This repository implements the code to benchmark the computation costs of the BLS threshold signature from the paper titled "Adaptively Secure BLS Threshold Signatures from DDH and co-CDH".

Paper link: https://eprint.iacr.org/2023/1553

Code structure

The relevant implementations are part of the src/ directory, and the directory has the following structure.

    ├── README.md
    └── src
        ├── adaptive_bls.go             // implements new BLS threshold signatures
        ├── adaptive_bls_test.go        // implements the tests and benchmarking code for our scheme
        ├── boldyreva.go                // implements both Boldyreva-I (RO based DLEQ verification) and Boldyreva-II (pairing based verification)
        ├── boldyreva_test.go           // implmenets the tests and benchmarking code for Boldyreva-I and Boldyreva-II
        ├── utils.go                    // implements some common interfaces
        └── utils_test.go               // implements test case for our common funcitionalities

The code has been tested on a M2-pro Apple laptop with go version go1.21.5 darwin/arm64

Benchmarking of BLS threshold signatures

To benchmark the code go to the src/ directory and run the following commmands.

To benchmark boldyreva run

go test -cpu 1 -benchmem -run=^$ -bench BenchmarkBLS

Here -cpu [NUM_CPU] lets us configure the number of CPUs used for evaluating the schemes.

The benchmakr outputs the following results:

1. For Boldyreva-I
   • B1-pSign measures the partial signing time
   • B1-pVerify measures the partial signature verification time
   • [T]-B1-agg measures the time to aggregate partial signatures from T signers. Our benchmark outputs numbers for a threshold of 64, 256, and 1024.
2. For Boldyreva-II
   • B2-pSign measures the partial signing time
   • B2-pVerify measures the partial signature verification time
   • [T]-B2-agg measures the time to aggregate partial signatures from T signers
3. Common
   • [T]-verify measures the cost to verify the final signatures with T signers. This is also applicable to our scheme.

To benchmark adaptive BLS run

go test -cpu 1 -benchmem -run=^$ -bench BenchmarkABLS

The benchmakr outputs the following results:

1. Our scheme
   • ABLS-pSign measures the partial signing time
   • ABLS-pVerify measures the partial signature verification time
   • [T]-ABLS-agg measures the time to aggregate partial signatures from T signers