Comment analyser les interférences WiFi et trouver le meilleur canal

How to Scan for WiFi Interference and Find the Best Channel. A Purple WiFi Intelligence Briefing. Welcome to the Purple WiFi Intelligence Series. I'm your host, and today we're getting into something that sits right at the intersection of RF physics and operational reality: how to systematically scan for WiFi interference and identify the best channel for your deployment — with a particular focus on the 5 gigahertz band, where the real performance gains are hiding. If you're managing WiFi across a hotel, a retail estate, a stadium, or a conference centre, this is not an academic exercise. Poor channel selection is one of the single most common causes of throughput degradation, client roaming failures, and the kind of guest complaints that land on the CTO's desk on a Monday morning. The good news is that it's entirely fixable — and it doesn't require replacing hardware. Let's get into it. First, let's establish the landscape. The 2.4 gigahertz band has three non-overlapping channels in most regulatory domains: 1, 6, and 11. That's it. In a dense venue — say, a conference centre with 40 access points — you are sharing those three channels across every AP, every neighbouring business, every guest's mobile hotspot, and every Bluetooth device in the room. The interference floor is almost always elevated before your first client even connects. The 5 gigahertz band is a fundamentally different proposition. In the UK and most of Europe, you have access to 19 non-overlapping 20-megahertz channels. Spread across UNII-1, UNII-2, and UNII-3 sub-bands, this gives you genuine channel reuse flexibility — particularly important when you're designing for high-density environments. The best channel for 5 gigahertz in your specific deployment depends on three variables: your regulatory domain, the presence of DFS-triggering radar sources nearby, and the channel utilisation of neighbouring networks. Let me explain DFS, because it trips up a lot of deployments. Dynamic Frequency Selection is mandated by the IEEE 802.11h standard for channels 52 through 144 — the UNII-2 band. These channels share spectrum with weather radar and military radar systems. When an access point detects a radar pulse on a DFS channel, it must vacate that channel within 10 seconds and cannot return for 30 minutes. In an airport, near a port, or in a city centre with dense radar infrastructure, DFS events can cause sudden, unexplained client disconnections. If you're seeing intermittent drops with no obvious cause, check your controller logs for DFS events before you do anything else. For most enterprise deployments, the pragmatic starting point for 5 gigahertz channel selection is the UNII-1 block — channels 36, 40, 44, and 48 — and the UNII-3 block — channels 149, 153, 157, 161, and 165. These are DFS-free in most regulatory domains, which means no radar-triggered channel changes and faster client association. The trade-off is that UNII-3 channels operate at higher frequencies, which means slightly reduced propagation through walls and floors. In a hotel with concrete construction, that's actually a feature, not a bug — it limits co-channel interference between floors. Now, how do you actually scan for interference? There are three tiers of tooling, and the right choice depends on your budget and the complexity of the environment. Tier one is built-in controller scanning. Every major enterprise WiFi platform — Cisco Catalyst, Aruba Central, Juniper Mist, Ruckus SmartZone — has some form of RF scanning built into the access point firmware. Dedicated radio scanning mode, sometimes called monitor mode or air monitor mode, puts one radio on a continuous passive scan across all channels, collecting RSSI data, channel utilisation percentages, and neighbouring BSSID information. This is your baseline. Run it for at least 24 hours to capture the full temporal pattern — interference in a hotel kitchen at lunch is very different from interference in a conference room during a morning keynote. Tier two is spectrum analysis. Tools like Metageek Chanalyzer with a Wi-Spy adapter, or Ekahau Sidekick, go beyond 802.11 frames and capture the raw RF spectrum. This is where you find non-WiFi interference sources: microwave ovens operating at 2.45 gigahertz, baby monitors, older cordless DECT phones that haven't been fully migrated, and — in industrial environments — frequency-hopping Bluetooth devices running legacy profiles. A spectrum analyser will show you a characteristic signature for each interference type. A microwave oven produces a wide, duty-cycled burst across the 2.4 gigahertz band every time it cycles. A Bluetooth device produces a characteristic frequency-hopping pattern. Knowing the source tells you whether the fix is a channel change, a hardware replacement, or a physical separation of equipment. Tier three is purpose-built site survey platforms. Ekahau Pro and iBwave are the industry standards here. You import a floor plan, walk the space with a survey adapter, and the platform builds a heat map of signal strength, channel utilisation, co-channel interference, and adjacent-channel interference across your entire floor plate. For a greenfield deployment or a major refurbishment, this is non-negotiable. For an existing deployment with persistent performance issues, a targeted survey of the problem zones is often sufficient. One metric that's frequently overlooked is the channel utilisation percentage. Most controllers report this, but few teams act on it. A channel utilisation above 70 percent on any AP is a red flag — you're approaching saturation, and latency will spike non-linearly under load. The fix is either channel reassignment, reducing transmit power to shrink the cell and reduce co-channel contention, or — in genuinely high-density environments — deploying additional access points with tighter cell sizing. Channel width is the other lever. 80-megahertz and 160-megahertz bonded channels deliver higher peak throughput for individual clients, but they consume a much larger portion of the available spectrum. In a dense deployment, 20-megahertz or 40-megahertz channels on 5 gigahertz will almost always outperform 80-megahertz channels in aggregate throughput, because you can run more non-overlapping cells simultaneously. Reserve wide channels for low-density, high-throughput scenarios — a boardroom, a back-office server room, or a dedicated IoT network segment. Now let me give you the practical framework I use when advising clients on channel optimisation. Start with a passive scan during peak operational hours. Do not run your initial scan at 2am on a Sunday — you will not see the interference environment that your users actually experience. For a hotel, scan during check-in and check-out peaks. For a retail environment, scan on a Saturday afternoon. For a conference centre, scan during a live event. Second, document your findings before making changes. Take a baseline of throughput, latency, and client association rates. This is your before state. Without it, you cannot demonstrate ROI or diagnose regressions after a change. Third, implement channel changes incrementally. Do not reassign every AP in a building simultaneously. Change one zone, validate for 48 hours, then proceed. Simultaneous changes make it impossible to isolate the cause of any new issues. Fourth, disable automatic channel selection — Auto-RF or RRM — in high-density deployments unless your controller is specifically tuned for your environment. The default RRM algorithms are calibrated for typical office deployments, not for a stadium with 500 APs. Uncontrolled automatic reassignment during a live event is an operational risk. The most common pitfall I see is over-reliance on the default channel plan. Most access points ship with auto-channel enabled, and most IT teams never revisit it. In a venue that has grown organically — additional APs added over time, neighbouring tenants installing their own networks — the default plan will be increasingly misaligned with the actual RF environment. A manual audit every 12 months, or after any significant physical change to the venue, is the minimum standard. The second pitfall is ignoring the 2.4 gigahertz band entirely because everyone uses 5 gigahertz now. IoT devices — door locks, environmental sensors, point-of-sale peripherals, digital signage controllers — frequently operate exclusively on 2.4 gigahertz. A congested 2.4 gigahertz band will not affect your laptop users, but it will cause intermittent failures in your operational technology layer, which is often harder to diagnose. Now for a few rapid-fire questions. Should I use DFS channels in a hotel? Generally yes, if your controller supports DFS well and you're not near an airport or port. The additional channel availability is worth it. But monitor your controller logs for DFS events in the first 30 days. What's the best channel for 5 gigahertz in a dense venue? There is no single answer — it depends on your neighbours. Run a scan, find the least utilised channels in the UNII-1 and UNII-3 blocks, and assign those. Channel 36 and channel 149 are often the least congested starting points in urban UK deployments. How often should I re-scan? Quarterly as a minimum. After any major event, any physical building change, or any new tenant moving into adjacent space. Can Purple's platform help with this? Yes — Purple's WiFi analytics layer gives you continuous visibility into client density, session quality, and throughput patterns across your estate, which feeds directly into channel optimisation decisions. It's the operational intelligence layer that sits above the controller. To bring this together: WiFi interference scanning is not a one-time activity — it's an ongoing operational discipline. The best channel for 5 gigahertz is the one with the lowest utilisation and the least interference in your specific environment, at your specific peak load times. That answer changes as your environment changes. The practical next steps are: run a passive scan during peak hours this week, pull your channel utilisation data from your controller, identify any channels above 70 percent utilisation, and make one targeted change. Validate it. Then build a quarterly review cadence into your network operations calendar. If you want to go deeper on any of this — site survey methodology, DFS event analysis, or how to integrate RF data with Purple's guest WiFi analytics platform — the links in the show notes will take you to the full technical guide and the Purple team's contact page. Thanks for listening. Until next time.