Skip to main content

How to Implement Time and Bandwidth Restrictions on Guest WiFi

You will be able to set per-device and aggregate bandwidth caps, RADIUS session and idle timeouts, and SSID schedules for your guest WiFi. You can then choose values that fit your venue type, test each control from a real device, and fix the failures that most often stop limits from working.

By Tom HackettPublished Updated
📖 16 min read3,559 words3 worked examples3 practice questions12 key definitions

Video overview

Listen to this guide

View podcast transcript
How to Implement Time and Bandwidth Restrictions on Guest WiFi A Purple WiFi Intelligence Briefing [INTRODUCTION & CONTEXT - approximately 1 minute] Welcome to the Purple WiFi Intelligence Briefing. I'm your host, and today we're getting into something that sits right at the intersection of network performance, compliance, and guest experience - implementing time and bandwidth restrictions on guest WiFi. If you're running a hotel, a retail chain, a stadium, or a conference centre, this is one of the most operationally impactful decisions you'll make about your network. Get it wrong, and you've either throttled your guests into frustration, or you've left your corporate network exposed to a bandwidth free-for-all. Get it right, and you've got a scalable, compliant, and commercially intelligent guest access layer. In the next ten minutes, we're going to cover the technical architecture, the implementation steps, real-world case studies from hospitality and retail, the common pitfalls, and what good looks like from a business impact perspective. Let's get into it. [TECHNICAL DEEP-DIVE - approximately 5 minutes] Let's start with the fundamentals. When we talk about time and bandwidth restrictions on guest WiFi, we're really talking about two distinct but complementary policy layers - and understanding the difference is critical before you touch a single configuration screen. Bandwidth restrictions govern throughput. How many megabits per second can a single guest device consume? How much aggregate traffic can the entire guest SSID push through your uplink? These are enforced through Quality of Service mechanisms - specifically the IEEE 802.11e standard, which underpins WiFi Multimedia, or WMM. WMM defines four traffic access categories: voice, video, best effort, and background. Guest traffic should almost always be classified as best effort or background, ensuring your corporate and operational traffic retains priority. Time restrictions govern session duration. How long can a guest stay connected before they're required to re-authenticate? This is enforced at the captive portal layer, through session timeout parameters, and increasingly through RADIUS Change of Authorisation - CoA - which allows your authentication server to dynamically terminate or modify a session without requiring the client to disconnect and reconnect. Now, the architecture that makes all of this work cleanly is VLAN segmentation. Your guest SSID should live on a dedicated VLAN - let's call it VLAN 30 - completely isolated from your corporate network on VLAN 10 and your operational network on VLAN 20. The firewall sits between these segments and enforces inter-VLAN routing policies. Guest traffic on VLAN 30 should have no path to your internal servers, point-of-sale systems, or any device on the corporate LAN. This is not optional - it's a PCI DSS version 4.0 requirement under Requirement 1.3, which mandates network segmentation between payment environments and any network accessible to untrusted devices. Let's talk about the actual enforcement mechanisms. There are three primary approaches, and the right one depends on your infrastructure. The first is controller-based enforcement. If you're running a centralised Wireless LAN Controller - from Cisco, HPE Aruba, Juniper Mist, or similar - you can apply per-client and per-SSID bandwidth policies directly on the controller. A typical configuration for a hotel might set a per-client downstream cap of 25 megabits per second, an upstream cap of 5 megabits, and an aggregate SSID cap of 500 megabits to protect the uplink. Session timeouts are configured in the RADIUS attributes returned during authentication - specifically the Session-Timeout attribute, which tells the access point exactly how many seconds a session is valid for. The second approach is firewall-based policy enforcement. Platforms like Fortinet FortiGate, Palo Alto Networks, or pfSense allow you to apply traffic shaping policies at the firewall level, scoped to the guest VLAN. This is particularly useful in environments where the wireless infrastructure doesn't natively support per-client rate limiting, or where you need more granular control over application-layer traffic - for example, blocking peer-to-peer file sharing or video streaming during peak hours. The third approach is cloud-managed enforcement. Platforms like Purple, Cisco Meraki, and Juniper Mist push policy configurations from a central cloud dashboard to distributed access points. This is the preferred model for multi-site deployments - a retail chain with 200 stores, for example - because it eliminates the need for on-site configuration at each location. Policy changes propagate automatically, and you get centralised visibility into usage patterns across the entire estate. Now, let's talk about time-based scheduling, which is a slightly different concept from session timeout. Scheduling means the guest SSID itself is only active during defined hours. A retail store might only broadcast the guest SSID between 09:00 and 21:00, matching trading hours. Outside those hours, the SSID is suppressed entirely, reducing your attack surface and eliminating the risk of unauthorised access overnight. Most enterprise access points support SSID scheduling natively, and cloud-managed platforms make this trivial to configure across a large estate. One more mechanism worth calling out is data volume quotas - sometimes called daily data caps. Rather than restricting speed, you restrict total consumption. A guest gets, say, 500 megabytes per day. Once that quota is consumed, the session is either terminated or throttled to a very low speed - perhaps 1 megabit - sufficient for basic messaging but not for streaming. This is particularly effective in environments with constrained backhaul, such as remote hotels on satellite or fixed wireless connections. The technical standard underpinning all of this is IEEE 802.1X for port-based network access control, combined with RADIUS for authentication, authorisation, and accounting. The RADIUS server returns attributes that the access point or controller uses to enforce policy - including Session-Timeout, Idle-Timeout, and vendor-specific attributes for bandwidth limits. If you're running a cloud RADIUS deployment, Purple's platform integrates directly with your wireless infrastructure to deliver these attributes dynamically, based on the user's authentication method and the policies you've defined. [IMPLEMENTATION RECOMMENDATIONS & PITFALLS - approximately 2 minutes] Right, let's get practical. Here are the implementation steps I'd walk any client through. Step one: Define your policy matrix before you touch any hardware. For each venue type - hotel, retail, stadium, conference centre - define the session time limit, the per-client bandwidth cap, the aggregate SSID cap, the daily data quota, and the schedule window. Document this. It becomes your baseline configuration and your audit trail. Step two: Segment your network. If you don't have VLAN separation between guest and corporate traffic, stop everything else and fix that first. No bandwidth policy in the world compensates for a flat network where guest devices can reach your internal systems. Step three: Configure your captive portal with appropriate session parameters. Set the Session-Timeout RADIUS attribute to match your policy - 7,200 seconds for a two-hour session, for example. Enable idle timeout to reclaim sessions from devices that have disconnected without formally logging out. This is critical for capacity management in high-density environments. Step four: Apply per-client rate limits at the controller or access point level. Test these under load - not just with one device, but with a realistic number of concurrent clients. A 10-megabit-per-client cap feels generous when there are 5 guests, but when there are 200 guests in a conference room, your aggregate SSID cap becomes the binding constraint. Step five: Enable client isolation on the guest SSID. This prevents guest devices from communicating with each other over the wireless network, which eliminates a significant class of lateral movement attacks. Now, the pitfalls. The most common one I see is over-provisioning. Operators set generous bandwidth caps because they're worried about guest complaints, and then they're surprised when a handful of guests streaming 4K video saturate the uplink for everyone else. The right approach is to set conservative caps and monitor usage data. If your analytics show that 95% of guests are consuming less than 5 megabits, you can confidently tighten the cap without impacting the guest experience. The second pitfall is forgetting about MAC address randomisation. Modern iOS and Android devices randomise their MAC addresses by default, which means your per-device quotas and session tracking may not work as expected. Your captive portal and RADIUS infrastructure need to track sessions by authenticated identity - email address, phone number, or social login - rather than MAC address alone. The third pitfall is neglecting GDPR compliance. If you're collecting personal data at the captive portal as part of your authentication flow - and you should be, for accountability purposes - you need a lawful basis for that processing, a privacy notice, and a defined retention period for your session logs. Under GDPR Article 5, you cannot retain personal data longer than necessary for the purpose for which it was collected. [RAPID-FIRE Q&A - approximately 1 minute] Let me run through a few questions I get asked regularly. "What's the right bandwidth cap for a hotel?" For mid-scale properties, 15 to 25 megabits downstream per client is the sweet spot. Luxury properties should consider 50 megabits or higher, particularly if they're marketing themselves as business-friendly. "Should I use time limits or data quotas?" Use both. Time limits manage session concurrency. Data quotas manage throughput abuse. They solve different problems. "Can I apply different policies to different guest tiers?" Yes, and you should. A loyalty programme member who has authenticated via your app should get a better experience than an anonymous walk-in. RADIUS attributes can return different bandwidth profiles based on the user's tier. "What about WPA3?" Enable WPA3 Opportunistic Wireless Encryption on your guest SSID. It provides per-session encryption without requiring a password, which is exactly what you want for an open guest network. [SUMMARY & NEXT STEPS - approximately 1 minute] To wrap up: implementing time and bandwidth restrictions on guest WiFi is not a set-and-forget exercise. It's an ongoing operational discipline that sits at the intersection of network engineering, compliance, and guest experience management. The core principles are: segment your network with VLANs, enforce policy at the controller or firewall layer using RADIUS attributes, set conservative bandwidth caps and adjust based on usage data, use captive portal session timeouts to manage concurrency, and ensure your data collection practices are GDPR-compliant. If you're looking to go deeper on the authentication layer, Purple's guide on implementing 802.1X authentication with Cloud RADIUS is an excellent next step. And if you're evaluating your overall guest WiFi strategy, the Purple platform gives you the analytics and policy management tools to operationalise everything we've discussed today across your entire venue estate. Thanks for listening. Until next time.

Part of our core series: Guest WiFi Guide →

Interactive Technical Tool

Guest WiFi Time & Bandwidth Policy Calculator

Size per-user rate limits, session expiration timers, and DHCP subnets based on venue type and WAN uplink. Generates production CLI configs and RADIUS attributes.

High visitor turnover with bursty mobile usage. Prevents table camping while ensuring guests enjoy fast social media and streaming speeds without starving POS terminals.
50 Mbps1 Gbps2 Gbps
10 Devices5001,500+
Free tier capped at basic speed with premium upgrade pass.
Recommended down cap
5 Mbps
Upload: 2 Mbps
Session timeout
1.5 hrs
Idle Disconnect: 15 mins
DHCP subnet sizing
/23
510 Usable IPs (2h Lease)
Circuit utilisation
~30%
Peak demand: 150 Mbps (assumes 20% of users active at once)
# Cisco Meraki Dashboard Traffic Shaping & Splash Configuration
# SSID: Guest-WiFi | VLAN: Guest (DHCP Lease: 2h)
Wireless > Configure > Access control > Splash page:
  - Splash frequency: Every 90 minutes (or "Custom interval")
  - Client inactivity timeout: 15 minutes

Wireless > Configure > Firewall & traffic shaping:
  - Bandwidth limits per client:
    - Down: 5 Mbps
    - Up: 2 Mbps
  - Rules:
    - Restrict peer-to-peer / torrents: Deny
    - SpeedBurst: Enabled (5 seconds burst for fast page loads)
Useful? Link to this tool

You restrict guest WiFi in two layers. Your access points or firewall enforce bandwidth: a per-device cap (for example 10 Mbps down, 2 Mbps up) and an aggregate cap on the guest SSID. Your RADIUS server enforces time: Session-Timeout and Idle-Timeout attributes end sessions, and SSID scheduling switches guest access off outside opening hours.

What do time and bandwidth restrictions on guest WiFi actually do?

Restrictions stop guest traffic from crowding out the systems that run your venue. Without them, a handful of devices streaming high-resolution video can fill a shared broadband circuit. Your card terminals, booking system and staff calls then compete for whatever is left.

Bandwidth restrictions set how fast each device, and the guest network as a whole, can go. Time restrictions set three things:

  • How long a session lasts.
  • How long an idle device keeps its place.
  • When guest access exists at all.

Segmentation keeps guest traffic away from your internal network, whatever the limits are.

Three terms need defining before the detail:

  • RADIUS is the authentication protocol, defined in RFC 2865, that your access points use to ask a central server whether a device may connect, and on what terms.
  • VLAN (virtual LAN, IEEE 802.1Q) is a logically separate network carried over the same switches and cables.
  • Captive portal is the web page a guest sees before getting online.

Each control lives in a specific place and maps to a named standard or vendor feature.

Control Enforced by Standard or mechanism What it protects
Per-device rate limit Access point or controller Vendor feature, or RADIUS vendor-specific attribute Stops one device saturating the link
Aggregate guest cap Firewall or controller Traffic shaping on the guest VLAN Reserves WAN capacity for operations
Traffic priority Access point IEEE 802.11e (WMM) Puts staff voice and payments ahead of guest browsing
Session timeout RADIUS server, applied by access point RFC 2865, attribute 27 Controls concurrency and prompts re-authentication
Idle timeout RADIUS server, applied by access point RFC 2865, attribute 28 Frees IP addresses and airtime from departed devices
Mid-session change RADIUS server to controller RFC 5176 (Change of Authorisation) Enforces data allowances and tier upgrades
SSID schedule Controller or cloud dashboard Vendor feature Removes guest access outside opening hours
Segmentation and isolation Switch, firewall, access point IEEE 802.1Q plus client isolation Corporate network and PCI DSS scope

Where Purple fits

Purple Guest WiFi runs as a cloud overlay on the access points you already own. It provides the captive portal and the cloud RADIUS layer. Purple decides who connects and returns the session terms. Your hardware enforces those limits on the air and at the switch.

That split matters when you troubleshoot. A limit can be set correctly in RADIUS and still be ignored by firmware that does not support the attribute.

What do you need before you restrict guest WiFi?

Hardware that supports each control

Purple is hardware-agnostic. It works with Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet.

Before you design policy, confirm your firmware supports four features:

  • Per-client rate limits.
  • The RADIUS Session-Timeout and Idle-Timeout attributes.
  • Change of Authorisation (CoA).
  • SSID scheduling.

Purple's support articles for each manufacturer cover the configuration steps. Follow those rather than generic menu paths, because the menus differ between controller versions.

Numbers to measure first

Set limits from evidence, not habit. Gather four numbers before you touch a dashboard:

  • WAN capacity and peak utilisation. Pull 30 days of interface data from your firewall. Note the busiest hour, not the daily average.
  • Business traffic floor. This is the bandwidth your card terminals, property management or booking system, and staff voice need at peak.
  • Dwell time. Find out how long visitors actually stay. WiFi Analytics reports session length and visit frequency, so your timeouts reflect real behaviour.
  • Peak concurrent devices. Count the devices in the busiest hour, not across the whole day.

Network and compliance groundwork

  • A VLAN ID and a subnet sized for peak devices. A /24 gives 254 usable addresses, a /22 gives 1,022 and a /20 gives 4,094.
  • Firewall administration access and a change window. You also need a test phone and a laptop.
  • A publicly trusted TLS certificate for any portal hostname you host yourself.
  • A GDPR position. You need a lawful basis under Article 6, a privacy notice, and a retention period for session records under the storage limitation principle in Article 5(1)(e).
  • A PCI DSS map. Know where your cardholder data environment (CDE) sits and which subnets reach it.

How do you set up time and bandwidth limits on guest WiFi?

Work from the wire upwards: segmentation first, then caps, then identity and time. Each step depends on the one before it.

Step 1: Segment the guest network

Create a dedicated guest VLAN on your core switch. Trunk it to every access point that broadcasts the guest SSID.

Give the VLAN its own DHCP scope with short leases. Two to four hours suits hotels. 30 to 60 minutes suits high-churn venues such as stations and shopping centres.

Enable DHCP snooping and dynamic ARP inspection on your switches. Together they block rogue DHCP servers and address spoofing on the guest segment.

Step 2: Lock down the firewall and cap the aggregate

Write explicit rules that drop all traffic from the guest VLAN to every internal subnet. Permit only internet-bound traffic, plus the DNS and DHCP services guests need.

Then apply a traffic-shaping policy that caps the guest VLAN's total throughput. Set the cap at your circuit size minus your business traffic floor. That cap is your guarantee. Whatever guests do, the reserved bandwidth stays available to operations.

Step 3: Configure the guest SSID and per-device limits

  • Client isolation. Switch on client isolation, also called peer-to-peer blocking. Guest devices then cannot reach each other, which limits lateral malware spread and interception between devices.
  • Encryption. Enable Opportunistic Wireless Encryption (OWE, RFC 8110), part of WPA3. OWE encrypts each device's traffic individually on an open network, with no shared password. Use transition mode if older devices still need to connect.
  • Traffic priority. IEEE 802.11e defines WiFi Multimedia (WMM), which has four access categories: voice, video, best effort and background. Map guest traffic to best effort or background, so staff voice and payment traffic win contention for airtime.
  • Per-device rate limits. Set asymmetric caps such as 10 Mbps down and 2 Mbps up as a general starting point. Guest traffic is mostly downloads, so a lower upload cap costs guests little.
  • Bursting, where supported. Some platforms let a device exceed its cap for a few seconds, then throttle back. Pages load quickly while sustained load stays within the cap.

Step 4: Connect the captive portal and cloud RADIUS

Point the guest SSID at Purple's captive portal and cloud RADIUS, following the support article for your hardware. When a guest authenticates, RADIUS returns the session terms:

  • Session-Timeout (attribute 27). This sets the maximum session length in seconds before re-authentication. A value of 7,200 gives a two-hour session.
  • Idle-Timeout (attribute 28). This ends a session after a period with no traffic. A value of 900 ends it after 15 minutes of inactivity.

RADIUS can also carry bandwidth limits, but no single standard attribute exists for them. Manufacturers use vendor-specific attributes (attribute 26 in RFC 2865), so match the attribute to your hardware.

Finally, configure the walled garden. These pre-authentication rules let unauthenticated devices reach DNS, DHCP and the portal hostname. Everything else redirects to the login page.

Step 5: Add mid-session control with Change of Authorisation

Change of Authorisation, defined in RFC 5176, lets the RADIUS server alter a live session without disconnecting the device. Use it for two jobs:

  • Throttle a guest who has used their daily data allowance.
  • Lift a guest onto a faster tier after they log in with a loyalty account.

Data allowances also need RADIUS accounting (RFC 2866), which reports the bytes each session uses. Your controller must accept CoA messages. Your firewall must also let them through on UDP port 3799, the port RFC 5176 specifies.

Step 6: Schedule the guest SSID

Map the guest SSID to a time profile that matches your opening hours, plus a short buffer. Configure the access points to stop broadcasting outside those hours, rather than just refusing new connections. An SSID that is switched off gives nobody anything to connect to overnight.

Step 7: Add tiers if your venue needs them

Tiered access lets you reward loyalty without raising the baseline load. A free tier might allow 5 Mbps and one-hour sessions. A premium tier might allow 50 Mbps and 24-hour sessions, authenticated by room number or a loyalty account. CoA lifts the cap on the live session, so the guest never reconnects. Keep the free tier usable, because a free tier that feels broken generates complaints.

Got questions about your specific setup?

Our team works with venue operators, IT managers, and network engineers across 80,000 venues. Book a 20-minute call and we will show you how others like you solved it.

What limits should you set for your venue?

Start from dwell time and your business traffic floor, not from a vendor default. The values below are design starting points. Validate them against your own utilisation data in the first month.

Venue Plan for a stay of Session timeout Idle timeout Per-device cap (down/up) SSID schedule
Hotel One or more nights 24 hours 30 minutes 15/3 Mbps Always on
Retail store Under 1 hour 90 minutes 10 minutes 5/1 Mbps Store hours plus 30 minutes
Shopping centre 2 to 3 hours 3 hours 15 minutes 5/1 Mbps Centre opening hours
Conference centre 1 event day 10 hours 15 minutes 4/2 Mbps Event days only
Stadium Event plus arrival 5 hours 10 minutes 3/1 Mbps Event days only
Train 1 trip 3 hours 20 minutes 2/0.5 Mbps Whenever in service
Hospital Hours to days 24 hours (patients), 4 hours (visitors) 30 minutes 5/1 Mbps Always on
Library or council building Under 2 hours 2 hours 15 minutes 5/1 Mbps Building hours

The guest noun changes with the venue, and so does the policy logic:

  • Hotels. Guests stream and join video calls, so a 24-hour session avoids a daily login.
  • Retail. Shoppers stay briefly, so tight caps rarely affect them.
  • Trains. Passengers share one cellular backhaul across the whole train, which is why caps are the tightest in the table.
  • Healthcare. Separate patient and visitor profiles. A patient on a ward needs longer sessions than a visitor in the café.

Worked scenario: a 200-room hotel

The situation. A 200-room hotel runs a 500 Mbps circuit. It carries guest traffic, the property management system, card terminals and staff voice. Evening streaming pushes the circuit to its limit, and front-desk calls break up.

What was done.

  • An aggregate cap of 350 Mbps on the guest VLAN, which reserves 150 Mbps for operations.
  • A per-device cap of 15 Mbps down and 3 Mbps up.
  • A 24-hour session and a 30-minute idle timeout.
  • WMM mapping that places staff voice ahead of guest traffic.

The measurable outcome. Operations keep 150 Mbps at all times, 30% of the circuit. One device can never take more than 3% of the circuit. 23 devices can run at full speed before the aggregate cap engages. The test is simple: firewall reports should show business traffic never starved during the evening peak. You then decide on a circuit upgrade from data, not from complaints.

Worked scenario: a 40-store fashion retailer

The situation. Each store has a 100 Mbps broadband line shared by guest WiFi, card terminals and stock systems. Stores open from 09:00 to 18:00.

What was done.

  • A 40 Mbps aggregate cap and a 5/1 Mbps per-device cap.
  • A 90-minute session and a 10-minute idle timeout.
  • An SSID schedule of 08:30 to 18:30, rolled out as one template across all 40 sites.

The measurable outcome. Every store reserves 60 Mbps for transactions and stock systems, 60% of the line. The guest SSID is off for 14 hours a day, which removes guest access entirely outside trading. Short sessions also mean returning shoppers re-authenticate on each visit, which gives cleaner visit-frequency data.

Worked scenario: a conference centre hosting 2,000 delegates

The situation. A conference centre on a 1 Gbps circuit hosts a 2,000-delegate event. Planning assumes two devices per delegate.

What was done.

  • A /20 guest subnet with 4,094 meusable addresses, because a /22 would cover only 1,022 devices.
  • A 700 Mbps aggregate cap and a 4/2 Mbps per-device cap.
  • A 10-hour session (36,000 seconds) and a 15-minute idle timeout.
  • An SSID schedule active on event days only.

The measurable outcome. Every planned device can get an address, with headroom to spare. 175 devices can run at full rate at once before the aggregate cap shares capacity. That is enough for the bursty demand of session breaks. The idle timeout returns leases and airtime from delegates who have left the hall.

How do you check the restrictions work?

Test each control from a real device before you announce the change. Use a phone and a laptop on the guest SSID, and run these checks in order.

  1. Segmentation. From the guest SSID, try to reach an internal address such as a printer or a server. The attempt must fail.
  2. Client isolation. Connect two guest devices and try to reach one from the other. The attempt must fail.
  3. Per-device cap. Run a speed test. Results should sit at or just below your cap, allowing a brief burst if you enabled one.
  4. Aggregate cap. Run downloads on several devices at once. Confirm the firewall graph flattens at the guest cap while business traffic continues.
  5. Idle timeout. Leave a device inactive beyond the idle window. It should have to re-authenticate.
  6. Session timeout. Temporarily set a short session value on a test profile. Confirm the device re-authenticates at the right moment, then restore the production value.
  7. MAC randomisation. Forget the network and rejoin. If your limits are identity-based, the guest's time and data allowance should carry over.
  8. CoA. Trigger a tier change or an allowance threshold. The new speed should apply without a disconnect.
  9. Schedule. After closing time, scan for the SSID. It should not appear.

Then check the records. RADIUS accounting logs should show the Session-Timeout and Idle-Timeout values on accepted sessions. In WiFi Analytics, compare session length against your timeouts.

After go-live, review two signals monthly for the first quarter:

  • Peak utilisation on the WAN. Sessions that cluster at the timeout suggest guests want longer.
  • Re-authentication volume. A spike in repeat logins within minutes suggests the idle timeout is too aggressive.

What goes wrong with guest WiFi limits, and how do you fix it?

Guests reset their time limit by reconnecting

Apple introduced private addresses in iOS 14, and Android 10 randomises MAC addresses by default. If you key limits on MAC address alone, a randomised device looks brand new and the clock resets.

The fix is identity. Tie session time and data allowances to what the guest authenticates with at the captive portal, such as a verified email address, phone number or loyalty account. The RADIUS server then holds the session state, not the device.

New devices cannot get an address

In high-churn venues, 24-hour DHCP leases hold addresses for devices that left hours ago. New arrivals then find the pool empty. To fix it:

  • Shorten guest leases to 30 to 60 minutes.
  • Widen the subnet, for example from a /24 to a /20.
  • Enable DHCP release on disconnect where your controller supports it.

A shorter idle timeout also helps, because it ends abandoned sessions sooner.

The login page will not load

This is the most common guest complaint, and it usually has one of three causes:

  • DNS blocked before login. The walled garden is missing DNS, so the device cannot resolve the portal hostname. Permit DNS before authentication.
  • Untrusted certificate. A self-signed or expired certificate on a portal hostname triggers browser warnings that block the page. Use a publicly trusted certificate and renew it before expiry.
  • HTTPS start page. Browsers cannot redirect an HTTPS page without a certificate error. Rely on the operating system's captive portal detection, which uses plain HTTP probes, rather than a guest's homepage.

The limits are set but not enforced

You see the right values in RADIUS, but speeds stay uncapped. The usual cause is a vendor-specific attribute your hardware does not recognise, or firmware that ignores it.

Confirm the attribute matches your manufacturer, and check the access point's own client view. If CoA changes never apply, look for a firewall rule blocking UDP port 3799. Also check whether the controller has CoA disabled.

Time limits annoy guests more than they protect you

Short sessions in a hotel or hospital generate complaints and front-desk calls. Match session length to the stay, and use idle timeouts to recover resources instead. Re-authentication for a returning, identified guest should be a single tap, not a fresh form.

What do guest WiFi restrictions cost, and what do you get back?

What it costs

  • Hardware. Rate limiting, client isolation, WMM and SSID scheduling are features of the access points you already run. You need no new hardware, and nothing is ripped out and replaced.
  • Engineering time. Budget for design, a change window and testing.
  • Licensing. The captive portal, cloud RADIUS and analytics layer sit within Purple's Connect, Capture and Engage plans.

What you get back

  • WAN headroom. An aggregate cap turns a circuit upgrade from a reaction into a planned decision. You upgrade when utilisation data says so, not when guests complain.
  • Protected transactions. Segmentation and WMM priority keep card payments and staff voice ahead of guest traffic, even at peak.
  • Smaller compliance scope. PCI DSS v4.0.1 does not mandate segmentation. Effective segmentation can, however, keep the guest network out of scope. Requirement 1.3 restricts network access to and from the CDE. If you rely on segmentation, Requirement 11.4.5 requires penetration testing of those controls at least every 12 months and after any change.
  • First-party data. Session limits create regular, legitimate re-authentication points. With conscious-choice opt-ins, guests who consent join your CRM, and Purple integrates with CRM and marketing platforms. Guests who decline still get online.

Proof behind the platform

Purple runs at 80,000+ live venues in more than 90 countries, with 500 million users and logins a year. The platform is ISO 27001 certified and GDPR compliant. Purple is also a certified B Corp.

Frequently asked questions

Do you need new access points to limit bandwidth on guest WiFi?

No. Per-client and per-SSID rate limits are standard on enterprise access points from Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet. For time limits and mid-session changes, check that your firmware supports the RADIUS Session-Timeout attribute and Change of Authorisation. Purple runs as a cloud overlay on that hardware, so nothing is ripped out and replaced.

Will Purple work with the guest network you already run?

Yes. Purple is hardware-agnostic and layers its captive portal and cloud RADIUS on top of your existing access points and controllers. You keep your VLANs, firewall rules and rate limits. You repoint the guest SSID's authentication to Purple, following the support article for your manufacturer. Purple is ISO 27001 certified and runs at 80,000+ live venues in more than 90 countries.

Can guests get round a time limit by reconnecting?

Not if you tie session state to identity. iOS 14 and later, and Android 10 and later, use randomised MAC addresses, so a device can look new each time it joins. If your limits rely on MAC address alone, reconnecting resets the clock. Link session time and data allowances to what the guest authenticates with at the captive portal, such as a verified email address or loyalty account.

Does putting guest WiFi on its own VLAN take you out of PCI DSS scope?

It can reduce scope, but only if the segmentation is effective and tested. PCI DSS v4.0.1 does not require segmentation, but without it, systems sharing a network with cardholder data can fall into scope. If you rely on segmentation, Requirement 11.4.5 requires penetration testing of those controls at least every 12 months and after changes. Your Qualified Security Assessor makes the final scoping decision.

Do session time limits affect GDPR compliance?

Time limits do not, but the data collected at each login does. Every re-authentication must sit under the lawful basis and privacy notice you showed at first login. Session records linked to an identifiable guest are personal data, so set a retention period and enforce it. Purple uses conscious-choice opt-ins, so marketing consent stays separate from network access.

Should you cap data or cap time?

Cap speed and time for most venues, and add data allowances only where backhaul is metered or scarce. Per-device rate limits and session timeouts need only standard access point features and RADIUS attributes. Data allowances also need RADIUS accounting to count bytes, plus Change of Authorisation to throttle a live session. Trains and remote sites on cellular backhaul benefit most from data allowances.

What per-device speed should a hotel set?

Start at 15 Mbps down and 3 Mbps up per device, then adjust against your own utilisation data. Hotel guests stream and join video calls, so caps designed for retail feel slow in a room. Keep an aggregate cap on the guest SSID that reserves bandwidth for your property management system, card terminals and staff voice. Review peak utilisation monthly for the first quarter.

Can you apply the same restrictions across every site in your estate?

Yes. Cloud-managed controllers let you push SSID, rate-limit and schedule settings to many sites as a template. Purple applies one captive portal and RADIUS policy across your estate from a single pane of glass. Set per-site exceptions only where circuit size or opening hours differ, and record each one so audits stay straightforward.

Key Definitions

RADIUS

Remote Authentication Dial-In User Service, defined in RFC 2865. Access points ask a central server whether a device may connect and receive the session terms as attributes in the reply. Attribute 26 carries vendor-specific attributes.

Purple's cloud RADIUS returns the time limits and, through vendor-specific attributes, bandwidth limits. Your hardware must recognise the attribute or the limit is ignored.

Session-Timeout

RADIUS attribute 27 in RFC 2865. It sets the maximum session length in seconds before the device must re-authenticate; a value of 7,200 gives a two-hour session.

Your main time control. Set it from measured dwell time, for example 24 hours in a hotel or 90 minutes in a retail store.

Idle-Timeout

RADIUS attribute 28 in RFC 2865. It ends a session after a set number of seconds with no traffic; a value of 900 ends it after 15 minutes of inactivity.

Frees IP addresses and airtime from devices that have left. Too aggressive a value shows up as a spike in repeat logins within minutes.

Change of Authorisation (CoA)

Defined in RFC 5176. It lets the RADIUS server alter a live session without disconnecting the device, with messages sent on UDP port 3799.

Needed to throttle a guest who has used a daily allowance or lift a loyalty guest onto a faster tier. A firewall rule blocking UDP 3799 is a common reason CoA never applies.

RADIUS accounting

Defined in RFC 2866. The access point or controller reports usage for each session, including the bytes transferred, to the RADIUS server.

Required for data allowances, and the logs let you confirm Session-Timeout and Idle-Timeout values on accepted sessions.

VLAN

Virtual LAN, specified in IEEE 802.1Q. A logically separate network carried over the same switches and cables, identified by a VLAN ID and trunked to access points.

The guest VLAN is the foundation for firewall isolation, the aggregate cap and keeping guest traffic away from your cardholder data environment.

WiFi Multimedia (WMM)

Traffic prioritisation defined in IEEE 802.11e, with four access categories: voice, video, best effort and background.

Mapping guest traffic to best effort or background lets staff voice and payment traffic win contention for airtime at peak.

Opportunistic Wireless Encryption (OWE)

Defined in RFC 8110 and part of WPA3. It encrypts each device's traffic individually on an open network with no shared password; transition mode supports older devices.

Gives guests encrypted connections without a password on the sign, while the captive portal still handles login.

Captive portal and walled garden

The captive portal is the web page a guest sees before getting online. The walled garden is the set of pre-authentication rules allowing DNS, DHCP and the portal hostname, with everything else redirected to login.

A missing DNS entry, an untrusted TLS certificate or an HTTPS start page are the usual reasons the login page will not load.

Client isolation

An access point setting, also called peer-to-peer blocking, that stops devices on the same SSID from reaching each other.

Limits lateral malware spread and interception between guest devices. Test it by trying to reach one guest device from another.

MAC randomisation

Private, randomised MAC addresses used by default from iOS 14 and Android 10, so a device can present a new address each time it joins.

Limits keyed on MAC address alone reset on reconnect. Holding session state against the authenticated identity in RADIUS keeps allowances intact.

PCI DSS segmentation

PCI DSS v4.0.1 does not mandate segmentation, but Requirement 1.3 restricts network access to and from the CDE, and Requirement 11.4.5 requires penetration testing of segmentation controls at least every 12 months and after changes.

Effective, tested segmentation can keep the guest network out of PCI scope. Your Qualified Security Assessor makes the final scoping decision.

Worked Examples

A 200-room hotel runs guest traffic, its property management system, card terminals and staff voice over one 500 Mbps circuit. Evening streaming fills the circuit and front-desk calls break up.

The hotel capped the guest VLAN at 350 Mbps, which reserves 150 Mbps for operations at all times, or 30% of the circuit. Each device was capped at 15 Mbps down and 3 Mbps up, so one device can never take more than 3% of the circuit. A 24-hour session avoids a daily login for guests who stream and join video calls, and a 30-minute idle timeout recovers resources. WMM mapping put staff voice ahead of guest traffic. 23 devices can run at full speed before the aggregate cap engages. Firewall reports then show whether business traffic stays protected at the evening peak, so any circuit upgrade is decided from data.

A 40-store fashion retailer shares a 100 Mbps line per store between guest WiFi, card terminals and stock systems. Stores trade from 09:00 to 18:00.

Each store received a 40 Mbps aggregate guest cap and a 5/1 Mbps per-device cap, reserving 60 Mbps, or 60% of the line, for transactions and stock systems. Shoppers stay briefly, so a 90-minute session and a 10-minute idle timeout fit their behaviour. The guest SSID runs from 08:30 to 18:30, rolled out as one template across all 40 sites. That removes guest access for 14 hours a day outside trading. Short sessions also mean returning shoppers re-authenticate on each visit, which gives cleaner visit-frequency data.

A conference centre on a 1 Gbps circuit hosts a 2,000-delegate event, and planning assumes two devices per delegate.

Around 4,000 devices would overrun a /22, which covers only 1,022 addresses, so the team deployed a /20 guest subnet with 4,094 usable addresses. The guest VLAN was capped at 700 Mbps with a 4/2 Mbps per-device cap. That lets 175 devices run at full rate at once before the aggregate cap shares capacity, enough for bursty demand during session breaks. A 10-hour session (36,000 seconds) covers the event day. A 15-minute idle timeout returns leases and airtime from delegates who leave. The SSID runs on event days only.

Practice Questions

Q1. A major retail shopping mall experiences frequent DHCP IP address exhaustion on its guest WiFi network during peak weekend hours. The current configuration uses a /24 subnet (254 available IPs) with a 24-hour DHCP lease time. How should the network architect resolve this issue without expanding the hardware infrastructure?

Hint: Consider the relationship between average dwell time, DHCP lease duration, and the size of the logical subnet.

View model answer

The network architect should implement two immediate changes:

  1. Reduce the DHCP lease time from 24 hours to 30 or 60 minutes. Since the average dwell time in a shopping mall is 1 to 2 hours, a short lease time ensures that IP addresses are rapidly reclaimed from departed devices and returned to the pool.
  2. Expand the DHCP scope by changing the subnet mask from a /24 to a /21 (providing 2,046 available IPs) or /20 (providing 4,094 available IPs). This increases the logical size of the IP pool on Guest VLAN 30 without requiring any new physical switches or access points.

Q2. An IT manager notices that several users on the guest WiFi network are consistently bypassing the 500 MB daily data quota. The network uses MAC-based tracking to enforce quotas. How are the users likely bypassing this restriction, and what is the recommended enterprise-grade solution?

Hint: Modern mobile operating systems rotate their physical identifiers automatically.

View model answer

The users are bypassing the quota by utilising MAC Address Randomisation, a native privacy feature on modern iOS and Android devices. By toggling their WiFi connection off and on, or modifying their device settings, they generate a new randomised MAC address, which the network access point treats as a brand-new device with a fresh 500 MB quota. The recommended solution is to transition from MAC-based session tracking to Identity-Based Session Tracking. Configure the captive portal to require user authentication (e.g., email verification, SMS OTP, or social login). Associate the data consumption quota with the user's authenticated identity in the centralised RADIUS/policy database. When a user connects, regardless of what randomised MAC address their device presents, they must log in, and their session will be mapped to their unique identity, enforcing the 500 MB daily limit across all MAC addresses they use.

Q3. A hotel chain wants to ensure its guest wireless network complies with PCI DSS v4.0. During an audit, the QSA (Qualified Security Assessor) discovers that the hotel's property management system (PMS) and guest WiFi are on different subnets but connected to the same physical switches without firewall rules blocking inter-subnet traffic. What is the compliance risk, and how should it be remediated?

Hint: PCI DSS requires logical segmentation to be actively enforced, not just defined by subnets.

View model answer

The compliance risk is that the guest WiFi network is not segmented from the Cardholder Data Environment (CDE) where the PMS resides. In a flat physical network with inter-subnet routing enabled and no firewall restrictions, any guest device on the WiFi can route traffic directly to the PMS server. This brings the entire guest WiFi network into the scope of the PCI audit, representing a critical non-compliance finding. To remediate this:

  1. Enforce strict VLAN segmentation on the switches. Assign the guest WiFi to a dedicated VLAN (VLAN 30) and the PMS/CDE to a separate secure VLAN (VLAN 100).
  2. Implement firewall policies at the gateway/router level. Configure explicit Access Control Lists (ACLs) or firewall rules that drop all traffic originating from VLAN 30 destined for VLAN 100.
  3. Enable stateful packet inspection and perform regular penetration testing to verify that no guest device can establish a connection to any device within the CDE, thereby officially segmenting the guest network out of the PCI audit scope.

Frequently asked questions

How do I set session time limits for guest WiFi users?

Session time limits are enforced by sending the Session-Timeout RADIUS attribute (RFC 2865) in the Access-Accept packet during captive portal authentication, or by configuring inactivity and session timers directly on your wireless controller (such as Cisco Meraki, ArubaOS, or UniFi). Once the timer expires, the gateway deauthenticates the client or redirects them back to the captive portal for re-authorisation.

What is the recommended bandwidth limit per guest WiFi user?

For high-turnover public venues like cafes and retail stores, 5 Mbps download and 2 Mbps upload per user provides responsive 1080p video streaming, web browsing, and messaging without monopolising airtime. In hospitality or enterprise coworking spaces, offering 15 to 25 Mbps download with 5 to 10 Mbps upload ensures glitch-free video conferencing and VPN performance. Enforcing per-user QoS token buckets prevents single users from consuming the entire WAN uplink.

What is the difference between Session-Timeout and Idle-Timeout in guest WiFi?

Session-Timeout specifies the maximum total duration (in seconds) that an authenticated client may remain on the network before mandatory re-authorisation, regardless of traffic activity. Idle-Timeout (inactivity timer) monitors user data packets and disconnects a device if no traffic is transmitted for a set period (typically 10 to 15 minutes), freeing up DHCP leases and AP association table entries when visitors leave the venue.

How do you prevent guests from bypassing session limits with MAC address randomisation?

Modern mobile devices (iOS, Android, Windows) rotate private MAC addresses. Relying exclusively on physical MAC addresses allows users to cycle their connection and reset timers. Enterprise platforms like Purple tie guest sessions to verified identity credentials (SMS one-time passwords, social logins, or corporate emails) or issue Passpoint (Hotspot 2.0) encrypted profiles, ensuring time and bandwidth quotas persist across hardware address changes.

How does RFC 8910 (Captive Portal API) communicate time and data quotas to client devices?

RFC 8910 standardizes how wireless networks communicate session parameters to client operating systems via DHCP Option 114 and IPv6 Router Advertisements. Rather than intercepting DNS queries or dropping packets upon expiry, the gateway advertises an API endpoint that informs the device of remaining time or byte allowances, allowing native OS notifications and frictionless renewal splash screens.

Which RADIUS attributes enforce bandwidth rate limiting on enterprise access points?

Enterprise access points and controllers parse vendor-specific RADIUS attributes returned in the Access-Accept packet. The most common standard attributes include WISPr-Bandwidth-Max-Down and WISPr-Bandwidth-Max-Up (used by Meraki, Ruckus, and UniFi), Aruba-Bandwidth-Contract (ArubaOS), and Cisco-AVPair='rate-limit:downlink/uplink' (Cisco IOS-XE / Catalyst).

How does Purple automate guest WiFi bandwidth tiering and monetization?

Purple provides a cloud-managed captive portal and identity platform that integrates seamlessly with your existing AP infrastructure. It allows venues to offer tiered access models - such as a free 60-minute basic tier alongside a paid or loyalty-gated high-speed premium tier - automating payment processing, voucher generation, and RADIUS rate-limit assignment without manual network reconfiguration.

Got questions about your specific setup?

Our team works with venue operators, IT managers, and network engineers across 80,000 venues. Book a 20-minute call and we will show you how others like you solved it.