Best practices to protect their connected devices Papers in systems nice about dynamic conservatism


Embedded art And Memfault Connect together to organize a quarterly embedded field of discussion that focuses on the technical aspects of the structure of embedded systems on a scale. We will present guest panel members who are up to date with the latest embedded development. Our goal is to spread advantageous techniques and practices throughout the industry.

Our fifth event focuses on Best practices to protect their connected devices. This discussion will take place on September 7 at 0900 PST. Phillip and Tyler are accompanied by Benjamin ‘Bambam’ Winston. Benjamin has worked on everything, from safe elements to middleware, from door access systems to large online game platforms on a long list of companies that include Tesla, Facebook and Fitbit.

Robust security for Ioot Devices are crucial to protect sensitive data, protect the privacy of users and to ensure the integrity of embedded systems. By implementing strict security measures, you can prevent data injuries, maintain the reliability of the devices and further promote trust among your customers.

Our expert body for embedded engineering has successfully built and distributed IoT products in several industries. You understand how important it is to ensure that customer and company data are protected with suitable security measures. You will discuss:

  • How to include the security in your product development process
  • Which areas should you ensure – transport, actual data, keys
  • How to block your hardware in relation to the sensitivity of the data

You can register for the live event and find the recording On this page.

Table of contents

  1. Written answers
  2. Read more
  3. Previous panel discussions
  4. Written answers
  5. Read more
    1. Blog posts of embedded art posts
    2. Interrupt posts
    3. Other websites
    4. Only membership content
    5. Previous panel discussions

Written answers

Which prefabricated secure boat solutions are available?

Mcuboot and UEFI are two common options.

What kind of safety layers did Philip have?

Instead of referring to specifically defined layers, I said that they should think about defenses as a shift approach. You cannot rely on a single defense strategy and call it “good enough”. The more obstacles you can set up, the more difficult or time -consuming it can be to take advantage of your system.

Rob Johnson expanded my answer:

I heard what I think that they talk about Swiss cheese model. Not every layer helps with a certain susceptibility to security, but overall you should act as a sieve to filter out weaknesses.

This is also the case – just because they were not burned if they have not used a certain reduction, this does not mean that it has no benefit. Perhaps you may have had enough lead that something else was dealt with with this problem.

Is Zero Trust the highest level of safety for hardware devices?

I don’t know that I have the confidence in relating to the “highest level”, but I think it’s a reasonable security seal. Especially in view of the fact that some of the Top error Relate to the wrong/inappropriate/missing authentication, authorization management, default authorizations, etc.

Where (if somewhere) did you see that a safe element is used effectively? I have often seen it as engineering requirementAnd then it is ultimately used as a Secret shop, which (almost how) can be implemented effectively via Flash -reading protection.

In view of the fact that one of the purposes of safe elements (SES) is to use them as a safe secret, this does not seem to be a problem for me. In this light it is often important to create hardware root for A Safety boot Procedure. Some SES also offer the option of performing their own software. SEs that carry out the processing are common in contactless payments (e.g. Apple Pay, Square/Block Hardware) and access controls. If you select parts that can do this processing but do not use it, it is certainly a waste, and you could have selected a cheaper component that only focuses on safe storage.

It may be that for the risk model of your system and probably attacker vectors, reading protection are good enough. However, I would not use too much trust in the Readback protection: It is not foolproof and can be defeated. There are several reasons why you want to use an SE or another dedicated component for this purpose:

  • Ses are often supplied with additional flash memory, which you can use as Secure data storage For PII or other secrets. In this way you can avoid implementing your own Secure the shopWhich is much prone to errors.
  • Saving keys (and other secrets) in an SE makes it more difficult to access when the system is compromised compared to the storage in Flash (there is another level to get through)
  • SES is so constructed that they are resistant to physical and software manipulation, probably Moreso as her flash memory.
  • You will find SES that are prepared with keys and certificates and simplify the manufacturing process if you are interested in clear device key.

When do you really need crypto accelerator in the HW in contrast to FW? If you look at Zephyr, there is little to no support for these devices, which implies that they are either too complex, or there is no sufficient demand

With hardware acceleration compared to software implementation, you get rather meaningful services and electricity efficiency. And of course memory/flash savings if you don’t have to do this in SW. I have worked on several systems in which it is not sufficient for software for an encryption approach.

I do not think so zephyr Is really a benchmark for whether this is a useful ability or not, still a measuring device with sufficient demand. Zephyr’s crypto support is in a bit of river – they still use Tinycrypt, which is no longer a well -kept project. And they can have both TinyCrypt and MbedtLs in their application, since different modules use different implementations. There was a lot of conversation/interest in it PSA -APISWhich would better enable support for the hardware accelerator for some device classes (mentioned several times in this problem). But that hasn’t happened yet.

If you use MBbetls as a crypto driver in Zephyr, you have the option of Redefinition of the Cyrpto primitive to use the hardware accelerator for your device.

Noah Pendleton also gave a written answer:

Phillip is completely right! It depends on your application. For example, if you expect you to carry out a lot of TLS data transmission (more than a few KB per day), you will probably want to accelerate the HW, especially the Handshake -overhead (ECDH etc.) to facilitate handshake (ECDH etc. .). what can be very slow in the software- sometimes for all seconds. Of course, this is particularly important for battery -operated devices, which is why they see many MCUs with low power with low power consumption nowadays with low power with low power consumption. And it can also be important if your system has real -time restrictions (maintenance sensors or displays), the setting of the CPU can affect the other requirements with crypt cooperations.

Read more

It is difficult to treat everything in a temporary discussion! Below you will find additional resources to continue researching the topics discussed by us.

  1. Blog posts of embedded art posts
  2. Interrupt posts
  3. Other websites
  4. Tools
  5. Only membership content

Blog posts of embedded art posts

Interrupt posts

Here are useful articles from The interrupt blog:

Other websites

Tools

Only membership content

Our members Can find more information about security in the Feldatlas.

The recommended high -ranking entries include:

Safety -related practices, functions and tools:

Security risks and weaknesses that were mentioned:

Safety -related regulation:

Previous panel discussions

If you liked this panel discussion, read our other discusions:



Source link