Written by By Hans Brandl, Infineon Technologies
Personal Computers (PCs) with built in Trusted Computing (TC) technology are already well known and widely distributed in the marketplace with nearly all new notebooks containing a Trusted Platform Module (TPM), which can be used to increase trust and security features in a number of daily applications.
Trusted Computing (TC) has become an established technology for the verification and implementation of integrity and security applications on personal computers. Similar applications are also required for embedded computing platforms, which similar to personal computers, also have trust and security problems. The issues here are mainly due to the increasing complexity and consequential instability of Operation Systems (OS) and applications, in addition to Internet connections that are prone to security risks and attacks.
Until now, the TC standard has focused on PCs because of their large resource of available code space, specific bus interfaces and large computing power, elements which are not always available with embedded platforms. However, as there are an increasing number of embedded computing platforms like PCs in the field, it has become a necessity to adapt the current TC standard and implementation, to cover embedded platforms, such as mobile communication devices, industrial controls or automotives.
Trusted embedded systems
As with all new technology, a systematic approach should be taken for the development of trusted embedded systems to ensure that all components are considered:
- Trusted hardware and Trusted Platform Modules should be integrated as part of the VHDL design, which can then be adapted to different host systems together with processors to support trust architecture;
- Trusted operating systems should be based on the upcoming new virtualization / hypervisor architecture, which is already in use on PCs, adapting it to the specific requirements of small platforms and trusted modules.
- Security layers for implementing easy and accessible security mechanisms should be included
- Trusted protocols – Elementary TC protocols like TSS (Host interface API) and TNC (trusted network connect) an advanced secure communication protocol, will also be required for embedded platforms, but with a restricted amount of resources.
Trusted platform modules
One important point in the deployment of TPMs in PCs, is the activation of the TPM itself. Due to privacy reasons and misgivings, in a typical PC system TPMs are delivered in a deactivated state, which have to be activated at deployment by the new owner of the computer. Today, as the use of TPMs is increasing, especially in large organizations, ITC infrastructures require a different and more efficient method of handling the modules and the new Trusted Computing Management System (TCMS) [1] is often the first option for streamlining efficiency in large TPM infrastructures.
By combining existing TCG and PC standards, TPMs can now be activated in a secure and productive way, remotely by the organization’s ITC administrator, with the help of a dedicated TCMS server. Similar use of the TCG standard features, together with a TCMS server, enables new basic functionalities for the secure transfer and management of TPM stored certificates and other critical information. Based on these fundamental security mechanisms, secure and remote management, together with secure message handling, paves the way for new TPM applications and use scenarios.
For embedded applications, the same procedures are required, but on a much larger scale. For example, mobile phone communication protocols are usually controlled by the network provider – it’s not expected that the owner of a mobile phone should activate the security features via a TPM and enter the relevant details. A user expects to be able to use the system straight from the box, while the network provider does the rest. In addition to this, the owner expects that security parameters are set to a suitable level so that an attacker or third party is unable to interfere with it.
A possible future application could be the security and safety enhancements of future embedded platforms in complex, mission critical applications, such as automotive applications. It is not possible to manually activate a TPM based security feature during the production, sales or after sales stage because the manufacturer, salesperson or user doesn’t have the knowledge and understanding concerning such security induced handling and its reach. As a consequence, it is necessary to expand the first approaches of TPMs remote security management to a much broader product base at embedded applications and increase its flexibility to a very broad range of embedded security platforms for example. Further standardization of such functionality will also be necessary for the broad penetration into the fields of use.
Trusted computing in the future
The mobile phone is considered a future application for embedded trust, because of the following reasons for example:
- The network provider requires control over the network access, the authentication of the user and to control specific services;
- Modifications and enhancements to the platform’s hard and software (as well as additional software) may need authentication and verification from the integrity of the platform;
- The user may want to store private data (electronic purse, payment, access control applications) on the device and be confident that this cannot be accessed by anyone else. Similar requirements are also made by next generation automotive systems:
- A car manufacturer needs a trusted infrastructure for checking the integrity of the car system itself and may need a TPM function for starting a trusted virtual operating system, with error tolerance, for internal safety functions;
- The repair shop may need access to specific critical functions, such as engine and power features inside the car or the odometer value, which should not be accessible by the user or any other unauthorized third party;
- It is also interesting to note that future wireless car-to-car communication runs onlyin a safe and secure, authenticated manner.
All the application scenarios (and many more that arise in the future) may require independent instances of trust and security, however, in order to streamline costs, only one module should be implemented, which can carry out all of the scenarios.
One of the Trusted Computing Group Mobile Phone Work Group’s (TCG MPWG) first deliverables has created and published a specification [2], which offers new potentials for implementing trust in mobile computing platforms (but not only in these) by introducing a new, hardware-based trust anchor for mobile phones and devices.
This Mobile Trusted Module (MTM) has properties and features comparable to a standard TPM. The MPWG proposed a much more universal security architecture for mobile phones and devices on a higher abstraction level. The specification is called a TCG Mobile Reference Architecture (RA) and abstracts a trusted mobile platform as a set of tamper resistant trusted engines, operating on behalf of different stakeholders.
This architecture offers a high degree on flexibility from the design and implementation of the trusted components, to all participants in hard- and software development, also at non-mobile phone systems. An important aspect of the TCG Mobile Reference Architecture is the potential to make some parts of a trusted mobile platform virtual. The implementation of the MTM chip depends on the security requirements of its specific use-case. For high levels of protection and isolation, an MTM could be implemented as a slightly modified Trusted Platform Module (TPM) – this enables cost effective implementation of new security critical applications and various innovative business models, in both the mobile and generic computing domain.
A trusted mobile platform, based on a MTM, is characterized as a set of multiple tamper-resistant engines, each acting on behalf of a different stakeholder. Generally, such platforms have several major components:
- Trusted engines (TE)
- Trusted services (TS)
- Customization by trusted resources (TR)
All of these trusted engines can be implemented on the same piece of silicon and with today’s expertise in chip security, the different compartments and their data can be protected against each other to enable all functions to sit on one piece of silicon, and therefore create cost savings. Due to their flexible design, the new standards are not restricted to specific application areas, but allow new products, which require trust and security to take advantage of them.