Comment configurer l'authentification WiFi 802.1X : Un guide étape par étape

How to Configure 802.1X WiFi Authentication: A Step-by-Step Guide A Purple Enterprise WiFi Intelligence Podcast [INTRODUCTION — approximately 1 minute] Welcome back. I'm speaking today as a senior solutions architect, and if you're listening to this, you're probably staring down a network security project that involves 802.1X authentication — either because your compliance team has flagged it, your insurer has asked about it, or you've just inherited a network that's running on a shared PSK and you know that's not good enough anymore. So let's get straight into it. 802.1X is the IEEE standard for port-based network access control. It's the backbone of enterprise WiFi security — the mechanism that ensures every device connecting to your network has been positively identified and authorised before it gets a single byte of traffic through. This isn't optional for organisations handling payment card data under PCI DSS, it's not optional for healthcare environments under GDPR and NHS data security standards, and frankly, for any organisation running more than a handful of access points, it's the right architecture. Over the next ten minutes, I'm going to walk you through the technical architecture, the RADIUS configuration, certificate deployment, and the real-world scenarios where this gets complicated. Let's go. [TECHNICAL DEEP-DIVE — approximately 5 minutes] Right, so the 802.1X framework has three components. You've got the Supplicant — that's the client device, the laptop, the phone, the IoT sensor. You've got the Authenticator — that's your access point or your network switch, sometimes called the NAS, the Network Access Server. And you've got the Authentication Server — almost universally a RADIUS server in enterprise deployments. Here's how the handshake works. When a device tries to connect to an 802.1X-protected SSID, the access point doesn't just let it on. Instead, it opens what's called a controlled port — a limited channel that only passes EAP traffic, the Extensible Authentication Protocol. The AP sends an EAP-Request Identity to the device. The device responds with its identity. The AP then forwards that to the RADIUS server, wrapped in a RADIUS Access-Request packet. The RADIUS server runs the authentication — checking credentials against Active Directory, a certificate store, or whatever identity backend you've configured — and sends back either an Access-Accept or an Access-Reject. Only on an Accept does the AP open the full data port and assign the device to the appropriate VLAN. Now, the EAP method you choose here matters enormously. There are five you'll encounter in enterprise deployments. EAP-TLS is the gold standard. Both the client and the server present X.509 certificates. There are no passwords involved. It's the most secure option and the one required for the highest PCI DSS compliance tiers. The catch is that you need a full PKI — a Public Key Infrastructure — to issue and manage client certificates. That means a Certificate Authority, certificate lifecycle management, and a mechanism to push certificates to every device. For organisations with Microsoft Active Directory and Active Directory Certificate Services, this is very achievable. For organisations without that infrastructure, it's a significant investment. PEAP-MSCHAPv2 is the most widely deployed method in practice. It creates a TLS tunnel using only a server-side certificate, then passes username and password credentials inside that tunnel. It's compatible with virtually every device out of the box, integrates directly with Active Directory via NPS on Windows Server, and doesn't require client certificates. The trade-off is that it's vulnerable to credential harvesting attacks if users are tricked into connecting to a rogue AP — because the client doesn't validate the server certificate by default. You must enforce server certificate validation in your supplicant profiles. EAP-TTLS is similar to PEAP but more flexible in the inner authentication method. It's common in Linux environments and where you need to support legacy authentication backends. EAP-FAST was developed by Cisco as a response to LEAP's weaknesses. It uses Protected Access Credentials rather than certificates. It's primarily relevant if you're in a Cisco-heavy environment or dealing with legacy devices that can't support the others. EAP-SIM and EAP-AKA are used in carrier-grade deployments — think OpenRoaming or Passpoint — where authentication is tied to a SIM card or USIM. These are increasingly relevant for public venue WiFi where you want seamless, secure onboarding without a captive portal. Now let's talk RADIUS configuration. Whether you're deploying Microsoft NPS, FreeRADIUS, Cisco ISE, or Aruba ClearPass, the core configuration steps are the same. First, you define your RADIUS clients — these are your access points or wireless LAN controllers. Each client is registered with its IP address and a shared secret. That shared secret is used to authenticate the RADIUS messages between the AP and the server. Use a minimum of 22 characters, randomly generated, and unique per NAS device. Second, you configure your network policy. This is where you define who gets access to what. In NPS terms, you're creating a Network Policy that matches conditions — group membership in Active Directory, device type, time of day — and assigns attributes — VLAN ID, session timeout, bandwidth limits. The RADIUS attribute you'll use most is VLAN assignment, specifically Tunnel-Type set to VLAN, Tunnel-Medium-Type set to 802, and Tunnel-Private-Group-ID set to your VLAN number. Third, you configure your connection request policy. This tells NPS how to handle incoming RADIUS requests — whether to authenticate locally or forward to another RADIUS server. In a distributed deployment, you might have a central RADIUS server with NPS proxies at each site. On the certificate side, for PEAP and EAP-TLS, your RADIUS server needs a server certificate trusted by your clients. The simplest path is to use a certificate from a public CA — DigiCert, Sectigo, Let's Encrypt — because those root certificates are already trusted by all major operating systems. If you're using an internal CA, you need to push the root certificate to all client devices via Group Policy or your MDM platform. For EAP-TLS specifically, you also need client certificates. In an Active Directory environment, you'd use ADCS with auto-enrolment via Group Policy to push certificates to domain-joined devices. For BYOD devices, you'd use your MDM — Intune, Jamf, VMware Workspace ONE — to push both the certificate and the WiFi profile. On the access point side, the configuration is straightforward. You create a new SSID, set security to WPA2-Enterprise or WPA3-Enterprise, point the RADIUS authentication server to your NPS IP on UDP port 1812, set the RADIUS accounting server on UDP port 1813, enter the shared secret, and enable dynamic VLAN assignment if you're using it. Most enterprise AP platforms — Cisco Meraki, Aruba, Ruckus, Extreme — have a GUI for this that takes about ten minutes once your RADIUS server is ready. [IMPLEMENTATION RECOMMENDATIONS AND PITFALLS — approximately 2 minutes] Right, let's talk about where deployments go wrong, because this is where I earn my consultancy fees. The most common failure point is certificate validation. I've seen organisations deploy PEAP-MSCHAPv2 correctly on the server side, then leave client supplicant profiles configured to accept any certificate. That completely undermines the security model. Every supplicant profile — whether pushed via Group Policy or MDM — must specify the trusted root CA and the expected server name. Without that, you're vulnerable to evil twin attacks. The second common issue is RADIUS shared secret management. I've seen production networks running with the shared secret set to "radius" or the vendor default. These secrets are the keys to your authentication infrastructure. Generate them randomly, store them in a secrets manager, and rotate them on a schedule. Third: VLAN misconfiguration. Dynamic VLAN assignment is powerful — it lets you put staff devices on the corporate VLAN, contractors on a restricted VLAN, and IoT devices on an isolated VLAN, all from the same SSID. But if the RADIUS attributes aren't configured correctly, or the switch trunk ports aren't carrying the right VLANs, devices will either fail to connect or land on the wrong segment. Test this thoroughly in a lab before rolling out to production. Fourth: redundancy. Your RADIUS server is now a critical piece of infrastructure. If it goes down, nobody connects. You need at minimum a primary and secondary RADIUS server configured on every AP. In large deployments, consider RADIUS proxy clusters with health monitoring. Fifth, and this is specific to hospitality and retail environments: guest versus corporate separation. Your 802.1X corporate SSID and your guest WiFi SSID should be completely separate — different VLANs, different firewall policies, different DNS. A platform like Purple handles the guest side with its own captive portal and analytics layer, while your 802.1X infrastructure handles the corporate side. These are complementary, not competing, systems. [RAPID-FIRE Q&A — approximately 1 minute] Let me run through the questions I get most often. Can I run 802.1X on a cloud-managed AP platform? Yes — Meraki, Aruba Central, and Ruckus Cloud all support it. You configure the RADIUS server details in the cloud dashboard, and the APs handle the EAP proxying. Do I need Active Directory? No. FreeRADIUS can authenticate against LDAP, SQL databases, flat files, or even REST APIs. But AD integration via NPS is by far the most common enterprise path. What about IoT devices that don't support 802.1X? Use MAC Authentication Bypass — MAB — as a fallback. The device's MAC address is sent to RADIUS as the username and password. It's not as secure as EAP, but it lets you onboard IoT devices while keeping them in a restricted VLAN. Does 802.1X work with WPA3? Yes. WPA3-Enterprise is essentially WPA3 with 802.1X authentication. It adds stronger encryption — 192-bit in the high-security mode — and is the recommended standard for new deployments. [SUMMARY AND NEXT STEPS — approximately 1 minute] So to bring this together: 802.1X is not a nice-to-have. For any organisation handling sensitive data, processing payments, or operating in a regulated environment, it's the baseline for enterprise WiFi security. The architecture is well-established, the tooling is mature, and the deployment path is clear. Start with your EAP method selection — PEAP-MSCHAPv2 if you need quick wins and broad compatibility, EAP-TLS if you have the PKI infrastructure and need the strongest security posture. Get your RADIUS server configured and redundant before you touch a single AP. Push your supplicant profiles via Group Policy or MDM before you go live. And keep your guest WiFi completely separate — use a purpose-built platform for that layer. If you're operating a multi-venue environment — hotels, retail chains, stadiums — the complexity scales with the number of sites, but the architecture doesn't change. The key is centralised RADIUS with site-local redundancy, and a consistent MDM-pushed supplicant profile across your device fleet. Thanks for listening. The full written guide, architecture diagrams, and configuration checklists are available at purple.ai. If you're planning an 802.1X deployment and want to talk through the specifics of your environment, reach out to the Purple team directly.