Brute Force OOK using Flipper Zero

Brute force subghz fixed code protocols using flipper zero, initially inspired by CAMEbruteforcer

This repo aims to collect as many brute force files/protocols as possible, so if you can or want to contribute you are more than welcome to do so!

How it works

Using flipperzero-bruteforce.py you can generate bruteforce .sub files for subghz protocols that use fixed OOK codes. Inside the script it is also possible to specify your own protocol in case it's not present.

To generate all the files simply run:

python3 flipperzero-bruteforce.py

It will generate bruteforce files for all the specified protocols organized in many folders with the following structure:

sub_files/
└── PROTOCOL_NAME
    ├── SPLIT_FACTOR
    │   ├── <parent_file>_000.sub
    │   ├── <parent_file>_001.sub
    │   ├── ...
    │   └── <parent_file>_NNN.sub
    └── debruijn.sub

For each protocol there are 6 sub folders, containing 1, 2, 4, 8, 16 and 32 files, SPLIT_FACTOR (the directory's name) indicates the number of keys per .sub file. <parent_file> simply indicates the parent file of the current .sub file, for example, inside folder 64 we have 003_006.sub, its parent file is 128/<parent_file>_003 and its children will be 32/006_<file_id>. This is useful when trying to get a close guess to the key.

For example, let's say you are trying to bruteforce a gate using CAME 12 bit protocol:

Play the single file (4096/ folder) to make sure the attack works
Play the two files inside 2048/ folder, to see which half contains the correct key (suppose the second one works, 000_001.sub containing keys from 2048 4095)
Play its children (001_002.sub, 001_003.sub) inside the 1024/ folder to narrow the search
Keep doing this until you reach the last files inside the 128/ folder, these files take less than 10 seconds to send, almost the same as having the actual remote.

If you wanted to narrow the search even more you could modify the script to generate your own files containing less keys.

The debruijn.sub files are there just for reference I have not really been able to use them succesfully, they try to simulate a debruijn sequence for the protocol, although this is quite unfeasible because most protocols have some sort of stop bit or pilot bit appended or prepended to the actual key that differs from a normal bit. Also I have not yet found a gate using a bitshift register.

Currently supported protocols

Protocol ℹ️	Frequency 📡	Bits	Timing ⏱	Testing 🧪
CAME	433.92	12	287 sec	Tested
CAME	868.35	12	287 sec	Untested
CAME-fast	433.92	12	224 sec	Tested
CAME-fast	868.35	12	224 sec	Untested
Chamberlain	300	9	123 sec	Untested
Chamberlain	315	9	123 sec	Untested
Linear	300	10	212 sec	Tested
Linear	310	10	212 sec	Untested
NICE	433.92	12	628 sec	Untested
NICE	868.35	12	628 sec	Untested
PT-2240	433.92	24	800 hours	Untested
Ansonic	434.075	12	276 sec	Untested
Holtek	315	12	387 sec	Untested
Holtek	433.92	12	387 sec	Untested
Holtek	868.35	12	387 sec	Untested
Holtek	915	12	387 sec	Untested

Currently supported Pagers

Pagers for the Italian restaurant "Spacca", listed as Spacca_pager

Adding a protocol

Adding a protocol is very straight forward, inside the script protocols are defined at the bottom, inside the protocol list:

protocols = [
    Protocol("CAME", 12, {"0": "-320 640 ", "1": "-640 320 "}, "-11520 320 "),
    Protocol("NICE", 12, {"0": "-700 1400 ", "1": "-1400 700 "}, "-25200 700 "),
    Protocol("8bit", 8,  {"0": "200 -400 ", "1": "400 -200 "}),  # generic 8 bit protocol
    ...
]

A protocol is defined by a few parameters passed to the constructor in the following order:

name: the name of the protocol
n_bits: the number of bits for a single key
transposition_table: how 0s and 1s are translated into flipper subghz .sub language
pilot_period: aka preamble, a recurring pattern at the beginning of each key, inexistant by default
stop_bit: a recurring pattern at the end of each key, inexistant by default
frequency: working frequency, defaults to 433.92
repetition: number of transmissions per key in the bruteforce

Optimization

After a little testing working with CAME 12 bit (together with @BitcoinRaven), it seems that as long as the ratio between the bits' lengths is respected (x: short pulse, 2x: long pulse, 36x: pilot period) the actual duration of x can be lowered from the original 320 microseconds. From testing it seems that 250 microseconds is stable, shortening the bruteforce by a good minute (224 seconds against 287 seconds). From a code point of view:

protocols = [
    # Old CAME
    Protocol("CAME", 12, {"0": "-320 640 ", "1": "-640 320 "}, "-11520 320 "),
    # New CAME
    Protocol("CAME", 12, {"0": "-250 500 ", "1": "-500 250 "}, "-9000 250 "),
]

Other attempts we made to shorten the bruteforce were:

Combining two 12 bit codes → ❌
Removing/adding 1Bit from/to the end/beginning → ❌
Sending only the first pilot signal → ❌
Removing the pilot signal between the repetion → ❌
Remove half a bit from the end → ❌
Reducing pulse length keeping the same ratio as the original → ✅

Further testing will be performed for the other protocols.

Timing

To compute the time it takes to perform a bruteforce attack, we need to sum the time it takes to send each code:

(pilot_period + n_bits * bit_period) * repetition * 2^n_bits

For example, computing this for CAME turns out to be:

[(9000 + 250) + 12 * (250 + 500)] * 3 * 2^12 = 224.256.000 microseconds ~ 224 seconds