billautomata / rsa_computer

msp430 based rsa computer
MIT License
0 stars 0 forks source link

hilarious numbers #2

Open billautomata opened 8 years ago

billautomata commented 8 years ago

On an msp430g2553:

An 8 bit key can be cracked in ~100ms. A 16 bit key can be cracked in 1.5 seconds. A 32 bit key can be cracked in 900 minutes.

To decrypt a value using an 8 bit key, takes 1.5 minutes. To decrypt a value using a 16 bit key, takes 2-100 days. To decrypt a value using a 32 bit key takes ????????

Fun keys to crack, can't be used in the encrypt / decrypt operations. To decrypt you need to perform N operations where N = the private key value. The private key for two 8 bit primes is >300,000.

nodechomsky commented 8 years ago

Those are bizarre figures! Nice!

William Van Justice III, M.S., DIDD Approved Behavior Analyst

On Feb 18, 2016, at 7:59 PM, Bill Automata notifications@github.com wrote:

On an msp430g2553:

An 8 bit key can be cracked in ~100ms. A 16 bit key can be cracked in 1.5 seconds. A 32 bit key can be cracked in 900 minutes.

To decrypt a value using an 8 bit key, takes 1.5 minutes. To decrypt a value using a 16 bit key, takes 179 days. To decrypt a value using a 32 bit key takes ????????

Fun keys to crack, can't be used in the encrypt / decrypt operations. To decrypt you need to perform N operations where N = the private key value. The private key for two 8 bit primes is >300,000.

— Reply to this email directly or view it on GitHub.

billautomata commented 8 years ago

Number theory to the rescue. I found an efficient method of solving this in a way that should take 100ms at most. Taking calculations down from 1 billion operations to a couple hundred, at most.

https://github.com/billautomata/rsa_computer/blob/gh-pages/efficient_modular_exponentiation.md

We can use 32 bit keys without it taking years to decrypt with them.

nodechomsky commented 8 years ago

That's great!

William Van Justice III, M.S., DIDD Approved Behavior Analyst

On Feb 18, 2016, at 11:22 PM, Bill Automata notifications@github.com wrote:

Number theory to the rescue. I found an efficient method of solving this in a way that should take 100ms at most. Taking calculations down from 1 billion operations to a couple hundred, at most.

https://github.com/billautomata/rsa_computer/blob/gh-pages/efficient_modular_exponentiation.md

We can use 32 bit keys without it taking years to decrypt with them.

— Reply to this email directly or view it on GitHub.

billautomata commented 8 years ago

15 bit keys are the limits for the encryption. That's up from 12 bit earlier today. I don't think it can be pushed past 15 bits.

15 bit keys are crackable in a few seconds. What if we had multiple msp430 sources creating and transmitting messages to each other, simulating the internet. Screw the user interaction for now. 15 bit keys crack in a few seconds, but what if they have 200 to crack?

Let's see how many msp430's can generate keys before the msp432 (arm board) chokes.

Big i2c bus of keys being transmitted. We can simulate the telco switcher, have a central hub that (needlessly) routes the packets, but also routes them to the NSA (to make our lives easier). Sniffing sounded fun, but blegh.

nodechomsky commented 8 years ago

I like that idea a lot! I can make quite a few of the design I sent you with parts I already have laying around. So we can make specialized nodes that can represent various devices in such a system. The interactive stuff would be interesting, but I think simulating a busy network would be more visually rich.

William Van Justice III, M.S., DIDD Approved Behavior Analyst

On Feb 19, 2016, at 9:23 PM, Bill Automata notifications@github.com wrote:

15 bit keys are the limits for the encryption. That's up from 12 bit earlier today. I don't think it can be pushed past 15 bits.

15 bit keys are crackable in a few seconds. What if we had multiple msp430 sources creating and transmitting messages to each other, simulating the internet. Screw the user interaction for now. 15 bit keys crack in a few seconds, but what if they have 200 to crack?

Let's see how many msp430's can generate keys before the msp432 (arm board) chokes.

Big i2c bus of keys being transmitted. We can simulate the telco switcher, have a central hub that (needlessly) routes the packets, but also routes them to the NSA (to make our lives easier). Sniffing sounded fun, but blegh.

— Reply to this email directly or view it on GitHub.