enjun / symmetric-reconciliation

Symmetric information reconciliation for the QKD post-processing procedure

Symmetric information reconciliation for the QKD post-processing procedure

1. General description

Repository contains simulation of symmetric blind reconciliation algorithm realized in Python 2.7.

2. Files contents

test_error_correction.py

Launches a simulation of the symmetric blind reconciliation for the current number of tries and QBER values. Keeps the result in output.txt.

error_correction_lib.py

Contains basic procedures for performing a test of reconciliation protocol. It includes:

• generation of random keys (bit strings);
• adding errors in accordance with given level of quantum bit error rate (QBER);
• choosing an appropriate rate of LDPC code among a given range and numbers of shortened and punctured bits;
• generation of positions for shortened and punctured bit;
• extending key with shortened and punctured bits;
• syndrome encoding;
• syndrome decoding;
• performing symmetric blind reconciliation for given pair keys;
• testing of the full procedure of information reconciliation, including generation of keys, adding errors, and collection of statistics.

codes_1944.txt

Pool of four standard LDPC codes of block length 1944 together with positions for untainted puncturing. The set of code rates is {5/6, 3/4, 2/3, 1/2}.

codes_4000.txt

Pool of nine LDPC codes with block length 4000, constructed with improved progressive edge growing algorithm with particular distribution polynomials. The set of code rates is {0.9, 0.85, ..., 0.5}.

file_utils.py

Contains some auxiliary procedures for reading files with codes.

3. Notes about storage of parity-check matrices

The storage of parity-check matrices is based on two variables: s_y_joins and y_s_joins. They contains positions of nonzero elements for each row and column correspondingly. For example, for the matrix

H = 
1 1 0 1
1 0 1 1
0 1 1 0

one has

s_y_joins = [[0,1,3], [0,2,3],[1,2]]
y_s_joins = [[0,1],[0,2], [1,2],[0,1]]