Zero Trust Architecture for the Internet of Things

Discover how Zero Trust Architecture is transforming IoT security, moving beyond outdated perimeter defenses to ensure every device, user, and service is authenticated and secure. Learn why adopting ZTA is essential for modern IoT design and how it can boost security, flexibility, and trust in your connected systems.

A similar change is now coming to the Internet of Things (IoT), even in the Smart Home. This white paper explains how IoT security is being transformed by Zero Trust Architecture, why this change is happening, and what IoT designers must do to adapt.

Problems with perimeter security

For many years, cybersecurity used perimeter-based security with firewalls. This approach harkens back to the time of castles and moats in the Middle Ages. By keeping the bad guys outside of a castle’s walls with a moat, the good (trusted) guys inside thought they were safe. When inside, no extra protection, such as armor, was necessary. How easy and efficient!

Unfortunately, perimeter security does not work well in modern cybersecurity. Attackers can use an email with a malicious link or a tempting ad on a web page to infect machines inside the perimeter. An infected machine becomes a base for observing unencrypted network traffic, scouting the target network, and infecting other machines. A more fundamental problem with the perimeter model is that users today spend much of their time outside the office and trusted services are often located in external clouds. Accommodating remote work in a perimeter security model requires extending the perimeter to include remote users and cloud services, which is a tremendous extension—and a substantial risk.

Even in those workplaces where everyone and everything is in the office, perimeter-based security means that it only takes one infected device or malicious user to cause a network compromise. This has always been a problem.

Zero Trust Architecture: The modern cyber security approach

Over the last 20 years, a much better approach for cybersecurity has been developed: Zero Trust Architecture (ZTA). With ZTA, nothing is blindly trusted. Everything must be verified before it can be trusted.

Here is a brief summary of the principles of ZTA [1]:

1. No network is trusted so all communications are secured.
2. All parties (users, services, devices, etc.) are authenticated.
3. Dynamic policies determine which access requests are authorized.
4. The security of users, networks, devices, software, and data is monitored.
5. When a problem is detected, it is addressed promptly.

These ZTA principles are simple, but they are revolutionary and have many advantages:

• Successful attacks are more difficult because easy targets are eliminated.
• Users, devices, and services can be located at any location desired.
• Any network can be used because all network traffic is secured.
• Each user, device, and service gets only the access to which it is entitled.
• Unauthorized parties get no access.
• If an attacker does compromise a user, network, device, or service, the compromise can be detected quickly leading to rapid containment and response.

The main challenges of ZTA are:

• Legacy systems are hard to change.
• Additional processing power is needed to encrypt and authenticate all messages.
• Monitoring the security of networks, devices, and software requires automation to scale.

Despite these challenges, the ZTA principles have been developed, tested, adopted, and refined over the last several decades across academia, the private sector, and the public sector. They have proven effective at reducing the impact of security compromises while increasing flexibility. And they continue to be refined.

Now ZTA is moving into widespread adoption. Government agencies have been especially vocal in advocating the use of ZTA. Back in 2019, the UK’s National Cyber Security Centre (NCSC) laid out the principles of ZTA and recommended adoption. In 2020, the US National Institute for Standards and Technology (NIST) further defined ZTA and detailed several options for ZTA deployment, including variations that permit ZTA to be overlaid onto legacy systems that do not support ZTA natively. For example, ZTA can be implemented in software agents and hardware gateways that protect legacy resources.

With the announcement on May 12, 2021 of Executive Order (EO) 14028 (Improving the Nation’s Cybersecurity), US Government Federal agencies were tasked with implementing a ZTA approach to security [2]. Every Federal agency is required to immediately develop a plan and start to transition to ZTA.

Across the private sector, ZTA adoption is also growing. A recent survey [3] shows that more than 86% of companies have started to implement ZTA. However, only 2% have fully implemented all of the ZTA tenets. Results so far are promising. Companies that have implemented all ZTA tenets see half the incidents as compared to those that have implemented none.

Zero Trust Architecture for embedded systems

When building embedded systems, security has often been an afterthought. System designers hoped that nobody would bother to hack embedded systems and they worried that including proper security would add cost and complexity.

However, several new trends are leading to the conclusion that building in good security from the start actually reduces cost and complexity while increasing likely sales.

Low-power processors such as the Infineon PSOC 4, PSOC 6, and PSOC Edge microcontroller units (MCUs) include built-in hardware cryptographic accelerators that make encryption easy and fast.

The recent rise in government regulations on and certification programs for IoT security [4] [5] increasingly block access to important markets unless products are security certified.

Customers, who have consistently said they want more secure IoT products, are learning to look for security certification labels when shopping and say that they are willing to pay more for products that disclose strong security practices using those labels [6].

Manufacturers are learning that patching a security problem when products are in the field is much more costly than fixing that problem when the products is in development or design [7].

New IoT security standards, like Matter [8], take a fresh look at IoT security. Because of this fresh approach, Matter includes features that support all of the ZTA principles.

With these trends, embedded systems developers are finding that strong security is no longer an expensive and thankless task but rather a best practice that boosts sales with little extra expense.

Zero Trust Architecture support in Matter

The Matter standard illustrates how the ZTA principles are implemented in modern IoT products:

1. No network is trusted so all communications are secured. The Matter standard defines secured communication protocols for typical IoT transactions and enable extensibility to support new transactions.
2. All parties (users, services, devices, etc.) are authenticated. Every Matter participant has a unique identity that is cryptographically authenticated in the Matter protocols.
3. Dynamic policies determine which access requests are authorized. Matter defines a standard format for access control policies and a standard method for dynamically revising those policies.
4. The security of users, networks, devices, software, and data is monitored. Matter includes hooks that can be used to add security monitoring capabilities, but these are not enabled by default. IoT owners must opt in.
5. When a problem is detected, it is addressed promptly.

As noted, Matter enables administrators to immediately revoke access privileges.

For embedded systems, Matter embodies the principles of ZTA. Now consider how it can be implemented in IoT and in embedded systems more broadly.

Designing Zero Trust Into IoT and embedded systems

Creating secured IoT and embedded systems requires a modernized, forward-looking approach. Consider each ZTA principle:

1. No network is trusted so all communications are secured. Using the Matter standard is a good way to get secured communication protocols. However, this is not always practical. In that case, choose a well-vetted, secured communication protocol such as DTLS or TLS.
2. All parties (users, services, devices, etc.) are authenticated. MCUs can support cryptographic authentication with a unique identity. For example, Infineon’s OPTIGA Trust M secure elements support this approach, including optional support for Matter. Infineon’s PSOC MCUs include Cloud ID which provides a unique identity and supports Matter connectivity.
3. Dynamic policies determine which access requests are authorized. Dynamic policies are implemented in application layer software. The open source Matter SDK (which runs on the PSOC 62S2 Wi-Fi BT Pioneer Kit and other platforms) includes support for these policies.
4. The security of users, networks, devices, software, and data is monitored. Security monitoring can be performed in the cloud, on a network gateway, or in devices. When performing system design, choose the optimal implementation strategy for each design.
5. When a problem is detected, it is addressed promptly.

To enable rapid response to security problems, build in remote management capabilities to revoke authorizations if a hostile user, device, or service is detected. Then move on to recovery by including cyber resilience capabilities such as a recovery mode that automatically downloads and installs the latest firmware.

The role of AI in Zero Trust Architectures

With the role of artificial intelligence (AI) increasing in both good guy and bad guy efforts, AI impacts ZTA in a few ways. One technique that has been used for more than a decade through neural networks (NNs) is constant monitoring. Instead of human monitoring, AI can thoroughly analyze logs, especially as a first pass. Training the NN to identify a suspicious vs. a not suspicious connection initiates the process. If the NN identifies a questionable situation, it can flag the user or person responsible for dealing with unauthorized access. This is most often associated with unauthorized credit card usage. Network intrusion detection also uses NN/AI for full-time monitoring for suspicious traffic. AI use in these types of applications will only increase and improve with increasing AI capabilities.

AI can initiate (be an active participant) by sending messages. In these instances, the AI itself needs authentication to ensure that it can be trusted just like any other user. For example, in a smart home, an AI node may automatically turn on lights as the sun sets or initiate some other welcome home action. As with other ZT aspects, the AI with its unusual capabilities needs to have fine-grained access control to avoid worst case scenarios. The rules must be established so AI knows its place and stays within its lane in the system.

Implementing zero trust in legacy IoT

Many enterprises still operate in the old security structure/model and transitioning to Zero Trust is not as simple as flipping a switch. As with any aspect of cybersecurity, implementation is all about risk. The highest risk situations need to be protected first. Higher risk items could be the payroll system, finance system, and/or intellectual property for an IT system or manufacturing company. No matter what the priorities are, they need to be identified and ranked to provide the right focus for the high-risk assets.

One of the first areas to address could be system administrators that typically have a very high level of access. In many organizations, all the system administrators share the same password—the root password. For Zero Trust, this is very dangerous because it is impossible to know who specifically performed a particular action, after the action has been performed. A starting point for establishing Zero Trust would be providing system administrators with their own unique credentials. Then, multi-factor authentication should be implemented to protect the administrator from cyber attackers. This type of attack is how the Ukrainian power grid was hacked in 2016—by obtaining the access password from a system administrator’s computer through a key-stroke logger.

Further protection for high-risk assets can be accomplished by always requiring mutual authentication for high-risk systems and having fine-grained access control—access to only the systems and operations that each person needs to do their job. No one is inherently trusted, so access must be carefully controlled and restricted.

Some legacy systems or subsystems cannot support ZTA and cannot be upgraded to do so. These systems should be put into a protected zone so that the only way to access them is through a firewall or security gateway. That gateway can implement all of the ZTA principles and protect the legacy system.

In some instances, two legacy devices in two different locations may need to connect to each other. For example, this could occur in industrial systems, where the control room must communicate across a distance with some legacy manufacturing equipment and neither supports a modern security protocol. In this case, they can be connected by a secure communications channel, with a gateway in front of each legacy device as shown in Figure 1. In this manner, the control room can communicate securely with the legacy manufacturing equipment even though both legacy systems are several decades old.

Trust Infineon for IoT security and ZTA

All of the ZTA aspects can be implemented with Infineon product lines such as the OPTIGA Trust M, OPTIGA Authenticate, and PSOC 64 line/64 Secure MCUs. As shown in Figure 2, the OPTIGA Trust M can easily and securely interface with a PSOC 6 device, such as the KBA23540. This applies to any PSOC 6 device and uses ModusToolbox software. Evaluation kits are available to support this process.

OPTIGA Trust M is the first security platform to receive the Cyber Security Agency of Singapore CLS-Ready certification [10]. Another embedded security option is the OPTIGA Trust M MTR—a special version of the OPTIGA Trust M that includes support for the Matter protocol.

Finally, the new OPTIGA Trust M IoT security kit provides the easiest way to develop and evaluate end-to-end security use cases for IoT devices. The kit enables customers to prototype various security use cases and build full-featured IoT applications.

Making the transition to a new trusted environment

Zero Trust Architecture is a ground-breaking approach to establish trust for users, devices, and software. With Zero Trust, security must be taken more seriously than ever before. In addition to industry associations providing direction and capabilities to establish greater trust, such as CSA’s Matter standard, when governments enter the picture, the need to implement their requirements rises to the next level. Fortunately, there are hardware-based solutions available now. For embedded systems designers who are ready to implement improved security, controllers and development tools in a variety of trusted families already exist and are continuously expanding to improve trust in IoT connectivity. 

RESOURCES
Infineon | www.infineon.com

REFERENCES
[1] Zero Trust Architecture, Published August 10, 2020, NIST SP 800-207
[2] Executive Order (E.O) 14028, Improving the Nation’s Cybersecurity
[3] Security Outcomes for Zero Trust: Adoption, Access, and Automation Trends
https://www.cisco.com/c/en/us/products/security/zero-trust-outcomes-report.html
[4] Consumer IoT Device Cybersecurity Standards, Policies, and Certification Schemes https://csa-iot.org/wp-content/uploads/2023/02/Consumer-IoT-Device-Cybersecurity-Standards-Policies-and-Certifica- tion-Schemes.pdf
[5] Certification Mark – U.S. Cybersecurity Labeling Program for Smart Devices https://www.fcc.gov/cybersecurity-certification-mark
[6] Are Consumers Willing to Pay for Security and Privacy of IoT Devices? https://iotsecurityprivacy.org/research/AreConsumersWilling
[7] Boehm, Barry & Papaccio, Philip. (1988). Understanding and Controlling Software Costs. IEEE Transactions on Software Engineering, Vol. 14, No. 10.
[8] Matter: The Foundation for Connected Things https://buildwithmatter.com
[9] OPTIGATM Trust M: Interfacing OPTIGATM Trust M with … – Infineon Developer Community
[10] Infineon’s OPTIGATM Trust M first to receive CLS-Ready certification from Cyber Security Agency of Singapore https://www.infineon.com/cms/en/about-infineon/press/market-news/2022/INFCSS202207-107.html

PUBLISHED IN CIRCUIT CELLAR MAGAZINE • JUNE 2025 #419 – Get a PDF of the issue

Sponsor this Article