ECE Graduate student Munibun Billah recently won the award for the top graduate student paper at the 2023 International Telemetry Conference for his paper titled “Artificially Noise-Injected Low-Density Parity-Check Codes For the Gaussian Wiretap Channel.”
One of Munibun’s inspirations for pursuing engineering is the cool gadgets featured in science fiction movies and books. In this project, he found an innovative method of securing communication channels against eavesdroppers.
Munibun explains wiretap coding as a means of securing data transmission from an eavesdropper, which can ensure minimal information leaks. His paper calculates the information leakage for Gaussian channels by using a general framework that could be used to calculate different communication channels’ leakage.
He says, “To determine how much security the wiretap codes provide, we calculate information leakage between the confidential message that is to be transmitted and what the eavesdropper has received. Calculation of this information leakage depends on the eavesdropper channel.”
Munibun shares that for some channels the calculation is very straightforward, but for others it is very difficult, and no closed form solution exists.
“To overcome this difficulty, we used a deep neural network called Mutual Information Neural Estimator (MINE) to calculate information leakage for Gaussian channels. Also, we devised a wiretap coding scheme using one of the most popular error correction coding techniques called Low Density Parity Check Codes.”
The simulation results show that if wiretap coding is used, it can provide security against eavesdroppers for gaussian channels.
Munibun’s paper was recognized at this year’s conference in October. Congratulations, Munibun!