Energy Consumption Framework and Analysis of Post-Quantum Key-Generation on Embedded Devices

TL;DR

The study addresses energy efficiency of cryptographic key generation in a post-quantum world by experimentally benchmarking classical (RSA, ECC) and post-quantum (ML-KEM) key-exchange methods on a Raspberry Pi 5 using OpenSSL 3.5. A Dockerless, end-to-end framework pairs the Pi (DUT) with a Windows-based TC66C energy meter to quantify energy per key generation and subtracts baseline NULL runs to isolate computation energy. Results show PQC methods, particularly ML-KEM, achieve energy profiles comparable to or better than ECC at higher security levels and far superior to RSA as key sizes grow, with extensive extrapolations suggesting substantial scale energy savings if RSA is replaced. The work underscores the practical viability and potential cost savings of adopting PQC in embedded environments, advocates for accelerated PQC integration in standard libraries, and outlines a roadmap for broader hardware and end-to-end evaluations.

Abstract

The emergence of quantum computing and Shor's algorithm necessitates an imminent shift from current public key cryptography techniques to post-quantum robust techniques. NIST has responded by standardising Post-Quantum Cryptography (PQC) algorithms, with ML-KEM (FIPS-203) slated to replace ECDH (Elliptic Curve Diffie-Hellman) for key exchange. A key practical concern for PQC adoption is energy consumption. This paper introduces a new framework for measuring the PQC energy consumption on a Raspberry Pi when performing key generation. The framework uses both available traditional methods and the newly standardised ML-KEM algorithm via the commonly utilised OpenSSL library.

Figures (8)

• Figure 1: Combination of four different subject areas: the Compute Platform (Pi), Library (OpenSSL3.5), Key Exchange (Classical+PQC ML-KEM), Energy Use (measured via a TC66C meter).
• Figure 2: Block Diagram of the experimental setup - showing the Pi DUT which performs the key generation, and the Windows machine recording the electric supply characteristics from the TC66C meter. Trigger messages are conveyed using the network illustrated to START and STOP the measurements in line with the state of each experiment.
• Figure 3: Photograph of the experimental setup illustrating the Raspberry Pi 5 Device Under Test (DUT) and the TC66C test meter inline with the Power Supply for the Pi, measuring its energy usage characteristics.
• Figure 4: Plot of Energy Rate in Joules/1,000 key generations of each algorithm, categorised by NIST security level and type of algorithm. Experiment performed over a typical 500,000 key generation count for each algorithm. (A tablulated version is provided by Table \ref{['tab:energy_levels']}).
• Figure 5: Similar plot to Energy Utilisation in Figure \ref{['EnergyComparePlot']}, but detailing time to generate 1,000 keys for each algorithm on the Pi 5 Device Under Test.