Opinion: Security is the Achilles heel

8/16/2012 6:19 AM EDT

One potentially huge new market that is emerging from the convergence of cloud computing, machine-to-machine communications, wireless sensors, and the broadband Internet—both wireless and wired, is the so-called Internet-of-Things (IoT). The market got a boost recently when ARM helped form an industry forum to shape the IoT. If there any is doubt about ARM’s motive the observer needed only look back to March this year when ARM announced its Flycather CPU core designed especially for controlling the IoT.

While it is not obvious now, the one Achilles heel confronting this exploding market potential is security. The lack of security in the universe of machine-to-machine communications, notably industrial controllers was brought into clear focus by the U.S. Government’s successful cyber-attacks on Iran’s nuclear enrichment program using the Stuxnet computer worm.  According to the New York Times, the worm “temporarily took out nearly 1,000 of (Iran’s) 5,000 centrifuges.” The implication was clear that any machine connected to the Internet could be successfully attacked and its intended purpose subverted.

Security risks have never deterred the advancement of new technology such as the IoT as the prevalence of on-line commerce will attest. However, just as previous technology advances confronted the issues of security liability, so too, will the IoT likewise rise to the challenge. One of the first changes that must occur is in the way IoT System-on-Chip devices are designed and manufactured. 

In his remarks during the DAC Pavilion Panel discussion “Is "Lifecare" the Next Killer App?” Kristopher Ardis, Director of Business Development, Smart Grid Solutions at Maxim Integrated Products described this new methodology as designing security in from the start. “When you design a device it should do exactly what it was designed to do and nothing more,” he stated. “Engineers must think of all threats to a design from initial concept through completion of the design.”

One particular vulnerable element in the embedded controller within an IoT device is its program.  Software has traditionally been executed from programmable memory after the system is booted up using code stored in read-only memory (ROM) or non-volatile memory (NVM). This approach must be re-architected given the vulnerability demonstrated by recent cyber-attacks. 

To ensure that a device is resilient to cyber-attack, its control program should not be changeable in the field. One way to prevent changes is to execute the code from ROM or one-time programmable (OTP) NVM. The advantage of executing from ROM or NVM is that a hacker cannot change the code remotely. Furthermore, both provide security from hackers using physical attacks, however, of the two ROM is more vulnerable than NVM to physical attacks.

ROM is the absence or presence of a metal via deposited during the final stages of chip fabrication, which means that the program code is part of the metal mask layer of a chip—revisions require a new mask set. Thus, hackers using non-invasive techniques such as power analysis or more invasive reverse-engineering techniques such as de-processing and Focused Ion Beam analysis can read the contents of ROM memory.  This is the one advantage of OTP NVM based on anti-fuse technology. 

Storing a bit of data in anti-fuse bit cells occurs when a programming voltage is applied to the gate of a logic CMOS transistor to produce a breakdown in its gate dielectric, thus producing a logic “1” (changing a high resistance path into a low resistance).  This storage mechanism provides very high security.  Since the bit cell does not store a charge there is no physical evidence of its state. The bit determines an initial “0” or programmed “1” through the process of a very low sensing current, not voltage, thus making the cell less vulnerable to low-cost passive attacks—Glitching and Data Remanence—as well as semi-invasive attacks such as UV attacks, Fault Injection, and Voltage Contrast.

Finally, the anti-fuse NVM has one other benefit that contributes to increased security of an IoT device: its ability to be programmed during final test. The obvious plus is that one design can be configured for a number of different applications.  ROM would demand a unique mask set for each version.  The security advantage this provides over ROM is the ability to completely erase the memory upon detecting the devise has been or is being compromised. A programming voltage applied to the memory array turns all bits to “1”.

As the design community begins designing more IoT devices and realize the security vulnerabilities when these devices are installed in the field, the unique benefits of anti-fuse NVM will become more apparent. Thus, this technology that has provided secure storage of encryption keys in set-top boxes and other conditional access applications will offer similar capability to an entirely new generation of SoC designs. 

About the author

David Hsu is senior field marketing and applications manager at Kilopass Technology Inc. He graduated from The Ohio State University and received his Master’s degree in Electrical Engineering from Purdue University. Hsu started his career in Siemens Components in the telecommunication division; later at Datapath Systems’ read channel program and as principal engineer at LSI Logic HyperPHy SerDes group from1998 to 2006. Between 2006-2009, he worked at TSMC in the role of customer support and in IP/Library quality management.