A SURVEY OF INFORMATION SECURITY IMPLEMENTATIONS FOR THE INTERNET OF THINGS By Arlen Baker, Principal Technologist, Wind River WHEN IT MATTERS, IT RUNS ON WIND RIVER A SURVEY OF INFORMATION SECURITY IMPLEMENTATIONS FOR THE INTERNET OF THINGS ABSTRACT This paper examines the implementations of the well-known information security1 components of confidentiality, integrity, and availability (the CIA triad2) as applied to the Internet of Things (IoT), and how these implementations can be used to defend against various attacks. The approaches taken by this paper are widely applicable to IoT devices in a variety of mar- kets, including aerospace3, automotive4, defense5, industrial6, medical7,8, and networking9, and are directly applicable to the protection of the intellectual property (IP) of the vendor. The CIA Triad is authoritatively defined in: United States Code, 2006 Edition, Supplement 5 Title 44 - Public Printing and Documents Chapter 35 - Coordination of Federal Information Policy Subchapter III - Information Security Section 3542 - Definitions TABLE OF CONTENTS The CIA Triad . 3 Defense-in-Depth Approach . 3 Confidentiality for IoT . 3 Privacy Implementations . 3 Separation Implementations . 5 Key Management Implementations . 6 Integrity for IoT . 7 Data Integrity Implementations . 7 Boot Process Implementations . 7 Authentification/Authorization/Accounting (AAA) Implementations . 9 Availability for IoT . 11 Whitelisting Implementations . 11 Intrusion Protection Implementations . 12 IoT Device Management Implementations . 14 Countermeasure Implementations . 15 Case Studies—How to Apply the CIA Triad . 16 Case Study: Aerospace Market . 16 Case Study: Medical Market . 16 References . 17 2 | White Paper A SURVEY OF INFORMATION SECURITY IMPLEMENTATIONS FOR THE INTERNET OF THINGS THE CIA TRIAD DEFENSE-IN-DEPTH APPROACH The CIA triad is the foundational security principle for the protec- No single security principle by itself could provide complete pro- tion of an asset. Its three components can be thought of as similar tection for an IoT device. Rather, it is the proper layering of these to the components of security for the contents of a home: defenses that will provide a much stronger, multifaceted protec- tion for the IoT device. The concept of layering these principles • Confidentiality is defined as maintaining theprivacy of an together is known as defense in depth.10 asset. Solid doors, walls, and window coverings provide privacy for the contents of a residence. Many factors dictate the security components that need to be • Integrity is defined as maintaining thecontent of the asset. included to protect an IoT device; the security assessment will An alarm system, a fence, and locks on the doors and windows uncover the required components. maintain the integrity of a residence, such that the contents of the residence are kept intact. CONFIDENTIALITY FOR IOT • Availability is defined as theaccessibility of the asset. The con- Confidentiality implementations are used to protect the privacy tents of the residence are available to the residents via pass- of data in IoT. This protection includes data passing throughout codes to the alarm system and keys to the door locks. IoT (data in motion), data that are stored on the IoT device such as on disk drives and/or in non-volatile memory (data at rest), and The CIA triad can be further broken down into sub-principles, which data that are being processed on the IoT device (data in process). can then be broken down into implementations, as shown in Figure Confidentiality can be partitioned into three sub-principles: pri- 1. The remainder of this paper will discuss these sub-principles and vacy, separation, and key management (as shown in Figure 2 along how each can be used to secure an IoT device. with their associated implementations). Application of the CIA triad begins with the security assessment. The security assessment determines which CIA implementations Confidentiality are required based on vulnerabilities, risks, regulatory require- ments, and IP protection needs, and balances those needs against cost, performance, and the operational environment. The security Key Privacy Separation assessment will provide the Security Policy, which defines the secu- Management rity objectives for the IoT device: what the security-related events are, how they are to be constrained, when they are to be reported, and what actions to take in response to the events. The security Data in Motion Partitioning Key Generation assessment also provides the processes within the development cycle to assure that the security-related principles are implemented. Covert Key Data at Rest Channels Distribution Security Assessment Security Policy Confidentiality Integrity Availability Sanitization Privacy Data Integrity Countermea- sures Separation Boot Process Whitelisting Figure 2. Confidentiality implementations Key Authentication Intrusion Management Authorization, Privacy Implementations & Accounting Protection Management Privacy is achieved through the use of cryptographic algorithms Development Processes on the data (encryption), making it nonsensical to a non-autho- rized individual. An authorized individual can restore the data to Trusted Platform its original form (decryption). Just as there are different types of Figure 1. CIA triad principles 3 | White Paper A SURVEY OF INFORMATION SECURITY IMPLEMENTATIONS FOR THE INTERNET OF THINGS door locks11, there are different types of cryptographic algorithms Asymmetric cryptographic algorithms are also called public key based on the need. Types of cryptographic algorithms used for algorithms. This type of algorithm requires two keys (a key pair): confidentiality include the following12: one that is kept private and one that can be made public. The private key is kept tightly protected and is accessible by as few • Symmetric algorithms individuals as possible. The public key can be accessible by oth- – Stream cipher: Processing data one datum at a time ers, but does require a level of protection in an IoT environment, – Block cipher: Processing data one group (multiple data) at as its corruption could cause a denial-of-service (DoS) attack. The a time asymmetric algorithm provides for encryption to be completed by • Asymmetric (also known as public key) algorithms the public key and the decryption completed by the private key. The strength of the privacy provided is based on the combina- Figure 4 presents an asymmetric cryptographic workflow. tion of the cryptographic algorithm and the length of the asso- A downside of asymmetric cryptography is that it requires more ciated key13. Industry-approved cryptographic algorithms and processing power and longer-length keys to achieve a level of key lengths are provided in Barker and Roginsky, “Transitions: security comparable to symmetric cryptography. For this reason, Recommendation for Transitioning the Use of Cryptographic asymmetric cryptography is typically used for the generation and Algorithms and Key Lengths”14. verification of digital signatures. This use will be discussed in the The length of time in which a cryptographic key should be used “Integrity for IoT” section of this paper. is called a cryptoperiod. Cryptoperiods vary based on the algo- rithm, key length, usage environment, and volume of data that Private Key is being protected. Guidance for cryptoperiods can be found in Barker, “Recommedation for Key Management—Part1: General (Revision 4)”15. Wind River Encryption kdKSp8*_9(&\||akd^2 Decryption Wind River Symmetric cryptographic algorithms use the same key for both the (AES- Cipher Text encryption and decryption processing. This would be similar to a Plain Text Plain Text door lock that is keyed on both sides of the door for the same key: IoT Device #2-n IoT Device #1 lock the door from the inside (encryption); unlock the door from the outside (decryption). An example of a symmetric algorithm Public Key is the Advanced Encryption Standard (AES)16. AES is an industry- 17 approved symmetric algorithm for providing confidentiality of Figure 4. Asymmetric cryptographic workflow sensitive data. Figure 3 shows a typical symmetric cryptographic workflow. In the IoT arena, sharing a cryptographic key can be Protecting the confidentiality of data in an IoT device can be a challenging because of the large number of end-points involved. regulatory requirement, a method to protect IP, or an industry- This challenge will be addressed in the “Key Management recommended requirement (for example, in aerospace18, automo- Implementations” section of this paper. tive19, defense20, industrial21, medical22, and networking23. Shared Key Data in an IoT device can be in one of three states: in motion, at 0123456789abcdef0123456789abcdef rest, or in process. Data in motion is data passing throughout the IoT; data in process is data generated or consumed within an IoT Encryption Decryption Wind River BhT*%Fq.1hp)@\gd Wind River device; and data at rest is data stored on the IoT device. (AES-128) (AES-128)(AES- Cipher Text Plain Text Plain Text Data-in-Motion Privacy IoT Device #1 IoT Device #2 The data being passed over the network can be more than just the data being generated or consumed by an IoT device. The man- Figure 3. Symmetric cryptographic workflow agement data to and from the IoT device is just as critical. Updates 4 | White Paper A SURVEY OF INFORMATION SECURITY IMPLEMENTATIONS FOR THE INTERNET OF THINGS and patches, telemetry, and logging information can be of signifi- shorter key lengths than asymmetric