WiFi invité d'hôpital : Expérience patient et séparation du réseau

Hospital Guest WiFi: Patient Experience and Network Separation A Purple Technical Briefing — approximately 10 minutes [INTRODUCTION — approximately 1 minute] Welcome to the Purple Technical Briefing series. I'm your host, and today we're tackling one of the most operationally sensitive WiFi deployments you'll encounter in enterprise networking: hospital guest WiFi. If you're a clinical IT manager, a hospital CIO, or a healthcare network engineer, you already know the stakes here are different from any other venue. This isn't a hotel where a guest can't stream Netflix. This is an environment where a misconfigured VLAN could theoretically put clinical systems — EHR platforms, infusion pumps, imaging equipment — on the same broadcast domain as a patient's smartphone. That is not a theoretical risk. It has happened. And the consequences range from regulatory breach to patient safety incidents. So today we're going to cover three things: how you architect complete separation between clinical and guest networks, how you deliver a genuinely good WiFi experience for patients and visitors, and how you measure whether it's working. Let's get into it. [TECHNICAL DEEP-DIVE — approximately 5 minutes] Let's start with the architecture. The fundamental principle of hospital WiFi design is that clinical and guest traffic must never share a Layer 2 broadcast domain. Full stop. This is non-negotiable under NHS Digital's Data Security and Protection Toolkit, and it aligns with HIPAA's technical safeguard requirements in the United States. The standard approach is VLAN segmentation. You assign a dedicated VLAN — let's call it VLAN 10 — to clinical systems: EHR workstations, nurse call systems, medical IoT devices, PACS imaging servers. A second VLAN — VLAN 20 — carries all guest and patient WiFi traffic. These VLANs are trunked across your switching infrastructure and terminated at a next-generation firewall, where inter-VLAN routing is either completely blocked or very tightly controlled with explicit allow rules. Now, here's where a lot of deployments go wrong. Teams assume that VLAN separation at the switch layer is sufficient. It isn't. You need to enforce this at three levels: the access layer, the distribution layer, and the firewall. If your access points are dual-SSID — broadcasting both a clinical SSID and a guest SSID — those SSIDs must map to separate VLANs with no bridging between them. Your wireless LAN controller must be configured to prevent client-to-client communication on the guest VLAN, and you should enable AP isolation as a default. That means a patient on bed seven cannot probe the device on bed eight, even though they're on the same guest SSID. Authentication is the next layer. On the clinical network, you want IEEE 802.1X with EAP-TLS or PEAP-MSCHAPv2, backed by a RADIUS server — Microsoft NPS, FreeRADIUS, or a cloud-based RADIUS service. Every clinical device should have a certificate or a domain credential. No PSK on clinical networks. Ever. Pre-shared keys are a single point of failure — one compromised credential and every device on that SSID is exposed. For the guest network, the model is different. You're dealing with patients who may be elderly, unwell, or not technically confident. The authentication experience needs to be simple. A captive portal with a one-click accept or a simple SMS verification is appropriate here. You are not going to ask a patient recovering from surgery to configure 802.1X on their personal device. What you can do is use WPA3-SAE on the guest SSID to ensure over-the-air encryption without requiring per-user credentials. WPA3 Simultaneous Authentication of Equals eliminates the pre-shared key vulnerability by using a zero-knowledge proof exchange, so even if someone captures the handshake, they cannot brute-force the passphrase offline. Now let's talk about bandwidth. This is where a lot of hospital IT teams underestimate the requirement. A single patient in a bed today might have a smartphone, a tablet, and a smart TV or bedside entertainment unit. They're streaming Netflix or BBC iPlayer, making video calls to family, and potentially using a hospital patient portal. Netflix HD requires five megabits per second. A 4K stream requires twenty-five. A video call on FaceTime or Teams requires between one and three megabits per second each way. So per bed, you should be planning for a minimum of twenty-five megabits per second of available throughput — and that's before you account for concurrency factors. In a two-hundred-bed hospital where sixty percent of patients are actively using WiFi at peak time — say, seven in the evening — you're looking at three gigabits per second of aggregate demand on the guest network. Your uplink capacity and your access point density need to be sized accordingly. The rule of thumb I use is: one access point per ward bay, not one per ward. In a six-bed bay, you want an AP within ten metres of every bed, operating on the five gigahertz band for throughput-sensitive clients, with the two-point-four gigahertz band handling legacy IoT devices and older handsets. Channel planning matters enormously in a hospital. You have dense RF environments — thick concrete walls, metal bed frames, medical equipment generating interference. Use a wireless site survey tool before deployment, not after. Plan your channel reuse pattern on the five gigahertz band using non-overlapping channels from the UNII-1 and UNII-3 bands. Set transmit power conservatively — you want cells to overlap by about fifteen to twenty percent, not fifty percent. Over-powered APs cause co-channel interference and actually degrade throughput. For the clinical network, the RF design considerations are even more critical because you're supporting real-time applications. VoIP on nurse call systems, telemetry streaming from patient monitors, and barcode scanning at medication dispensing all require low latency and consistent signal. Target minus sixty-five dBm RSSI at every clinical endpoint, with a signal-to-noise ratio above twenty-five decibels. [IMPLEMENTATION RECOMMENDATIONS AND PITFALLS — approximately 2 minutes] Let me give you the top three implementation pitfalls I see in hospital WiFi projects. First: assuming your VLAN configuration is correct without testing it. I've seen deployments where a misconfigured trunk port allowed guest VLAN traffic to leak onto the clinical VLAN. The way to catch this is a post-deployment penetration test — specifically, attempt to reach clinical subnet addresses from a guest client. If you can ping anything in the clinical range, your segmentation has failed. This should be a mandatory sign-off criterion before go-live. Second: neglecting the captive portal experience. Hospitals often treat the guest WiFi portal as an afterthought. But a poorly designed portal — one that times out, doesn't render on mobile, or requires too many steps — directly impacts patient satisfaction scores. In the United States, HCAHPS survey results include communication and environment scores that are influenced by WiFi quality. In the NHS, Friends and Family Test responses frequently cite WiFi as a factor. A platform like Purple's Guest WiFi solution gives you a branded, mobile-optimised portal with analytics built in, so you're not just providing connectivity — you're capturing data on usage patterns that inform capacity planning. Third: not having a bandwidth management policy. Without QoS and rate limiting on the guest network, a single patient running a BitTorrent client can saturate the uplink and degrade experience for everyone else. Implement per-client rate limiting — typically five to ten megabits per second download per device — and use DSCP marking to prioritise video call traffic over bulk downloads. Block peer-to-peer protocols at the firewall level on the guest VLAN. [RAPID-FIRE Q&A — approximately 1 minute] Let me run through some quick questions I get asked regularly. "Can we use the same physical access points for clinical and guest?" Yes, absolutely — dual-SSID APs are standard practice. The separation is logical, at the VLAN level, not physical. Just ensure your AP firmware supports VLAN tagging and that your WLC enforces the separation. "Do we need a separate internet uplink for guest traffic?" Not necessarily, but you should use traffic shaping to ensure clinical management traffic — software updates, remote access — is never starved by guest usage. A dedicated guest uplink is a belt-and-braces approach if budget allows. "How do we handle medical IoT devices on WiFi?" Medical IoT — infusion pumps, telemetry monitors — should be on a dedicated third VLAN, separate from both clinical workstations and guest devices. This limits blast radius if a device is compromised. "What about GDPR and data collected through the captive portal?" Any personal data collected at login — email, phone number — must be processed under a lawful basis, typically consent. Ensure your portal terms are clear, your data retention policy is documented, and you have a data processing agreement with your WiFi platform provider. [SUMMARY AND NEXT STEPS — approximately 1 minute] To wrap up: hospital guest WiFi is not just a connectivity project. It's a patient experience initiative, a compliance requirement, and a clinical safety consideration all rolled into one. The architecture is straightforward — VLAN segmentation, 802.1X on clinical, WPA3 on guest, captive portal for access, QoS for bandwidth management — but the execution requires rigour at every layer. Your next steps: commission a wireless site survey if you haven't done one in the last two years. Review your VLAN configuration and test inter-VLAN isolation. Benchmark your current patient WiFi satisfaction against HCAHPS or Friends and Family Test data. And if you're evaluating guest WiFi platforms, look at Purple's Healthcare solution — it pairs with their HIPAA compliance guide to give you a full picture of the regulatory landscape. Thanks for listening. Full technical documentation, architecture diagrams, and implementation checklists are available in the accompanying guide on the Purple website.