Troubleshooting Wi-Fi networks and benchmarking performance is extremely difficult, due to the complex nature of how Wi-Fi works. The trouble is, the performance of any one individual client, is a factor of so many variables that are all in flux at the same time. So how can Service Providers measure overall Quality of Experience (QoE) and maintain Service Level Agreements (SLA) on their Wi-Fi networks? There are two parts to the puzzle – first, measuring QoE and second, adjusting the configuration to maintain QoE at a certain level.
In large enterprises, before deploying a single AP, it is standard practice to carry out detailed site surveys and develop a channel plan using predictive planning tools, to maximize coverage and capacity and minimize interference. By the time you actually deploy the gear, you have already established some realistic performance expectations for the minimum bandwidth available to each client, or for special applications such as VoIP, based on expected user volumes and usage assumptions.
This is a great starting point, but as everyone knows, wireless networks are constantly evolving and the environment changes. User demands continue to rise, cubicles, partition walls and furniture get moved, and new sources of interference emerge both from within the facility and from external sources such as neighboring Wi-Fi networks and non-Wi-Fi sources impacting the Wi-Fi spectrum, which are also evolving in parallel. Thus, to maintain the performance levels you originally planned for, enterprises need to continuously monitor the network to detect trouble spots, and make adjustments to AP settings and the channel plan to compensate for the environmental and usage changes.
Enterprise Wi-Fi equipment vendors’ network management platforms go some way toward giving you visibility of what is going on, albeit limited to a network-centric perspective. It’s not easy to connect all the dots, but you can drill down and examine data on packet losses and transmission retries for an AP, and SSID or an individual client using the stats collected by APs and controllers.
In addition, radios can be set to scan the spectrum for channel conflicts and interference either in a background or dedicated mode. Background channel scanning alongside client access is a popular approach, but in the most mission critical environments it is not uncommon for 1 in 10 radios to be dedicated to spectrum analysis. Indeed, many enterprise WLAN vendors now offer 802.11ac APs with three radios, one of which is intended for dedicated spectrum analysis. Obviously all these option have a cost. In fact, spectrum analysis and radio resource management is such a big business that many enterprise Wi-Fi equipment vendors have developed a portfolio of hardware and software components under imaginative brand names like CleanAir, EzRF, ARM and so forth.
Screenshot from Cisco’s CleanAir management console.
For a complete picture, you also need to measure performance from the user perspective, not the only network view, because that’s ultimately what the user experiences. This is no easy task either, because every client device has different capabilities. A single antenna in a smartphone does not compare with the MIMO antennas in a high-end laptop. So how do you first obtain these metrics and then represent them in a meaningful actionable way? This is possible. There are a few companies such as 7signal that specialize in this. They have developed Wi-Fi network probes which emulate the behavior of clients, and can continuously measure performance day and night, giving you excellent visibility of performance in an idle state and busy state. And this detailed information can be used to compare real-time performance against predefined SLAs for a wide variety of metrics, such as MOS score etc.
But here again, there is a cost, and still it takes human intervention to take corrective action when issues get identified. For service providers, even those going after Enterprise businesses, these various measures are cost prohibitive. It's different story for enterprises with a crew of experienced IT staff on hand and business stakeholders demanding an impeccable mobile experience. This is especially true in verticals such as healthcare and manufacturing where "mission-critical" is more than a buzzword.
What’s all the worry about? The latest Wi-Fi standard 802.11ac is faster than most Internet connections delivered to homes and small businesses. So what’s the issue? It’s true that 802.11ac is fast, but with that speed comes diminished range. The result is that homes that used to have decent coverage, albeit not very fast, find they have coverage holes as soon as they upgrade to 802.11ac.
As we all become dependent on anytime anywhere access the Internet, home owners want performance and coverage everywhere...even outdoors. This is driving them to buy supplementary 3rd party access points and ranges extenders. With the result that many homes now have multiple AP in homes each competing for a finite number of channels. Channel bonding features only serve to deplete those channels faster. Ultimately this means the 5 GHz band is starting to get clogged, just as the 2.4 GHz band has been for years.
Interference from neighboring APs is already an issue in dense urban areas, and especially in multi-tenant dwellings. The trouble is, the signals from other homes are usually not within a service provider’s control – the APs may be owned by competing providers, or they may be 3rd party APs that the consumer has purchased themselves.
If that is not enough, there is another looming threat to spectrum availability – This time coming from Cellular operators. Licensed spectrum is extremely expensive, and it is getting scarce. But Mobile data consumption is roughly doubling annually. While there are many viable tactics, including Small Cell deployment, that can help operators keeping pace with demand and improve utilization of their licensed spectrum, one of the most appealing and cheapest options, is to augment licensed spectrum with unlicensed spectrum. Small cells that combine Wi-Fi and 3G/4G service are already being widely deployed. And there is an even greater trouble ahead from variants of LTE-U.
LTE-U and LAA (Licensed-Assisted Access) both provide a way for Mobile Network Operators to scale capacity by combining licensed spectrum with unlicensed spectrum in the 5 GHz band. For the past few years, Wi-Fi industry advocates have been up in arms about this. There are great fears and there is considerable independent evidence that LTE-U/LAA does not play fair with Wi-Fi, and that the stability and performance of existing Wi-Fi networks could be substantially compromised.
A more recent, and even more insidious development, is Qualcomm’s MulteFire proposal. This is a new, LTE-based technology that solely operates in unlicensed spectrum, and doesn’t require an “anchor” in licensed spectrum at all. Yikes! Don’t count on the FCC to defend the interests of Wi-Fi users, that’s not how their bread is buttered!
So how can Service Providers avoid Wi-Fi interference from all these external sources so they can provide a stable foundation for the high-margin OTT services such as building automation, energy management, ambient assisted living and security they want to upsell in future? According to data from Statistica, each one has the potential to deliver double or triple the Average Revenue Per User (ARPU) over basic Internet service.
Wi-Fi interference issues are real...this is not speculation. Service Providers are already facing these issues in high density cities such as Chicago, New York and Los Angeles, London and Tokyo. It’s not just a big city problem though, even some suburban areas have experienced similar issues. And as competition for spectrum grows more fervent, they are going to run into this more and more.
To retain their customers, and drive OTT services revenue they will need to implement radio resource management mechanisms that can detect interference, compute the impact against expected user performance SLAs and then automatically choose more favorable channel / power settings. Without the means to ensure access to a stable medium, any efforts to control bandwidth utilization or prioritize applications such as voice, are inevitably only effective under ideal conditions.
To deliver a reliable Wi-Fi service, one must first ensure that the RF environment is stable and interference is contained. The most basic strategy is one of interference avoidance – that means choosing the cleanest available channel for each radio, and tuning transmit power to reduce overlap between cells. But this is not a set-it-and-forget it thing. Wi-Fi networks are constantly changing – from one day to the next, neighboring networks may get reconfigured, suddenly exposing your network to a channel conflict that wasn’t there before. So the monitoring of spectrum utilization needs to be a continuous background process, which triggers appropriate alerts and an automated corrective response when performance is measurably degraded.
In SMB deployments where multiple APs are the norm from the outset, Enterprise WLAN vendor solutions are a viable option, but still too expensive. But the management systems are not designed for the deployment scale required by Service Provider. Even if they were, Service Providers are reluctant to get locked into a proprietary vendor solution. But when Service Providers serve both SMB and home markets... this approach would still leave them without a practical answer for the home environment.
For Wi-Fi services to be profitable, in the home, Service Providers must minimize both the equipment and support costs by consolidating functionality into the least number of devices. This is accomplished by adding Wi-Fi into the broadband CPE or residential gear – not by having separate APs. Now with growing demand for multiple APs in homes, they need a way to deliver Wi-Fi in the Gateway, and separate APs, and for both types of devices to cooperate as a system. To accomplish this they must either build their own solutions or turn to broadband equipment vendors, not traditional enterprise WLAN vendors.
Historically, the Wi-Fi component on broadband CPEs is mostly unmanaged and reports very little about its health, users and how it is being used, to a Service Provider’s OSS back-end. It is also the case that many such devices have feeble radios and basic antenna systems. There are many other enterprise-grade features absent too, most notable is the lack of radio resource management. All this needs to change, and it is. More enterprise-grade features are finding their way into these devices, and Service Providers are realizing they need sophisticated radio resource management mechanisms that can scale to the millions of nodes that exist in their Wi-Fi networks.
Support calls must be minimized and truck rolls are to be avoided. A single truck roll to a home can pretty much erase a year’s worth of profit from that household. Therefore in addition to being highly automated, any solution needs to make the best use of radios and chipset used in the device without modification. It also needs to be hardware agnostic, and easily ported to the next generation chipset, or even a different chipset or OS, so that Service providers retain the flexibility to change equipment providers and platforms as next generation technologies mature.
Any solution must allow complete coexistence between old and new APs, puny and high-end radios, and chipsets with different capabilities, and within a small geographic zone, the radio management must be coordinated. It’s no use having radio management for each household or SMB operating in a silo, unaware of radios adjacent to them, which are also under the control of the same service provider.
Wi-Fi Chipsets from different vendors differ in what radio data and throughput statistics you can extract from them. But here are the main data sets of interest: RSSI, TX Errors, Clients, Channel utilization, Neighbors, Non-Wi-Fi interference events - a combination of which can be used to deduce some kind of real-time QoE metric, and compare it to historical benchmark data.
embedUR has more than a decade of experience developing Wi-Fi technologies for some of the biggest suppliers of enterprise networking and broadband access equipment. Since many of the products we have built for our customers are designed for service provider deployment, we gained early visibility of this looming interference problem and set out to solve it. We think we have. Learn more about embedUR’s Radio Resource Management solution for Service Providers.