How to improve Wi-Fi performance: Experts talk planning, troubleshooting

Wireless is the technology of choice and the default these days, and achieving optimal wireless performance is a key goal for IT departments in most buildings, campuses, metro hotspots, and wide area network settings.

Because radio frequency (RF) propagation always involves a high degree of variability, it is often difficult to predict the exact behavior of a given device. Variables include operating conditions, user and application traffic demands, and the capabilities and settings of individual vendor products. When considering mobility, Wi-Fi testing and validation can become very complex indeed.

[[251368]]

Performance variables include throughput, coverage, time-bound traffic (primarily phone calls and streaming video), reliability, security, speed, and most importantly, capacity, which is the ability to meet all traffic needs at any place and time.

As the price/performance ratio of Wi-Fi equipment continues to improve with advances in standards, technology, and implementation, many IT organizations have taken a brute-force approach: simply upgrading or adding access points (APs) and Wi-Fi controllers, Ethernet switches, and related hardware.

Unfortunately, this approach does not typically optimize performance and is expensive both in terms of equipment and the time required of network operations professionals.

To find a better way, we interviewed experts and came up with a set of best practices for Wi-Fi performance optimization. We focused on the three main phases of any large-scale Wi-Fi installation: planning and pre-installation; post-installation functional and performance testing and verification; and handling temporary performance issues (troubleshooting). We also sought advice on the necessary operational toolsets, as well as future directions for Wi-Fi performance optimization.

Interviewees included Dr. Eldad Perahia, distinguished technologist in the office of the CTO at Aruba; Matthew MacPherson, senior director of WLAN, Jim Florwick, senior technical market engineer, and Nilesh Doshi, software test manager, all from Cisco; Mika Hakala, CEO, and Jerry Olla, product manager, at Ekahau; Mike Leibovitz, senior director of product management, at Extreme Networks; Henry Youssef, chief mobility architect, at Henry Ford Health System; Bob Friday, co-founder and CTO, at Mist Systems; and Dr. Leigh Chinitz, CTO, at octoScope.

Wi-Fi Performance Optimization: Planning Phase

Even before deciding on a specific device and its subsequent installation, much can be done to optimize solution performance and prevent operational issues, experts say.

Hakala recommends designing for end-user quality of experience (QoE), capacity, and latency, not just coverage. Olla added that understanding user location, density, and application requirements are key to ensuring capacity. Olla also noted that measuring the physical space to understand RF behavior can save a lot of headaches later.

Chinitz recommends testing individual access points under actual operating conditions and loads to help understand the limitations of individual APs and capacity calculations. Adding a second AP can test roaming, load balancing, and band steering, and in most cases can reveal path loss and dead zones in specific locations.

Youssef said it’s useful to do an “AP-on-a-stick” test in a potentially difficult environment like a hospital. He’s even seen interior paint negatively impact RF performance.

Mist Systems is a leading advocate of combining indoor Bluetooth Low Energy (BLE) positioning technology with outdoor GPS technology. Friday believes that high-density AP spacing for indoor use cases can also help optimize Wi-Fi deployments.

Florwick recommends doing an initial RF scan for unlicensed bands, Wi-Fi or other potential interference, and choosing an initial Wi-Fi channel accordingly. The optimal channel bandwidth is usually a function of interference and channel utilization.

Perahia suggests that building an analytical model of the expected load, even just using a spreadsheet, can be used to analyze capacity assumptions.

Leibovitz said careful examination of the capacity, management and local services (such as DHCP) of the rest of the network (not just Wi-Fi) can avoid problems, as problems on the line often manifest themselves in the wireless elements of the network. It is also important to consider emerging, potential and expected traffic demands, not just current ones, such as those from the Internet of Things.

Wi-Fi Performance Verification

All of our participants agree that post-installation functional testing and performance verification is essential and should include at least a minimum sample of users with the production application under real operating conditions.

Youssef said he typically sees more issues related to the product, particularly the client devices, rather than the RF. Widespread changes in client behavior and compatibility are common, as well as issues such as driver settings being inappropriately changed by otherwise well-intentioned end users. He recommends a cursory re-survey of deployed spaces each year, noting that facilities organizations sometimes move or decommission APs for other purposes, too.

Perahia added that improper (out-of-spec) or defective AP cabling can cause unexpected performance degradation.

Doshi recommends alpha testing new deployments before allowing users to use them, measuring elements such as voice quality and roaming behavior. MacPherson recommends using analytical tools and leveraging the usual facilities in management consoles.

Leibovitz recommends occasionally spot-checking in multiple locations with specific end-user devices, even in installations that appear to be functioning normally.

Olla noted that documentation is critical at all stages of deployment, and even observational comments can provide solid clues when verifying feature performance.

Wi-Fi Troubleshooting and Toolset

Our participants noted that the software or hardware toolsets used to verify performance are also applicable when troubleshooting, so data collection has data collection and analysis capabilities covering RF, wireless network behavior, wired network, and certain data.

Leibovitz recommends that installations collect as much operational data as possible, because it is easier to spot patterns when large amounts of data are available. Analytics capabilities are crucial here.

Olla suggested checking the rest of the network value chain to see if there is a problem, even though users may think the problem is defined as “wireless,” and Chinitz added that talking to end users can yield valuable insights.

Youssef recommends using a third-party, independent Wi-Fi analysis and assurance tool that is independent of the chosen management console to get a second opinion outside of the management console and gain access to different troubleshooting strategies.

Friday said a real-time cloud-based management stack and API access, which enables better automation and troubleshooting for end users and third-party tool vendors, will allow for rapid development and deployment of new use cases. He also pointed to the availability of natural language query systems that can answer questions relevant to network domain experts, and expects them to gain greater speed and accuracy in the coming years.

Perahia recommends using a second AP from a different vendor to test particularly tricky situations. If the results are poor on both, then it's likely the client's fault. He also suggests using packet sniffing software, or even a simple benchmarking tool like Iperf can be used to quickly isolate troublesome clients, APs, and traffic types.

Even after nearly three decades of widespread use of wireless LANs in enterprises everywhere, it is clear that much work can still be done to enhance performance optimization and related operational elements.

Future Trends in Wi-Fi Performance Optimization

Wi-Fi experts we spoke to said they want to see more advanced analytics tools, including advanced capabilities based on artificial intelligence (AI) and machine learning (ML). These capabilities are already capable of automatically fixing problems before operators are aware of them. Friday calls these self-driving networks.

Chinitz discussed virtual benchmarking, which involves testing access points (and possibly even client devices) in a specially designed RF isolation chamber under a very large range of conditions, and then quantifying the results into a profile that can be used to predict real-world results.

According to Perahia, future operational performance analysis and optimization will depend on information collected directly from client devices, not just the infrastructure. The reason for this is the increasing use of beamforming, multi-user MIMO and other technologies that can cause traditional sensors and assurance tools to be in the RF null point during specific transmissions. WLAN chipset vendors may need to enhance client capabilities here in order to collect the required performance data.

MacPherson said SDN is likely to play an increasing role in data collection and optimization in the wireless portion of an organization's network. He also noted that management console interfaces need to be simpler, more precise, and take advantage of the rise of AI and ML.

At the very least, having a basic benchmark server, perhaps like speedtest.net, and something as simple as iperf implemented in the AP would make it easy to analyze the actual performance of a given air link. This is a big step up from just knowing the reported signaling rate, which can vary widely and rapidly in operation and is of little use in performance optimization anyway.