QFlow: A Fixed-Point FPGA Accelerator for Adaptive Key-Resource-Aware Route-Candidate Evaluation in Trusted-Relay QKD Networks
Supervised by Prof. Dr. Md. Nazrul Islam Mondal, Department of CSE, RUET. The work proposes a fixed-point FPGA accelerator for adaptive, key-resource-aware route-candidate evaluation, combining hardware-efficient route scoring with adaptive profile selection. The RTL architecture was prototyped on a Digilent Nexys 3 (Spartan-6 XC6LX16), with an offline reinforcement-learning policy mapped into deterministic hardware policy/profile ROMs.



