This Python implementation replicates and extends the concepts from the paper "Pancake: Frequency Smoothing for Encrypted Data Stores". It enhances the original Pancake model by adding a correlation between queries. This feature allows for more sophisticated analysis and understanding of query patterns over encrypted data stores.
- Pancake Model Replication: Implements the core ideas of frequency smoothing in encrypted data stores as described in the original Pancake paper.
- Query Correlation: Extends the model to include correlation between queries, enabling deeper insights into data access patterns.
- Frequency Calculation Enhancements: Offers improved methods for calculating frequencies and correlations within the data store.
- Python 3.x
- NumPy
Simply copy the pancake.py file to your project directory. Ensure that you have Python and NumPy installed.
To use the Pancake class, import it into your Python script:
from pancake import Pancake
Create an instance of the Pancake class:
pancake_instance = Pancake(num_states, s=2, smoothing=0.0, seed=None, add_rep=True, delta=-1)
num_states: Number of states (or keys) in the data store.s: Stage (default is 2).smoothing: Smoothing parameter (default is 0.0).seed: Random seed (optional).add_rep: Boolean to decide whether to add replicas to keys (default is True).delta: Delta parameter for correlation (default is -1).
The instance provides various methods for frequency calculations, adding replicas, and generating query sequences.
cal_corr(): Calculates the correlation between queries.gen_init(): Generates initial probabilities for keys.gen_transmat(): Creates a random transition matrix.cal_freq(): Computes frequency of each key.add_rep(): Adds replicas to keys to modify their frequencies.correlated_query_seq(): Generates a sequence of queries with correlation.