#lecture reflection

The root cause analysis

Blame everything on the last person to touch it

Culture

All cyber issues are due to human error ( fix easily and quickly )

Human Weaknesses

Honesty

Misdirectingand limited focus                                                                  People are bad picking the right features for focus ( if the condition is complex and useless)

Similarity matching

Frequency gambling

Availability heuristic - Kahneman 3

Satisficing and bounded rationality (满足和有限理性）

Satisficing is only doing good enough, rather than doing perfectly

Bounded rationality refers to how we have a limited amount of information, meaning our ability to make decisions is bounded by that information

People prefer positive statements

Cognitive strain（认知压力）

Group-think syndrome （群体思维综合症）

How people think when they’re in groups rather than as individuals

People prefer to keep the peace in a group rather than fight against collective ideas

The result is grouped become homogeneous(均匀）

Good for analyzing systems

Confirmation bias

People prefer the evidence that confirms what they believe

How to fight Human error (inevitable)

Fixing the problem isn’t about finding a single point of failure, maybe it’s just fixing everything you can, a tiny bit

Always assume that the person using the system is going to do whatever they can to attack it

Always assume that the environment of a system will be the worst-case scenario

Culture problem

Don’t just punish the person who made an error, or the last person who touched it

learning and stopping these situations in the future

Code problem

Bug bounty program

A deal offered by many websites, organizations and software developers by which individuals can receive recognition and compensation for reporting bugs, especially those pertaining to exploits and vulnerabilities

Crowd-Sourced Bug Bounty Websites

Public: Hackerone, Bugcrowd

Private: Synack

Tips

Look for software updates or assets that have recently changed

Look for publicly disclosed reports ( it is possible to happen again )

Fuzzing

An automated software testing technique that involves providing invalid, unexpected, or random data as inputs to a computer program

Not precise, can test a large number of inputs

Penetration Testing (渗透测试）

Definition

An authorized simulated cyberattack on a computer system

Performed to evaluate the security of the system

Identify strengths and weaknesses

Function

Allows us to discover vulnerabilities in our system before attackers do

AES (Advanced Encryption Standard)(高级加密标准)

Substitution-permutation network

Seems to not have been broken yet in the feasible time

Block Modes

Allow block ciphers to work with large data streams

The initialization vector is used to ensure distinct ciphertexts are produced even when the same plaintext is encrypted multiple times independently with the same key ( 使用相同的密钥对相同的明文进行多次独立加密也产生不同的密文）

ECB (Electronic Codebook Mode)

An algorithm that uses a block cipher to provide information security such as confidentiality or authenticity

Most modes require a unique binary sequence (an initialization vector (IV))

The IV has to be non-repeating or random

The message is divided into blocks

Each block is encrypted separately

Each block is decrypted separately

The disadvantage of this method is a lack of diffusion (离散）

CBC (Cipher-Block Chaining)

Each block of plaintext is XORed with the previous ciphertext block before being encrypted

Each ciphertext block depends on all plaintext blocks processed up to that point

To make each message unique, an initialization vector must be used in the first block

Disk Encryption

a technology which protects information by converting it into unreadable code that cannot be deciphered easily by unauthorized people

Wired Equivalent Privacy (WEP) (有线等效隐私)

a security protocol

designed to provide a wireless local area network (WLAN) with a level of security and privacy comparable to what is usually expected of a wired LAN

Wired Equivalent Privacy (WEP) type

A stream of bits (A) 

sent from one point to another

an initialization vector (初始化向量) — 2^24 keystreams 

secure and the key never use twice

A random stream of bits (B)

generated using an algorithm, such as RC4 (insecure)

The XOR function turns A into a stream of bits with the statistical properties of randomness.

A (xor) B ->  C (Encrypted Stream of bits) 

C (xor) B -> Gives back A again

figure out ???

Hash attack

Broken means violated faster than brute force

MD5, SHA0, SHA1 - broken

SHA2, SHA3 - not broken

SHA3 only one not vulnerable to length extension attacks

Merkle Damgard Construction

a method of building collision（碰撞）

resistant cryptographic hash functions from collision (抗冲击）

resistant one-way compression(单向压缩) functions 

used in the design of MD5, SHA1, SHA2

The way that Merkle Damgard Construction work 

Break the message into a series of blocks

Start with IV (initialization vector)

Given hash function f, it takes in IV and the block 1, output goes into f again with the block 2, etc.

After doing this with all blocks, the final hash is produced

Break the Digital Signatures RSA

RSA relies on people not being able to factorize the large number that is the mod

Brute force 

It is not the only way to break the cipher

The easiest way to calculate and need to find the domain of work

Guided Bruteforce

Brute force with heuristics ( 启发）

Entropy ( you can see the lecture 2 )

Degree of randomness ( 程度 ）

Complements patterns or order （ 补充性 ）

Less patterns = higher entropy

Possible combinations ( 5 letters word 5个字母的单词 )

2^25 possible combinations

Only 2^13 valid words

Each letter adds 2.5 bits on average

Different attack types

Replay attack 

an attack on a security protocol using replay of messages from a different context into the intended (or original and expected) context 

fooling the honest participants into thinking they have successfully completed the protocol run

Man-in-the-middle attack

an attack where the attacker secretly relays and possibly alters the communications between two parties who believe they are directly communicating with each other ( active eavesdropping / 主动窃听）

Hashing

One-way converts any form of data into a unique string of text (fixed size)

Hash functions return the same results each time (deterministic / 确定性)

Hash collisions ( Birthday attack ) — different items map to the same thing

Clustering in data causes clustering in the hash ( 数据中的群集导致哈希中的群集 )

The original message and hash can be tampered with ( 篡改 ）

How: computer the hash and check whether is the same   

Cryptographic Hashes 

quick to compute

return the same results each time (deterministic)

hard to reverse

hard to find two different texts that have the same hash

If the input is changed slightly, the output changes significantly (Avalanche property)

Avalanche property

Hard to find two messages with the same hash

If the input is changed slightly, the output changes significantly

Social Engineering

Always lie - consider a security question as another password

Use password managers and scrub your social media ( 清除 ）

Don’t reuse security questions and answers

Principles of Persuasion

key 

people can be easily exploited using persuasion

Reciprocity (互惠）

Increase the time delay between the initial gift and the later request

Social Cues

People believe in the social cues around them

Liking 

Authority

MAC

Append a password in the message to be hashed

Message Authentication Code

Public Key Cryptography ( RSA )

Revolutionary system & the first instance of public-key cryptography 

2 keys ( public and private )

The public key — Open 

The Private key — Only the person with the private key can decrypt the messages

Based on A^B mod C = ( (A mod C)^B ) mod C) 

How to calculate

Message M has a numerical equivalent m 

Choose p,q to be different primes (LARGE)

Calculate n = p*q

Calculate 𝛷(n) = (p-1)(q-1)

Choose an integer ‘e‘ such that 0<<e<<p,q and gcd(e, 𝛷(n)) = 1 (relatively prime)

Choose an integer ‘e‘ such that 0<<e<<p,q and gcd(e, 𝛷(n)) = 1 (relatively prime)

Choose an integer d such that ed≡1 mod 𝛷(n)

Encoding m:   m → m^e mod (n) 

Decoding c:    c → c^d mod (n)

Risk

Invisible

Always bad ( whether it happens or not )

See the outcome, not intelligence

Human are bad at assessing risk ( it is hard to get an actual probability of it happening )

Risk Type

Low-probability high-impact ( e.g. asteroids. Humans obsess or ignore it )

Low-impact high-probability

Risk matrix ( 矩阵 ）— 2D graph

Show the impact and probability

Allocation of resources

High impact and high likelihood (focus)

Low impact and low likelihood (ignore)

Risk registers 

Tool for documenting risks and actions to manage each risk

Compliance ( 合规 ）

With processes （ check at the beginning )

Measure info in bits

1 bit of info distinguishes between two cases

2 bits -> 4 cases

Bits of security - a cipher takes n operations to solve. This can be written in terms of 2^x where x is the number of bits of security required to solve a problem

Centralization ( 集中化 ）

Systems with a single point of failure often have very high-level security

Space-time tradeoff （ 时空权衡 ）

To defender — an algorithm can increase space usage to decrease time spent or vice versa

To attacker —use the space-time tradeoff, the attacker can roughly halve the effective number of bits

Work factor

The amount of effort required to break down a cryptosystem

Dealing with Risk

Prevention

Try to prevent an event from happening by removing the vulnerabilities which allow the event to occur ( like the rules )

Limitation

If the event cannot be prevented, then try and minimize it by limiting how bad the situation can get ( the installation of fire separation walls beforehand & lower the probability )

Passing the risk to a 3rd Party

Such as buying the insurance before

Wearing the risk