Abstract:The aim of this project was to develop a new Reservoir Computer implementation, based on a chaotic Chua circuit. In addition to suitable classification and regression benchmarks, the Reservoir Computer was applied to Post-Quantum Cryptography, with its suitability for this application investigated and assessed. The cryptographic algorithm utilised was the Learning with Errors problem, for both encryption and decryption. To achieve this, the Chua circuit was characterised, in simulation, and by physical circuit testing. The Reservoir Computer was designed and implemented using the results of the characterisation. As part of this development, noise was considered and mitigated. The benchmarks demonstrate that the Reservoir Computer can achieve current literature benchmarks with low error. However, the results with Learning with Errors suggest that a Chua-based Reservoir Computer is not sufficiently complex to tackle the high non-linearity in Post-Quantum Cryptography. Future work would involve researching the use of different combinations of multiple Chua Reservoir Computers in larger neural network architectures. Such architectures may produce the required high-dimensional behaviour to achieve the Learning with Errors problem. This project is believed to be only the second instance of a Chua-based Reservoir Computer in academia, and it is the first to be applied to challenging real-world tasks such as Post-Quantum Cryptography. It is also original by its investigation of hitherto unexplored parameters, and their impact on performance. It demonstrates a proof-of-concept for a mass-producible, inexpensive, low-power consumption hardware neural network. It also enables the next stages in research to occur, paving the road for using Chua-based Reservoir Computers across various applications.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to develop a new Reservoir Computing kernel based on the chaotic Chua circuit and study its application in Post - Quantum Cryptography (PQC). Specifically, the goals of the paper include: 1. **Develop a new reservoir computing model**: - Design and implement a new reservoir computer based on the chaotic Chua circuit. - Characterize the Chua circuit through numerical simulation and physical circuit testing. 2. **Evaluate its performance in classification and regression tasks**: - Use existing benchmarks to verify the performance of the new model in standard classification and regression tasks, ensuring that it reaches the level in the current literature. 3. **Explore its application in post - quantum cryptography**: - Apply the reservoir computer to the Learning with Errors (LWE) problem in post - quantum cryptography for encryption and decryption. - Research whether the Chua - based reservoir computer can handle the highly nonlinear problems required in post - quantum cryptography. 4. **Noise analysis and optimization**: - Analyze and mitigate the impact of noise on system performance to improve the stability and reliability of the system. 5. **Prospects for future work**: - Explore the use of multiple Chua reservoir computers combined into a larger - scale neural network architecture to handle more complex tasks, such as the learning with errors problem of high - dimensional behavior. The paper points out that although the reservoir computer based on the Chua circuit can achieve good results in traditional tasks, its complexity may be insufficient when dealing with highly nonlinear tasks such as post - quantum cryptography. Therefore, the future improvement direction is to study different combinations of Chua reservoir computers to achieve higher - dimensional behavior, thereby better solving the learning with errors problem. In addition, this research also shows the potential of the reservoir computer based on the Chua circuit in low - cost, low - power - consumption hardware neural networks and paves the way for subsequent research.

New Reservoir Computing Kernel Based on Chaotic Chua Circuit and Investigating Application to Post-Quantum Cryptography

Practical and Scalable Quantum Reservoir Computing

Classical and quantum reservoir computing: development and applications in machine learning

Prediction of chaotic dynamics and extreme events: A recurrence-free quantum reservoir computing approach

Configured Quantum Reservoir Computing for Multi-Task Machine Learning

Optimal training of finitely-sampled quantum reservoir computers for forecasting of chaotic dynamics

Memory-Augmented Hybrid Quantum Reservoir Computing

Quantum reservoir complexity by Krylov evolution approach

Reservoir Computing Using Measurement-Controlled Quantum Dynamics

Post-Quantum Cryptography Neural Network

A novel Chua's based 2–D chaotic system and its performance analysis in cryptography

Optimizing quantum noise-induced reservoir computing for nonlinear and chaotic time series prediction

Reconstructing bifurcation diagrams of chaotic circuits with reservoir computing

Hybrid quantum-classical reservoir computing for simulating chaotic systems

Quantum reservoir computing with repeated measurements on superconducting devices

Stochastic Reservoir Computers

Predicting Chaotic Systems with Quantum Echo-state Networks

Quantumness and Learning Performance in Reservoir Computing with a Single Oscillator

Quantum-activated neural reservoirs on-chip open up large hardware security models for resilient authentication