How is the security of IoT devices ensured?

How can IoT devices be secured?

Building protection across the connected system

Hello, friends. Securing Internet of Things devices means protecting the device, user account, home or business network, mobile app and cloud service together. A strong password is a useful starting point, but cannot fix unsupported software, an unnecessarily exposed management interface or unrestricted command permissions. The difference between IoT and IIoT also matters: a disconnected household lamp and a disrupted production control function have different consequences.

1. Establish what you are protecting and why

Confidentiality concerns unauthorised disclosure, integrity concerns unauthorised changes to readings and commands, and availability concerns loss of a required function. A data breach is only one possible outcome. Incorrect sensor readings and unexpected valve commands also deserve attention. NISTIR 8259A identifies device identification, configuration, data protection, interface access, software updates and cybersecurity-state awareness as baseline device capabilities.

The manufacturer supplies security capabilities, the installer configures them and the operator maintains them. Responsibilities must connect without gaps. In industrial cybersecurity, particularly, establish what function each device serves and who will respond when something goes wrong.

2. Ask security questions before purchasing

Investigate security-update support, the published support end date, unique credentials, data deletion and account recovery. A channel for reporting vulnerabilities to the manufacturer is also useful. ETSI EN 303 645 addresses these issues for consumer IoT; mentioning the standard does not itself demonstrate independent verification of a particular product. Support and replacement expenses contribute to long-term costs.

During installation, record the full model, hardware revision, running version, location, network connection and responsible person. Managing patches and permissions requires knowing which devices are present. An up-to-date mobile app does not establish that the device firmware is also current.

3. Separate human accounts from device identity

Use a strong, unique password for management accounts and enable multifactor authentication where supported. Give people individual accounts instead of sharing one service login, and remove access when someone leaves. Protect recovery methods as well. Usability contributes to security: if users cannot manage secure access and revert to a widely shared password, the installation is not sustainable.

For device-to-server connections, use supported per-device certificates, keys or access tokens. Human multifactor sign-in does not apply identically to every sensor. Aim to revoke and replace one compromised device's credentials without affecting the rest. A client name or serial number alone is not a secret authentication credential.

4. Permission to connect is not permission to do everything

Authentication establishes identity; authorisation determines permitted actions. For example, an MQTT temperature sensor can be limited to publishing its own measurement topic, without access to other devices' command topics. The OASIS MQTT 5.0 specification discusses restricting publish and subscribe permissions. Using MQTT does not configure these restrictions automatically. When establishing compatibility, define the necessary data flow rather than granting unnecessarily broad access.

5. Separate networks and control traffic between them

Separating IoT equipment from personal computers or critical controllers can limit the spread of a compromise. However, a separate SSID or VLAN is insufficient by itself; firewall and access rules must control traffic between the segments. Appropriate supported WPA2/WPA3 settings can protect Wi-Fi connections, but do not demonstrate that application data is encrypted throughout its journey.

Determine internet requirements by function: restrict unnecessary access when a device needs only specific services. Avoid exposing management interfaces directly to the internet. Provide remote access through a supported, authenticated method that is kept up to date. With continuous connectivity, unnecessary port forwarding and services should not remain enabled. After applying restrictions, test that updates, time synchronisation and required cloud connections still work.

6. Distinguish transport encryption from end-to-end encryption

TLS protects traffic between the parties to a connection. Use a supported modern TLS configuration and validate server identity and certificates. Disabling validation to bypass a certificate error weakens protection. SSL should not be recommended as a secure protocol option; SSLv3 has been deprecated. IETF RFC 8446 defines TLS 1.3.

Editor's note: Separate TLS connections from device to cloud and from cloud to app do not mean the cloud cannot read the data. If an MQTT broker can decrypt a message, transport protection has not concealed its contents from that broker. End-to-end content protection requires a separate design in which only the intended endpoints control the relevant keys and encryption.

7. Consider updates and secure boot together

Track security patches, the support period and the running version. Maintenance and updates require the correct hardware package, verification of its origin, a configuration backup and post-installation functional checks. On supported devices, signature verification and secure boot help prevent unauthorised software from running; a valid signature does not prove that software is free of defects. In production, test restart and output behaviour before deployment.

8. Include physical access and service interfaces

A device protected from network access can still be exposed to physical interference. Protect service ports, removable storage, reset buttons and setup information on labels as appropriate to its use. Restrict unused interfaces through supported settings, without unsupported modifications. Do not place private keys and passwords in project files or maintenance notes that everyone can read.

9. Collect less data and establish who can access it

For privacy and personal-data protection, establish what is collected, why it is needed, how long it is retained and who receives it. Keep unnecessary microphones, cameras, location access and remote sharing disabled. A function enabling continuous surveillance should not operate without limits merely because it is described as a security feature. Removing names does not necessarily anonymise data; activity times and device identifiers can still distinguish individuals. FTC guidance addresses data minimisation alongside access restrictions.

10. Prepare monitoring and recovery before an incident

Repeated failed logins, unexpected accounts, changed settings, unusual traffic and data gaps are signals to investigate, rather than proof of an attack on their own. Understanding normal device behaviour reduces false alarms. Logs should identify the device and time without recording passwords or access tokens. Establish who evaluates alerts and how quickly.

Plan access restriction, preservation of relevant logs, revocation of compromised credentials and recovery using a verified configuration. Restarting alone does not remove the cause of a compromise. Preserve necessary records before a factory reset; restoring the same weak settings and passwords can recreate the problem.

11. Coordinate industrial cybersecurity with physical safety

For example, a Modbus-connected gateway uploading data should not create an uncontrolled route into the control network. Assess read and write requirements separately. Disconnecting power or networking without considering the process can affect control, alarms or protective functions. NIST SP 800-82 Rev. 3 considers OT security alongside performance, reliability and safety requirements. Isolation and recovery must align with the facility's safe shutdown procedures.

12. Verify security throughout the service life

End of support, ownership changes and disposal are security events too. Plan replacement or safe isolation based on risk when security-update support ends. At disposal, follow the manufacturer's data-erasure procedure, unlink cloud accounts and revoke old keys. Electronic-waste management must also address data on the device and permissions held by external services.

At handover, obtain practical answers to these questions: Who can access the device? Can it issue commands to another device? What happens during an internet outage? Who responds to a vulnerability or fault? When evaluating a system that makes everyday work easier, look beyond its feature count. The aim is a manageable system that performs its required function with the least necessary privilege.

Related guides

Fundamentals: IoT guide, How IoT works and IoT versus IIoT.

Technical follow-up: Maintenance and updates, Data-breach risks, Continuous connectivity and cyberattacks and Device compatibility.

Data and service life: Privacy and personal data and Long-term costs.