Cleaning Up the Mesh with Synchronization

By TMCnet Special Guest

Brian Jenkins, Vice President, Product Management, SkyPilot Networks

 

Architects of citywide WiFi networks must contend with the three “C”s of wireless: Cost, Coverage, and Capacity. The combination of these constraints is common across all wireless networks and challenge architects to cost-effectively blanket an entire city with enough capacity to provide broadband service to a high density of WiFi (News - Alert) subscribers. If done effectively, the network can provide access to a litany of WiFi devices and provide an economic return for the network operator. However, challenges abound since radio frequencies (RF) pose a number of hurdles for network architects, including limited spectrum and potential interference with other devices.

 

Fortunately, the wireless industry continues to provide innovative solutions to challenging RF issues. Over the past few years, many of these innovations have found their ways into citywide WiFi networks but have shown mixed results, primarily since most innovations address only one or two of the three “C”s. As an example, an innovative solution might provide greatly increased capacity and coverage but with the burden of greatly increased costs. They key, as always, is to find the proper mix of cost, coverage, and capacity in a way that enables the network architect to build a carrier-class broadband wireless access solution. The challenges are even greater when services such as WiFi VoIP are added to the list of requirements.

 

With citywide WiFi networks, the first guideline to address is always coverage. The network requires WiFi coverage throughout the city to allow access from standard WiFi devices such as laptops, smartphones, PDAs, and WiFi VoIP phones. The industry standard solution is to deploy high-powered WiFi access points (APs) on light poles throughout the city. This provides the basis for broad-reaching outdoor WiFi coverage and allows WiFi to extend indoors through specialty WiFi CPE devices.

 

Light poles, however, do not have a connection to the Internet, so these light pole-mounted APs need wireless backhaul. Architects then must answer a few fundamental questions: Can light pole-mounted APs be cost-effectively backhauled? Can the backhaul provide enough capacity? Can the backhaul provide carrier-class features such as QoS, fault tolerance, and redundancy?

 

Since WiFi provides service to standard 802.11b/g clients and operates in the unlicensed 2.4 GHz frequency band, the most common method of providing wireless backhaul is utilizing the unlicensed 5 GHz frequency band. Using 5 GHz for mesh backhaul obviously helps address cost concerns since it has no fees. Using 5 GHz also addresses capacity concerns since it separates access and backhaul onto different frequencies, which is critical to scaling capacity. With this as the baseline, there have been a few different approaches to backhaul light pole-mounted APs.

 

A few of the early citywide WiFi networks used a point-to-multipoint backhaul solution. In this architecture, a base station sits at the top of an Internet-connected building and acts as the central hub for wirelessly connecting to the light pole-mounted APs. This is a good solution for capacity since it uses directional antennas to provide high modulation links; however, it also suffers from several significant weaknesses inherent to the network architecture.

 

A point-to-multipoint architecture does not provide mesh capabilities and uncovers a serious lack of many required capabilities. For instance, when an AP is outside of the coverage from the base station, it is stranded since coverage can’t be extended. Also, if the potential link is blocked by an obstruction, the AP is stranded since there is no way to route around the obstruction. In addition, the base station does not provide fault tolerance, redundancy, and self-healing capabilities. If, for any reason, the base station fails or loses Internet connectivity, all of the APs lose service and do not have a way to restore service until the base station is fully recovered. All of these issues seriously limit the attractiveness of point-to-multipoint as a backhaul topology.

 

To overcome the issues with point-to-multipoint, most architects choose a mesh topology for backhaul. Mesh networks allow the ability to extend the backhaul by interconnecting backhaul nodes, thereby overcoming the deficiencies found within point-to-multipoint networks. Mesh networking also enables carrier-class features such as self-healing failover and allows the network to easily overcome obstructions by routing around them. There are two fundamentally different types of mesh networking solutions, and they are typically defined by their antennas, either omnidirectional or directional.

 

The first generation of mesh networking devices used omnidirectional antennas to communicate with the other mesh devices that surround them. With this architecture, the coverage range can be easily expanded by adding “hops” between the nodes and the gateway to the Internet. These “hops” were automatically discovered and also provide the ability to have several traffic routes, enabling mesh networks to provide best-path routing, load balancing, and fault tolerance.

 

Many network operators now, however, realize that omnidirectional mesh systems fail to scale due to the interference created by thoughtlessly spraying every transmission over all 360°, regardless of where the intended recipient is located. This inefficiency causes the mesh to deteriorate as self-interference degrades throughput and undermines scalability.

 

A mesh network of directional antennas, though, effectively addresses the capacity concerns caused by omnidirectional mesh networks. Directional antennas provide a variety of benefits, including increased link budget that results in links with better modulation/throughput. In addition, directional-based mesh networks provide significantly better spectral efficiency — instead of mindlessly spraying transmissions over all 360 degrees, the directional antennas direct their transmissions to the intended recipient. This limits self-interference and results is much greater capacity. Even directional mesh systems, though, have their limitations.

 

The main issue with directional antenna systems is that they have to be properly pointed and are static. For any two directional antenna links, an RF technician needs to manually point both antennas at each other in order to establish a connection. This increases the cost and complexity of deployment and eliminates the networks ability to automatically discover new nodes and providing self-healing failover. In addition, changes to the network are not dynamic since they require technicians to re-point antennas, and getting the nodes pointed and tuned precisely for optimal performance can become an issue.

 

Fortunately, there is a way to provide the dynamic capabilities of omnidirectional mesh systems with the capacity efficiency of directional mesh systems. A dynamically switched directional antenna mesh system provides the capabilities throughout a 360 degree coverage area. In this system, multiple directional antennas are combined into an array that allows the radio to dynamically determine which antenna sector to use for each transmission. Each directional antenna in the array is capable of automatically creating a series of point-to-point links with the other directional antennas from neighboring nodes. The benefits are similar to directional antenna systems, including increase link budget, higher modulation rates, and interference mitigation, all of which results in better spectral efficiency and increased capacity.

 

However, unlike static directional system, a dynamic switched directional system is capable of automatically discovering all other units without requiring antenna pointing. Since the system can switch through all antenna sectors to “look” in all 360 degrees, it can automatically find all other nodes and establish point-to-point links with each one independently. This not only lowers the cost to deploy the system, but it also results in self-healing failover since the nodes can automatically and dynamically switch to connect when any changes are required throughout the network.

 

When considering the three “C”s of wireless, the dynamically switched directional antennas mesh systems provide a pragmatic approach. They provide a low-cost solution to integrated WiFi access and mesh backhaul and provide the ability to extend coverage easily with spectral efficiency and high capacity. The result is better performance at a lower total cost of ownership compared to omnidirectional or fixed directional solutions due to both technical and financial advantages.

 

One additional requirement in many citywide WiFi networks is the ability to handle VoIP traffic. To properly do this across a multi-hop mesh network requires mesh-wide synchronization to guarantee latency and jitter. One method of synchronizing transmission is the Time-Division Duplex (TDD) protocol that has been adopted by WiMAX. This protocol can be used to synchronize all transmissions throughout the mesh topology to coordinate traffic flow and mitigate self-interference. Mesh-wide synchronization requires a common (and wireless) timing source, which can be provided by the global network of Global Positioning System (GPS) satellites. The accurate, common clock allows the TDD protocol to coordinate simultaneous transmissions on multiple links throughout the mesh with great precision.

 

With traffic synchronization, latency and jitter are guaranteed, allowing the mesh to provide the more deterministic quality of service needed to support real-time applications such as voice over IP and video surveillance.

 

A dynamically switched directional antenna array provides the optimal design for today’s demanding wireless broadband networks. It offers the auto-discover and self-healing capabilities found in an omnidirectional system. This approach makes it simple to install and operate. At the same time, it includes the spectral efficiency and individual link optimization found in conventional directional antennas but instead utilizes dynamic links. Some may say this is the best of both worlds offering that clearly helps clean up the mesh.