Welcome back. This week we will first talk about physical attacks. From studying various ways of physical attacks, you will see the threats from hardware, or, how to break security from hardware side. You will also learn some of the countermeasures to these attacks. Probably one thing that you may find a very use for is how industry and government define and evaluate the security levels of a product. Most of the physical attacks require special skills and knowledge in addition to sometimes very expensive equipment and the tools. However, you do not need to have any special background to understand most of these attacks. Our goal in this lecture is to let people be aware of these attacks, not to teach them step by step, how to do such attacks. Next week, we will discuss in more details about side-channel attack, which is a very powerful physical attack. Over the years of lecturing on hardware security, I find it is very difficult to deliver this part of the material without a lab for demonstrations. For this Coursera course, I will also take away some of the images. Part of this course is developed based on a book chapter by Sergei Skorabogatov. On his web page, Sergei has also posted his lecture notes Physical Attacks and Tamper Resistance. If you are interested in this topic, you should definitely read this book chapter. Let's start with the question, what are the physical attacks? I cannot give you an accurate definition, but all the physical attacks have the following characteristics, or require the following conditions. First, to launch a physical attack, the attacker needs physical access to the chip, or at least should be able to contact, to contact, to connect certain wires in order to measure the target's signals. With more and more devices being connected wirelessly nowadays, physical attacks without physical access to the system have also been reported. For example, acoustic attacks, near field communication attacks, and some attacks to the RFID can all be performed close to the device without touching them. In addition to the necessary equipment and the tools, the attacker must also have the required skills and the knowledge to perform physical attacks. The second common thing about all the physical attacks is that they all have two phases. First, the Interaction phase where the attackers interact with some physical characteristics of the device to collect data. And then, the Exploitation phase, where the attacker analyze the collected information to reveal the secret and break the security. From this, without discussing any physical attack in details, what can we say about physical attack and hardware security? First, we can imagine that this requires, requirements make it harder to attack the system from hardware than from software or network. We have said that Physical attacks needs Physical access to the system. And those will most of the time need Specialized equipment, tools, and the knowledge to perform the task the, the attack. And normally, for software attack or network attack the, the bar is much lower. On the other hand, we know that the attackers loss will always be our gain. In this content, contexts this means that with the same costs and efforts investing on protection from hardware will most likely make the system more secure than adding security from network and software layers. This is also one of the reasons why hardware security has gained a lot of attention in the recent years. However, the downside of building security at hard lev, hardware is that this will also add another attacking surface. We have to make sure that the hardware is secure and trusted. Which means that we should consider the vulner, the vulnerabilities of hardware design and implementations. And make sure that the back doors will not become new threats. Attackers, not necessarily who launch physical attacks, can be categorized into three groups. Class I, clever outsiders. These attackers are often very intelligent, but they may not have sufficient knowledge about a system. They may not have access to the equipment and the tools necessary to break the system either. Class II, knowledgeable insiders. These attackers are familiar with the system, and have highly sophisticated tools and equipment to analyze and break the system. Class III, funded organizations. They normally have a team of expert, experts with all the necessary skills, tools, knowledge, and equipment to invent new attacks if necessary to break the target system. So what are the motivations for all these kind, kind of different attackers to break a system? There's one single motivation, money. Actually, this is also the motivation for almost all the attacks. Based on how money can be made from successful physical attacks, we can further partition the motivation for physical attacks into several groups. First, steal money or a service directly. Example of these type of physical attacks include, breaking smart cards and making bogus deposit, or breaking the TV set top box to watch programs without subscribing. Or, breaking game consoles to play games without paying. Second, make profit fits by sailing, selling or reselling products illegally. IP piracy, integrated circuits, overbuilding, and counterfeiting, are all examples of this kind. Third, post their own sales by interrupting or damaging the service provided by the competitors. For example, an attacker can insert malicious updates, patches or hardware tojan to make competitors device or system malfunction or suffer bad performance. This will give their own product an end, an unfair competitive edge. We have discussed IP protection last week, and we will cover hardware trojan later. This week our focus will be on how physical attacks can break the system. Particularly cryptosystems to steal service or money directly. When we say breaking a system, we do not mean physically destroying the system. Instead, the attackers want to learn information about the system that he doesn't have ac, the authority to access. In the case of cryptosystems, such information could be the secret key, or the secret data. In the case of design IP protections, it would be the design details. If on further discuss, discussing physical attacks, we would like to point out the difference between Physical Attacks and Cryptanalysis. They both have the goal of breaking the security. However, classic Cryptanalysis focuses on breaking the cryptosystem by mathematical analysis of the crypto, cryptographic algorithms. It had success in the past, for example, the birthday attack and the differential Cryptanalysis. But now, Cryptanalysis is becoming harder and harder. This is because most of the modern crypto systems are theoretically sound and unbreakable. Instead, there may exist flaws in the implementation of the crypto systems. Physical attacks attempts to find such flaws and break the system. Based on whether the targeted system or device will be damaged during or after the attack, Physical Attacks can be classified in three groups. Invasive attacks, this kind of attack requires direct access to the inside of the chip or device. Depends on whether after you attack, the system can be reassembled. The sys, the, the attack can be either Reversible, otherwise it called irreversible. Apparently, for irreversible attacks, it cannot be repeated. Normally, the device will be damaged after the attack. Sometimes tamper evidence will be, will left. The cost and required skills to perform invasive attack varies based on how the attacks will be performed on what kind of systems. But normally, the cost and effort will be very, very high. The second category is called a Non-invasive attack. In this case, the attacker will interact with the device, with a chip, with its interface, such as voltage, current, clock, input-output interface, etc. And such attack can also be Passive or active. Depending on whether the attacker only monitors and measures this physical measure, physical characteristics. Or, the attacker also injects input to the system to cause the system malfunction. Normally, such attack will not damage the device, and it will not leave any tamper evidence. Most of these thing, these attack, have very low cost and can be repeated. And in the middle, between invasive attack and non-invasive attack, there's another category called semi-invasive attacks. These attacks, normally they require access to the surface of the chip. But they do not need to create contacts with the internal wires. And usually did not damage the system, and depends on how the attacker is performed, they may or may not leave tamper evidence. And the cost of this depends on what kind of surface, what kind of interact they will have with the system. They can be high or low. But still be definitely much lower than the very expensive invasive attacks. And they also require some special skills. And most of the cyber invasive attacks they are repeatable. There are a few ways to perform physical attacks. And based on how physical attacks are performed, it can also be put into several different categories. Reverse engineering, this type of attack will study the chip's in, inside structure. Then trying to decide, determine the functionality of the system. Normally, this has very high cost, and the attacker should have similar capabilities of the designer. And normally have, also have the capability of manufacturing. And this is in general is in, invasive attack. The second group is called Microprobing. In this case it requires to direct access to the chip surface to observe, manipulate, and interact with the chip, and they are also invasive attacks. The third category is called a Fault generation. In this case, the attacker will generate, create f, Fault, normally a faulty input, or make the chip run eh, of normal environment conditions, and in the hope that the chip will malfunction, to leak information, or give the attacker additional access to the system. Depending how the fault is generated, the attack can be either be semi-invasive or non-invasive. And Side-channel attack, which we'll discuss next week, they are normally non-invasive attacks, and this type of attack has two phases. First, the attacker, while monitoring, will measure a chip's physical characteristics, such as power, current, timing, EM radiation, etc, during the chip's normal operation mode. And then, with the data collected, they will perform a data analysis to learn the hidden information. And finally, there's a category called Software attack. This again, it is non-invasive and the attacker will use normal input, output interface trying to explore known security vulnerabilities in the se, in the security protocols or algorithms were there software implementations to perform an attack. This attack, they are repeatable and non-invasive.