|
|
 |
| |
|
|
|
Deep Learning–Based Graph-Theoretic Modelling of Vulnerabilities in Post-Quantum Cryptosystems |
|
|
|
PP: 963-970 |
|
|
doi:10.18576/jsap/140632
|
|
|
|
Author(s) |
|
|
|
Sulieman. I. Mohammad,
Hamza Farhan Abu Owida,
A. Vasudevan,
Hanan Jadallah,
Mohammad Faleh,
Yogeesh N.,
|
|
|
|
Abstract |
|
|
| We introduced the provision of secured and trustworthy data transmission in network-coded communication systems is an important file in the dynamic network environment and adversarial attacks. The paper introduces a secure network coding scheme that is built on top of combinatorial designs that have been combined with key scheduling, which is fuelled by reinforcement learning. Balanced combinatorial designs ensure structure in coefficient generation to maximise algebraic security through the uniform distribution of source symbols across coded packets to minimise information leakage upon partial interception. Key scheduling is then designed as a reinforcement learning problem to enhance security further; the ability to select the key depending on the network state measurements and the historical key usage. The performance measures applied to the proposed framework comprise decoding accuracy, leakage rate of information and secrecy capacity. The successful outcome of the experiment confirms the effectiveness of the suggested strategy of secure transmission as it reports accuracy improvement, high decrease of the leakage of information, and increase of the secrecy capacity in comparison to the traditional random network coding and fixed key-based solutions. |
|
|
|
|
|
|
|
