Paper Discussion 7b: Sepecial Session - The Future of IoT Security

[albero94]

@AkinoriKahata Akinori Kahata, comprehensive:

• How does the resource use comparison between traditional TPM and new methods like SMART or TrustLite look like?
• Can edge computing help record the data provenance of IoT devices?

@pcodes Pat Cody, comprehensive:

• Why is blockchain more useful to log data than traditional approaches?
• How do you implement a global SDN?

@samfrey99 Sam Frey, critical:

• What makes the use of proof of stake challenging?
• One section talks about using device communication for attestation whereas another section talks about the challenges of secure communication, how can this problem be solved?

@anguyen0204 Andrew Nguyen, critical

• Why PoS is not a good option based on the paper?
• The remote attestation section says that it was designed for a single prover and then quickly moves into different protocols and functions without really explaining them.

@RyanFisk2 Ryan Fisk, critical

• Can edge computing help with attestation?
• Can blockchain be used for sensor fusion?

Shared concerns/questions:

• In the blockchain section, the authors barely explain why some consensus algorithms cannot be applied and why blockchain can really make a difference in security
• Some other sections, like scalable authentication, do not go deep enough in explaining the possible solutions to the mentioned security problems

[AkinoriKahata]

Reviewer: Akinori Kahata Review type: Comprehensive

1. The problem being solved.
   • The number of Internet-of-Things(IoT) devices is increasing rapidly, and IoT will change society. Especially, IoT devices treat a large amount of data, including personal data and industrial data. Then the trust of IoT devices and the data created at IoT devices is essential; however, due to the limitation of resources and power of IoT, improving the security of IoT is big challenge.
2. The main contributions.
   • The author chooses five technical trends of future IoT security and evaluates possibilities which trend has. The author points out that every trend has a strong point, but each of them is not developed for IoT devices originally, and every trend is facing the challenge of implementing IoT devices. Knowing these trends from a higher perspective is useful for researchers and developers to put forward the next level of IoT security.
3. Questions.
   • Remote attestation is one of the big issues of IoT devices, and SMART and TrustLite look good solutions. It is my interest, but I want to know how much the volume of using computer resources and power is different between traditional TPM and the new methods. The explanation of the paper is little abstract to me.
   • I think data provenance is very important for digitalized society. According to the paper, it is difficult for IoT devices to store the data provenance. Is there possibility that edge computer accumulate the data provenance of IoT devices? I think sharing task between IoT edge computer and IoT devices is good solution for solving resource problem.

[pcodes]

Reviewer: Pat Cody Review Type: Comprehensive

Problem Being Solved

The IoT is constantly growing, with huge numbers of devices being projected to be on the IoT in coming years. Many of these devices, however, are rushed to market and have security flaws. There is a growing need for systems to secure the IoT.

Main Contributions

This paper analyzes a variety of new technologies and how they could be applied to IoT security. The blockchain could be used as a way to securely log IoT application, and to verify sensor readings. Software-defined networks could be used to enhance security and network management. The paper also discusses improving risk assesment and threat modeling for IoT security flaws.

Questions

• Why is a blockchain specifically more useful for logging data, as opposed to more traditional approaches?
• How would you implement a "global SDN"? This seems like it would require a lot of coordination.
• What incentives exist for companies to adopt new threat models? The moral "we don't want to be hacked" reason doesn't seem to stop companies right now from ignoring security.

[samfrey99]

Reviewer: Sam Frey Review Type: Critical

Problem: The number of IoT devices is expected to reach 1 trillion by 2030, but these devices are often rushed to market with security flaws that could compromise the privacy of the user. Security must become a priority in the development of future devices.

Main Contributions: The authors explore many ways to improve IoT security such as a permission-based blockchain implementation, improved identity verification and authentication, and improved network reliability. They also propose a threat assessment framework to better define the the security issues facing IoT devices.

Questions:

1. What makes applying proof of stake to an IoT blockchain implementation so challenging? The paper doesn't specify here.
2. Would advancements in edge computing make a Byzantine agreement algorithm a feasible security addition for IoT?

Critiques:

• The section on sensor trustworthiness felt too short. They mention a few different technologies, but then just say that more research is needed. I wish there were a more concrete point here.
• Unless I misread the paper (which is entirely possible), it seems that the solution proposed to the issue of building a scalable authentication algorithm (section 5) is a multi-device platform, but the previous sections outline just how challenging secure communication between devices can be. I found this confusing. Is this actually a reasonable solution? A callback to the previous sections in the conclusion of section 5 could make this point much clearer.

[anguyen0204]

Reviewer: Andrew Nguyen Review Type: Critical

Problem: IOT is constantly growing and with new facets, applications, and interpretations. As a result, there is a concern as to how secure these new devices are and what are the solutions to patch these security holes efficiently. New technologies like blockchain and software defined networks propose new security solutions.

Main Contributions: The paper begins with a discussion and ideas to provoke thoughts of the trade offs between IOT and the environment it is in. For example, it may greatly benefit the economy and lifestyles of various people, but the security and behind-the-scenes concerns of IOT can't be overlooked. It then discusses possible solutions such as Blockchain, sensor data, and network reliabilities

Questions:

1. PoW is not feasible since it takes a lot of resources for blockchain but PoS (proof of stake) is adaptable. How so?
2. The paper says a problem for remote attestations is that is geared for a single-prover setting. What exactly is this saying?

Critiques: Shortly after talking about the remote attestation, the paper jumps into a lot of protocols and functions that were not previously explained or mentioned making it difficult to follow along and understand what exactly what the authors are trying to say since it goes to 0 to 100 fast. Next, I think overall the paper was written well to cover the various themes and topics of IOT security and its application globally/remotely. I would have liked to see a more fleshed out blockchain portion and more details regarding it. Although it was alright to follow along for the most part, I still am left with a sense of no resounding conclusion for how the security can be really put into this. So, it seems that the paper brushes on a few topics but doesn't spend enough time to really dive into them and fully tell what it wants to tell. As a result, the different themes were more 30%-60% fleshed out.

[RyanFisk2]

Reviewer: Ryan Fisk

Review Type: Critical

Problem

IoT security is an ever growing problem as new devices are used more frequently and store more sensitive information. Research into securing IoT devices and networks has yielded some progress, however there are still many security risks that need to be addressed.

Contribution

This paper examines some of the major security concerns with IoT devices and analyzes some of the proposed solutions for those problems. The authors also discuss the ways in which new technology could be used for IoT security, such as using blockchain for behavior logging or authentication. They also look at multiple levels of an IoT system, including sensor fusion and networking between devices.

Questions

1) Despite the security problems, could edge computing be a solution to the attestation problem?

2) Would using approximate byzantine algorithms for sensor fusion as the paper suggests cause more security problems by being less accurate?

3) Could the blockchain methods the paper suggests be used for sensor fusion?

Critiques

1) The paper goes very in depth with solutions to most of the problems they present with IoT security except in the section about scalable authentication. The analysis in the section is more of a proposal for security policies in IoT systems.

2) I'm not sure why the paper went from talking about sensor fusion to networking multiple IoT devices. There are security problems on the devices themselves and the paper seems to jump over those.

[gkahl]

Reviewer: Greg Kahl

Review Type: Critical

Problem:

As the scope of IoT continues to grow and even more devices, especially embedded devices, are being connected to the internet. Security, especially authentication, attestation, and threat detection/modeling.

Contributions:

This paper explored the current methods for handling these things and analyzed the viability of each of them in a distributed IoT environment. In addition to this they further discussed the use of blockchain and software defined networks and the multiple ways in which these may be leveraged to increase security across the Internet of Things.

Questions:

1 - In section VII-C they proposed the idea of dynamic attack taxonomy which would evolve as IoT evolves. How would this work? It seems difficult to know how the environment is going to change, and seems like the IoT environment changing is how we are in the position we are in now.

2 - In regard to the first two attestation techniques they mentioned, would it be feasible to combine both lightweight security architectures with program code written to ROM to prevent the code itself from being changed, and a monitoring service to analyze the control flow to further increase security?

3 - In the Authentication section they outlined what a secure authentication scheme would consist of, but didn't really conclude any existing schemes which would be effective. Are there any existing authentication methods for IoT devices which fit this bill?

Critiques:

• When they were discussing the viability of blockchain, they very briefly discussed the proof of work and proof of stake approaches, without really explaining how they are not viable. It felt like they were brushed aside too quickly.
• The section on threat modeling felt like it came out of nowhere. They went from discussing security asspects such as authentication and attestation to general threat modeling, and threw a long list of popular risk assessment approaches without discussing any of them. It left me feeling confused.