Systems' Cyrus Irani discusses the need for multiple radios in wireless mesh networks.
As wireless mesh networks grow in popularity -- with new public and private deployments announced almost daily -- they are also growing both in scope and complexity. When multiple hops are involved in a WiFi mesh covering large geographic areas, problems such as bandwidth degradation, radio interference, and network latency can arise. These problems wreak havoc on real-time service offerings, such as VoIP, which are high on the wish list of WiFi providers. Multi-radio, multi-RF, multi-channel WiFi mesh networks can solve each of these problems, providing a truly high-performance WiFi network that is optimized for real-time services like voice.
Why Mesh
While WiFi has clearly established itself as networking technology for data applications, there are still some inherent problems that need to be solved when deploying wireless VoIP (wVoIP). WiFi mesh networking addresses the shortcomings of WiFi and makes the planning, deployment, and operations of such networks significantly more cost effective, especially in hard-to-wire environments and/or metro scale deployments where wiring hundreds of nodes across tens of square miles is not economically feasible. A WiFi network that supports wVoIP in addition to data benefits even more from a mesh-based architecture due to the increased coverage and throughput density.
Four Requirements of a Mesh:
A mesh infrastructure must be able to deliver high throughput, low latency, and end-to-end quality of service not only between the wireless handset and access points, but also across the mesh links to the wired termination point. As such, the mesh backbone must provide:
High throughput across multiple hops
Regardless of the number of hops on the mesh, which is usually between 3 and 10, the mesh backbone must be able to support the traffic load. The ability to support high throughput equates directly to the number of voice and data users the system can support, so inadequate bandwidth across multiple hops results in unsatisfactory user density and requires additional equipment and a greater number of wired termination points within the network.
Low latency across multiple hops
To avoid such issues as jitter, each hop must minimize the packet latency. The holding time of a packet at any node in the mesh must be minimized and equivalent to wire line performance. As such, ideally, the packet should be forwarded even before all of the information is received from a previous node. Movement of packets across the mesh must be asynchronous in nature when any type of routing coordination is required.
End to end QoS - packet prioritization of voice
To deal with contentions and spontaneity of load demand, voice must be prioritized across the mesh backbone and terminated on a switch with traffic prioritization. This prioritization needs to be automatic, and handled well via separate VLANs/SSIDs for voice. 802.11e can not be counted on in the near future. It is no longer adequate to provide a class of service just between the wireless handset and the device-serving radio. Mesh introduces the requirement of QoS across the entire backbone to avoid any contention that can occur at each hop in the mesh.
Layer 2 switched network
Layer 2 networks minimize the roaming problem that occurs between layer 3 networks. Layer 3 networks also require careful planning around different type of higher level protocols. Both these factors contribute to performance issues and installation issues.
Each of these four parameters directly affects scalability (in terms of number of users and network coverage) and voice quality. If a multi-hop topology does not provide for these requirements, it lacks voice-caliber functionality.
Approaches to Wireless Mesh
Wireless mesh approaches vary, but most have their technology roots in the original concept of the Wireless Distribution System (WDS). WDS is a wireless AP mode that uses wireless bridging (where APs communicate only with each other and don't allow wireless clients to access them), and wireless repeating (where APs communicate with each other and with wireless clients). It is intrinsic to all mesh networks that user traffic must travel through several nodes before exiting the network (for example, to the wired LAN).
The root of the throughput and latency problem is due to 802.11 being a half-duplex technology, meaning it can only perform one function (either send or receive traffic) at any given time. As such, the number of radios dedicated to wireless handsets and the mesh backbone, along with their function for a particular topology, is vital and directly affects the 4 factors mentioned above.
The Single Radio Approach-Everything on the Same Channel
The single radio model is the weakest approach to wireless mesh as it uses only one radio (channel) per node to handle all three functions: client access, backhaul ingress, and backhaul egress. As more APs are added to the network, a higher percentage of the radio bandwidth is dedicated to repeating backhaul traffic, leaving very little capacity for wireless clients because wireless is a shared medium. Also, a node cannot send and receive at the same time, or send when another AP within range is transmitting, which introduces intolerable latency after just 3 hops. This contention for the available shared bandwidth is based on Ethernet-like collision avoidance rules for wireless (CSMA/CA).
This means that, in a single-radio mesh architecture, one radio must constantly switch from performing backhaul ingress, backhaul egress, and client access - which introduces significant latency.
Simple math shows that only limited throughput is possible per wireless client for the single-radio approach, regardless of the routing protocol used. For example, if you have 5 APs with only 20 wireless clients connected to each AP, with all APs and clients sharing the same 802.11b channel (5 Mbps), that equates to less than 50 Kbps per user-worse than a dial-up connection. And since all wireless clients and APs must operate on the same channel, network contention and RF interference results in unpredictable latency.
The Dual Radio Approach-Sharing the Backhaul
With the dual radio approach, one radio is dedicated to wireless client support while the other radio is dedicated to wireless backhaul (mesh) support-with the backhaul channel being shared for both ingress and egress traffic. Since a dual-radio approach provides a separate radio for both client access and backhaul, this relieves some of the client-side congestion (low throughput, low latency) but the backhaul mesh channel must be shared for both ingress and egress traffic because the backhaul radio must still constantly switch from performing backhaul mesh ingress and backhaul mesh egress, offering minimal improvement to the backhaul bottleneck with the same latency issues across the mesh.
The Multi Radio Approach-A Structured Wireless Mesh
In the multi-radio or "structured mesh" approach, there are at least three radios per network node -- enabling dedicated radios for client traffic, backhaul ingress traffic, and for backhaul egress traffic. This approach offers significantly better performance than either the single- or dual-radio approaches as it allows for dedicated mesh backhaul links that can transmit and receive simultaneously because each link is on a separate channel.
Because the three functions of client ingress, backhaul egress, and backhaul ingress are handled by dedicated radios:
Summary
In order to meet the demands of real time communications applications, like wVoIP, a WiFi mesh network requires a multi-radio, multi-RF, and multi-channel architecture. Multi-radio architectures cost-effectively provide the necessary capacity and coverage for high-throughput, low-latency, and high-priority voice traffic across multi-hop mesh networks by providing dedicated nodes for client access, mesh backhaul ingress, and mesh backhaul egress.
Cyrus Irani, VP of marketing and strategy at Strix Systems.