By Azimuth Systems
Wireless mesh networking, combining performance and simplicity with powerful economics, can be used when wired backhaul networks are impractical in applications such as municipal wireless networks, public safety, health care and education. In fact, a recent ABI Research report forecasts a $1.2 billion market for citywide Wi-Fi networks by 2010 and more than 50 municipalities, including Chicago, Houston, Philadelphia and San Francisco, have already installed wireless metropolitan networks with several others currently in the deployment process.
The maturity of the Wi-Fi industry and standards has transformed a low cost home network solution into a secure, high performance, enterprise-ready networking technology. The proliferation of Wi-Fi use at home, in the office and on the road has sparked the emergence of an entire range of targeted solutions for public access; many of which are based on Wi-Fi mesh.
Carriers and service providers understand the costs involved in the wide adoption of early-stage technologies and their customers will not tolerate unreliable new services. If an unpredictable network is deployed, the service provider bears the challenge and cost of customer support and troubleshooting.
If Wi-Fi is to effectively compete with technologies such as CDMA-2000 EV-DO, W-CDMA and UMTS, the technology must evolve from its low-cost, consumer roots to a carrier-grade infrastructure with improved performance and robustness. To guarantee the success of wireless mesh technology and to establish its credibility for metro-area network applications, carriers and service providers require reliable pre-deployment testing.
Step-by-step pre-deployment testing in a controlled laboratory environment is the most efficient way of introducing new and complicated networking technology such as wireless mesh networking. Thorough testing reveals technological flaws and weaknesses, helping vendors fix problems before product shipment. Such testing must be automated, or it lacks the scalability to test all possible load, motion, background interference and device configuration scenarios. The absence of automation may allow bugs and network vulnerabilities to remain unchecked.
Testing wireless mesh presents unique testing challenges, because this wide-area infrastructure supports mobile users and high-throughput data, voice and video applications. Wireless mesh nodes use sophisticated routing algorithms to direct traffic to its destination, instead of connecting through traditional cabled backhaul; therefore, mesh nodes cannot be tested in the same way as single access points – they must be tested together as a self-configuring, self-healing system.
This article reviews the requirements and challenges of mesh testing and discusses ways to address them.
Mesh Networking Performance Variables
Mesh networks are self-configuring and self-healing, which means that mesh nodes discover each other and determine the optimum frequency scheme for communicating with neighboring nodes and local clients. Because they must promptly respond to environmental changes and fault conditions by rerouting traffic or reconfiguring the frequency channel scheme, mesh nodes must constantly be aware of network conditions.
The most critical factors impacting user satisfaction for performance in a communications network are throughput and quality of service (QoS). In a Wi-Fi mesh system, throughput degrades as the number of hops increases. QoS for services such as voice and video is highly dependent on throughput, packet loss, delay and jitter, which also increase with each successive hop. When testing mesh networks for performance variables, measurements should be performed over different hop counts and plotted vs. hops.
A variety of device and network settings impact mesh performance and should be measured in a controlled laboratory environment. Some of these variables—such as range, equipment and vendor interoperability, interference, and others—are applicable to all Wi-Fi network testing. However, a number of variables affect mesh performance in particular, as outlined below.
Node Client Capacity shows the ability of individual mesh nodes to scale to a large number of clients. Because mesh deployments are primarily well-suited for and targeted at municipal or public access hotspots, conference centers, libraries, and hotels, the number of users may be high at time, fluctuate considerably, and have different demands.
Multi-hop Throughput Performance measures the throughput for varied topologies and node configurations. Network size is critical in mesh networking, because as the deployments succeed and the benefits are justified, customers will want to expand their networks to include bigger and bigger deployments over wider and more difficult coverage areas.
As deployment size increases, so will the number of nodes traversed from a user to the backhaul wired link. The performance will need to be tested. Each additional hop can potentially increase the network delay. In addition, the amount of background traffic and the number of network users increase (assuming each node is also a user’s node, not only a relay).
Call Capacity tests the capacity of the mesh network to deliver real time (delay critical) services like VoIP. Voice call capacity and quality is the measurement of voice traffic across multiple hops and with varying background traffic. The promise of fixed-mobile convergence/Wi-Fi to cellular roaming—such as better coverage, a decrease in dropped calls and high quality voice—has many service providers ready to jump in the market. Mesh networks must meet this new level of service, in addition to allowing regular data access. QoS-specific functions are a key part of supporting real-time communications traffic.
Node Roaming tests verify the robustness of the network design, auto-configuration and failure recovery through smooth node-to-node transition. Mesh networks can reroute traffic from a user through different routes; however, each vendor uses its own approach to make this happen seamlessly and efficiently. Evaluating the speed and disruption caused in making this transition will factor into the users experience as the system changes.
This is critically related to service applications on the network, such as public safety, or remote access to municipal data, billing and registration and as such to customer satisfaction. In addition, Wi-Fi access for radio, TV, phone and other services are an appealing concept, but if the service is spotty, the user base will revolt. Discovery and self healing of user connection and node to node connections is critical to operation.
Security tests include security performance and multi BSSID (Basic Service Set Identifier) isolation and security tests. Security performance measures the impact on performance for different security protocols. Private networks such as the library network, municipal employees’ network, or leased virtual corporate networks in a building or coverage area use security – and operate in parallel with other secure and open public networks. This test is focused at the load that security adds to the network and whether it affects other network functions.
Mesh Networking Test Methodology
Performance of wireless mesh devices and systems must be verified under controlled conditions that allow the test conditions and causes of poor performance to be easily isolated and managed. Multi-radio mesh test setups can be configured for controlled laboratory testing by interconnecting various mesh topologies using a conducted test setup.
By using multiple radios, mesh networks can segregate local client traffic and backhaul traffic to multiple simultaneous channels. Mesh networks that use multi-radio nodes with different channels that communicate with neighboring nodes and with local clients tend to have better throughput than early single-radio implementations, which shared one channel for client and backhaul traffic. As the number of radios in mesh nodes significantly impacts performance, perform measurements with different number of radios activated and compare the results. An example of how a simple mesh topology can be configured is shown in figure 1.
(Click image to enlarge)
Figure 1: Mesh test configuration. Mesh nodes are interconnected using programmable or fixed attenuators to emulate a variety of path losses causing the mesh to self-configure and select backhaul and client channels. Client traffic can be emulated using specialized test equipment. Throughput and QoS measurements can be performed at different points in the mesh and cover multiple hops.
Conducted test setups use shielded enclosures to isolate each mesh node from neighboring nodes (figure 2) to control signal flows through the intended conducted paths instead of coupling over the air.
(Click image to enlarge)
Figure 2: Mesh Test Isolation Chambers. Each mesh node in the test setup must be placed into a shielded isolation chamber with all the radio antenna ports cabled to the outside world for interconnection in a test network. This ensures that each mesh node communicates with clients or with neighboring nodes through conducted paths rather than through uncontrolled coupling. Isolation chambers also protect against external interference making the testing robust and repeatable.
By interconnecting a variety of network topologies in a conducted test setup, the factors affecting throughput and QoS, (packet loss, delay and jitter), can be measured in a controlled laboratory environment. Using programmable attenuators, path losses can be varied among mesh nodes, which allow the verification of a mesh network’s self-configuration and self-healing abilities. Programming high attenuation on an active backhaul interconnection forces the mesh to redirect the traffic flow, so fail-over conditions can be created dynamically and in a controlled manner. Throughput and QoS performance should be measured under both fail-over and normal conditions.
Settings for Mesh Tests that Impact Mesh Performance
When testing wireless mesh networks, it is also critical to measure a variety of device and network settings that impact mesh performance. As discussed in the previous section, these configurations should be varied in a controlled laboratory environment as the measurements are performed. The following outlines settings for mesh tests that impact mesh performance and highlights proper testing methodology.
Hop measurements: Throughput and QoS are directly impacted by the number of hops through the mesh network. Hops equal the number of nodes traversed between the user device and the high throughput wired backhaul network. Traffic load from groups of clients using data, voice and video services should be emulated, along with 802.11e prioritization for voice, data and video per group with a variety of traffic loads and packet sizes. Additionally performance should be measured for incrementing numbers of mesh hops.
Interfering traffic load: Each mesh node routes local traffic and forwards traffic from other nodes via its backhaul links. Depending on the efficiency of the routing algorithms, congestions can occur on backhaul links, impacting performance of the entire mesh. Configuring a traffic source and destination addresses in such a way as to exercise routing logic should be performed in a controlled laboratory test.
Direction of traffic flow: Throughput and QoS must be measured in upstream and downstream directions. The test application should allow controlled upstream, downstream and bi-directional measurements.
Security and QoS Settings: Throughput and routing efficiency may be impacted by security settings. To determine this impact, test with a variety of standard IEEE 802.11i security settings for groups of emulated clients. Emulate a mix of data, voice and video clients with different QoS profiles, while measuring throughput to test the ability of mesh infrastructure to prioritize voice and video over data. Additionally, the test should plot voice and video quality metrics vs. hops, load and other settings.
Distance between nodes: The RF conditions and loss between any two nodes like path loss and multipath conditions can have a significant affect on the performance of the system. Longer distances and other such variances between clients and mesh nodes should be used to measure and plot throughput vs. path loss and multipath models. This allows the ability to test the resilience of the system to changing conditions and estimate the optimal operating conditions. The test should also use programmable path loss and multipath to emulate failure conditions and cause traffic flow reconfiguration.
Interference including ACI (Adjacent Channel Interference) and co-channel interference: Throughput and QoS performance are affected by interference. Co-channel and adjacent channel interference are normal in mesh networks since neighboring radios can communicate on the same or adjacent channels and interfere with the channel under test. Perform throughput and QoS measurement in the presence of co-channel interference and ACI.
Mobility conditions: The most challenging and most important test parameter to control is mobility. Emulate motion of clients with respect to mesh nodes, motion of mesh nodes with respect to other mesh nodes (e.g. a bus with a mesh node moving through the city mesh.) and multiple clients moving at the same time. Different velocities of motion, for example, people walking, mesh nodes on busses and trains, etc. should be measured, along with different cell overlap conditions.
Conclusion
Wireless mesh networking technology is emerging as the primary infrastructure for several broadband services, including wide area voice and data transmission. A significant new market for the Wi-Fi industry, wireless mesh presents a set of unique testing challenges.
To properly establish its credibility for metro-area network applications, pre-deployment testing of mesh networks must benchmark performance testing in a controlled laboratory environment. Automated, standardized, repeatable test methodology will ensure the scalability of the system to support load, motion, background interference and device configuration scenarios, as well as identify bugs and network vulnerabilities prior to deployment.
Thorough and methodical lab testing evaluates feasibility and catches problems with early implementations prior to being introduced into the field, where finding problems is logistically challenging and costly. With thorough testing, mesh networks can be deployed on a large scale more confidently.