Comment configurer un WiFi invité : Guide de configuration d'entreprise sécurisé

How to Set Up Guest WiFi: A Secure Enterprise Configuration Guide A Purple Technical Briefing [INTRODUCTION AND CONTEXT - approximately 1 minute] Welcome to Purple's Technical Briefing series. I'm your host, and today we're covering a topic that sits right at the intersection of network security, compliance, and customer experience: how to set up guest WiFi securely in an enterprise environment. If you're an IT manager, a network architect, or a CTO responsible for a hotel group, a retail estate, a stadium, or a conference centre, this briefing is for you. We're going to move fast and stay practical. No theory for theory's sake. Just the decisions you need to make, the standards you need to meet, and the pitfalls you need to avoid. Let me frame the problem first. Guest WiFi is no longer a nice-to-have. It's a baseline expectation. Guests, shoppers, fans, and passengers arrive expecting immediate, reliable connectivity. But the way most organisations have deployed it - a shared password on a second SSID, maybe a splash page on top - is not a secure enterprise design. It's a home fix applied to a commercial problem. And when you're operating at scale, that gap creates real liability: regulatory exposure under GDPR, PCI DSS risk if your guest network touches payment infrastructure, and a network you genuinely cannot control once that password is out in the world. So let's fix that. Here's how. [TECHNICAL DEEP-DIVE - approximately 5 minutes] The foundation of any secure guest WiFi deployment is network segmentation. Before you think about captive portals, authentication methods, or analytics, you need hard separation between your guest traffic and everything else. That means a dedicated SSID mapped to its own VLAN - a Virtual Local Area Network - with firewall rules that deny access to internal subnets by default. Think of it this way: your guest network is a controlled external zone. You're not giving visitors keys to the building and hoping they stay in the right room. You're giving them a separate entrance, a separate corridor, and access only to the internet. Nothing else. The technical baseline looks like this. Create a guest SSID. Apply WPA2 or WPA3 encryption - we'll come back to WPA3 specifically. Enable client isolation so guest devices cannot see each other on the network. Block access to the primary LAN. Use a dedicated DHCP scope. And critically - test it from a real client device before you call it done. If a guest device can discover your printers, reach an admin interface, or cast to a meeting room screen, the network is not finished. Now, WPA3. If you're deploying or refreshing hardware in 2026, WPA3 should be your default. The Wi-Fi Alliance has required WPA3 certification for all new devices since July 2020, and most enterprise access points from Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, and Ubiquiti UniFi have supported it via firmware since 2019 or 2020. You may not need new hardware - just a firmware audit. What does WPA3 actually give you? Three things that matter operationally. First, Simultaneous Authentication of Equals - SAE - replaces the WPA2 Pre-Shared Key handshake. Under WPA2, if an attacker captures the four-way handshake between a client and your access point, they can run offline dictionary attacks against it indefinitely. SAE eliminates that. Even if someone captures every packet of your authentication exchange, they cannot derive the session key. Second, Forward Secrecy. Under WPA2, if an attacker records encrypted traffic today and later obtains your network password - through a disgruntled employee, a phishing attack, or a data breach - they can retroactively decrypt everything they recorded. With WPA3's SAE, each session generates a unique ephemeral key. Compromise the password tomorrow, and yesterday's traffic stays encrypted. For hospitality environments handling guest payment data, or retail networks processing loyalty transactions, this is a meaningful risk reduction. Third, Opportunistic Wireless Encryption - OWE. This is the game-changer for public WiFi. On a traditional open network, guest traffic is transmitted in plaintext. Anyone with a packet sniffer on the same network can read it. OWE automatically negotiates an encrypted connection between each client and the access point, with no password required and no change to the user experience. The guest clicks connect. Their session is encrypted. This directly addresses GDPR obligations around protecting personal data in transit. Now let's talk about the authentication layer - specifically, how guests actually get onto the network. The captive portal is still the most common mechanism, and it's still useful, but only if you're clear about what it is and what it isn't. A captive portal is an onboarding and policy tool. It is not your primary security control. The security comes from the VLAN segmentation and the WPA3 encryption underneath it. What the captive portal gives you is identity and consent. When a guest enters their email address and ticks a marketing opt-in checkbox, you've captured first-party data with explicit GDPR consent. That's valuable. Purple's platform processes 440 million logins annually across 80,000 venues - that data, captured compliantly through guest WiFi portals, is what turns a cost-centre network into a business intelligence asset. For environments where guests return regularly - hotel chains, transport networks, multi-site retail - you should be looking beyond the captive portal towards identity-based onboarding. Passpoint and OpenRoaming allow devices to authenticate automatically and securely on subsequent visits, with no portal interaction required. The device carries a credential, the network recognises it, and access is granted silently. This is what the industry calls seamless secure onboarding, and it's the direction enterprise guest WiFi is heading. For staff and corporate devices on the same physical infrastructure, IEEE 802.1X is the standard. It requires each device to authenticate individually against a RADIUS server - Remote Authentication Dial-In User Service - before network access is granted. The RADIUS server checks credentials against your directory - Microsoft Entra ID, Okta, or Google Workspace - and returns an accept or reject. If accepted, the access point places the device on the correct VLAN dynamically. Staff on the staff VLAN. Contractors on a restricted VLAN. Guests on the guest VLAN. All from a single physical infrastructure, all enforced automatically. This is what Purple calls Identity-Based Networks. The same hardware serves multiple user types, each with appropriate access, enforced at the authentication layer rather than through separate physical infrastructure. [IMPLEMENTATION RECOMMENDATIONS AND PITFALLS - approximately 2 minutes] Let me give you the implementation sequence, and then the three pitfalls I see most often. The sequence is: architecture first, then authentication, then the portal layer, then analytics. Do not start with the portal. Start with the VLAN. Get the firewall rules right. Confirm isolation is working. Then build the onboarding experience on top of a secure foundation. For hardware, Purple operates as a hardware-agnostic cloud overlay. You can deploy it on top of existing Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme, or Fortinet infrastructure. You do not need to rip and replace. The cloud overlay handles the captive portal, the consent flow, the analytics, and the identity layer, while your existing hardware handles the radio and the VLAN enforcement. Now the pitfalls. Pitfall one: confusing a separate password with actual network separation. A new SSID with a different password, sitting on the same broadcast domain as your staff network, is not a security boundary. It's cosmetic. VLAN segmentation is the boundary. If you haven't confirmed that guest traffic cannot reach internal subnets, you haven't finished. Pitfall two: leaving client isolation off. This is the setting that prevents guest devices from communicating with each other on the network. It's off by default on some controllers. Turn it on. A guest device that can reach other guest devices can be used to launch attacks against them - particularly relevant in hotel environments where guests are sharing the same SSID. Pitfall three: not testing fail-open behaviour. If your captive portal or authentication service becomes unavailable, what happens? Does the network block all guest access, or does it fail open and let everyone through without authentication? Decide this deliberately. For most venues, fail-open is the right call - guests should not lose connectivity because a portal server is temporarily unreachable. But fail-open without any policy enforcement is a risk. Configure your fallback state explicitly. [RAPID-FIRE Q AND A - approximately 1 minute] Let's do a quick Q and A on the questions I hear most often. "Do I need to replace my access points to get WPA3?" Probably not. Audit your firmware versions first. Most enterprise-grade APs from the major vendors support WPA3 via firmware update. "What's the minimum viable secure guest WiFi setup?" Dedicated SSID, dedicated VLAN, WPA2 or WPA3, client isolation enabled, firewall blocking internal subnets. That's the floor. Everything else - portals, analytics, identity - is built on top of that. "How do I handle IoT devices that don't support WPA3?" Put them on a dedicated SSID running WPA2, isolated on their own VLAN. This is standard segmentation practice. Do not mix IoT devices with guest traffic. "Is a captive portal enough for GDPR compliance?" It's part of the answer. You need explicit, informed consent - a clear opt-in checkbox, a link to your privacy policy, and a record of when consent was given. Purple's platform handles all of this and stores the consent records. But the portal alone is not compliance. The data handling practices behind it matter equally. [SUMMARY AND NEXT STEPS - approximately 1 minute] To summarise. Secure guest WiFi starts with architecture, not aesthetics. Get the VLAN segmentation right. Apply WPA3 where your hardware supports it - and it probably does. Use a captive portal for identity and consent, not as your primary security control. For returning visitors and multi-site estates, move towards identity-based onboarding with Passpoint or OpenRoaming. And layer your analytics platform on top to turn the network from a cost centre into a source of first-party business intelligence. The organisations doing this well - Premier Inn, Harrods, Manchester Airports Group, Pizza Express - are not running exotic infrastructure. They're running standard enterprise hardware with a cloud overlay that handles the identity, consent, and analytics layers. The result is a network that IT can trust, marketing can use, and guests actually want to connect to. If you want to go deeper, Purple has a full technical guide at purple.ai covering RADIUS configuration, WPA3 deployment, and captive portal architecture. The guide covers everything we've discussed today in detail, with worked examples and configuration checklists. Thanks for listening. This has been a Purple Technical Briefing.