As part of my master's degree in computer science, I am writing a project on the topic "Evaluation of future requirements for the energy efficiency of post-quantum cryptography (PQC) under consideration of green computing approaches". My basic assumption, which I would like to confirm, is that all PQC finalists (https://csrc.nist.gov/news/2022/pqc-candidates-to-be-standardized-and-round-4) can be exported on today's most common devices and that no new requirements need to be made. In addition, an extended result would be that, depending on the respective security level, criteria could be proposed to implement the algorithms with optimized parameters in terms of energy efficiency. If there are any basic advices on how I can check my thesis, I would be very happy to receive feedback! My biggest problem, however, is how I, as a "simple" student, can carry out the most scientifically relevant energy consumption measurements of PQC implementations without buying expensive measuring devices. Does anyone have any comments / experience in this area?
Some ideas: implement it on a "bare metal" system", so your measurements are not influenced by operations of the OS and the IO. Look for an evaluation board with a connection to measure power. As many manufactures see low powers as a sales point, they often have such a connection. You may have to add a low-pass filter (unless you have a more fancy integrator). Use 1 or 2 IO pins to signal the beginning and the end of the execution and a timer with sufficient resolution (or run the algorithm so many times that you can use a stopwatch :-).
If you want to measure a software implementation, modern 32 bit CPUs such as the ARM or RISC-V are readily available as embedded products, the x86 is somewhat rarer.
How to implement the software?
Naive approach: translate the specification into C code, optimize for speed and go! Modern C compilers are quite good at optimization. This may not deliver the most realistic results, see the next paragraph:
Most asymmetric algorithms have some core components that use most of the calculation resources, for example modular multiplication in the classical algorithms. Use an optimized library for these. Use a "profiler" to find where the algorithm spends the most time, this is an area for optimization. Look a the assembly code and check if the compiler did something stupid.
Assuming that PQ really takes off, some SoC vendors may implement part of it directly in silicon,so an FPGA might be the instrument of choice. If the algorithm is simple enough, you could implement the whole thing in VHDL or Verilog. Or implement only the heavy lifting in VHDL and use a CPU macro to do the complex but less resource intensive steps.
The hardware will cost about 10 Euro for a CPU to 50 Euro for an FPGA on Aliexpress (plan up to 5 weeks delivery time).
Hi Ronny Liebert
Given that two algorithms use a quantum processing unit (QPU) and a classical processing unit (CPU) to solve the same problem, utilising a green energy source. A green quantum advantage threshold can be determined when the computation cost of the former is more efficient than the latter under the same accuracy achieved. To provide a road map for answering your thesis question, we need to differentiate between three phases through the post-quantum cryptography adoption life-cycle:
1. Development: here, we are discussing developing an algorithm for standardisation (e.g. Crystals-Kayber) using NISQ hardware. Since there is more than one factor that can affect energy consumption experimentally, we can presume that the question of green energy may be answered through a literature review. For instance, Jaschke and Montangero - attached, provide the foundation to answer the green energy question through theoretical and empirical estimations based on the equality of fidelity and energy point (EFEP) between a Matrix Product State (MPS) and the quantum hardware via a criterion that can be calculated a priori as a function of the number of qubits, the generated entanglement, and the QPU’s gate properties to name the most important parameters. Also, they offered different simulation approaches, including the Rydberg quantum simulator and the trapped ion simulator. Finally, they provided a practical approach to using real hardware.
2. Implementation: IBM Quantum-Safe Explorer is a tool to implement PQC in vulnerable applications in the Cryptography Bill of Materials (CBoM). The high-level implementation steps are explained in the attachment. Please be advised that IBM developed three out of four algorithms approved by NIST. Also, we need to decide whether PQC will be layered on top of Quantum Key Distribution (QKD). The energy consumption estimation can be achieved by applying the abovementioned techniques for quantum hardware use.
3. Testing (hacking): this can be estimated, too. I am attaching a journal article that I hope you find useful.
All the best
Hi Ronny Liebert
Given that two algorithms use a quantum processing unit (QPU) and a classical processing unit (CPU) to solve the same problem, utilising a green energy source. A green quantum advantage threshold can be determined when the computation cost of the former is more efficient than the latter under the same accuracy achieved. To provide a road map for answering your thesis question, we need to differentiate between three phases through the post-quantum cryptography adoption life-cycle:
1. Development: here, we are discussing developing an algorithm for standardisation (e.g. Crystals-Kayber) using NISQ hardware. Since there is more than one factor that can affect energy consumption experimentally, we can presume that the question of green energy may be answered through a literature review. For instance, Jaschke and Montangero - attached, provide the foundation to answer the green energy question through theoretical and empirical estimations based on the equality of fidelity and energy point (EFEP) between a Matrix Product State (MPS) and the quantum hardware via a criterion that can be calculated a priori as a function of the number of qubits, the generated entanglement, and the QPU’s gate properties to name the most important parameters. Also, they offered different simulation approaches, including the Rydberg quantum simulator and the trapped ion simulator. Finally, they provided a practical approach to using real hardware.
2. Implementation: IBM Quantum-Safe Explorer is a tool to implement PQC in vulnerable applications in the Cryptography Bill of Materials (CBoM). The high-level implementation steps are explained in the attachment. Please be advised that IBM developed three out of four algorithms approved by NIST. Also, we need to decide whether PQC will be layered on top of Quantum Key Distribution (QKD). The energy consumption estimation can be achieved by applying the abovementioned techniques for quantum hardware use.
3. Testing (hacking): this can be estimated, too. I am attaching a journal article that I hope you find useful.
All the best
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence