Git Product home page Git Product logo

tss's Introduction

TSS: Threshold Signature Scheme library

Currently, the build-in supported schemes are:

  • ECDSA (binance's tss-lib)
  • EdDSA (binance's tss-lib)
  • BLS (builtin implementation)
  • PS (Pointcheval-Sanders)

Repository structure:

  • disc: Contains the discovery and synchronization module. Its role is to bootstrap the membership of the parties that are about to sign a message.

  • mpc: Contains implementations of various threshold signature schemes. Currently the only available one is the Binance ECDSA TSS scheme.

  • net: Contains an implementation of a network module. Its role is to provide end to end encrypted and authenticated communication.

  • rbc: Implements reliable byzantine fault tolerant broadcast. Its role is to ensure parties do not equivocate when broadcasting messages.

  • testutil: Contains utilities used in tests, mostly TLS key generation.

  • test: Contains integration tests that instantiate the TSS library with various threshold signature schemes found in mpc.

  • threshold: Contains the TSS library, which orchestrates instances of disc, mpc, rbc and optionally net.

  • types: Contains interface and type declaration.

Go Module structure:

The repository contains different go modules:

  • tss: The main module of the library, receives threshold signature schemes as a dependency injection.
  • mpc/binance: Wraps around the threshold signature scheme of binance-chain and presents an API that tss understands.
  • mpc/bls: Implements a threshold BLS using bn254 from Gnark-crypto.
  • test: Contains integration tests that instantiate tss with all implementations in mpc (currently only mpc/binance and mpc/bls)

The test module imports both tss and mpc/binance but neither tss nor mpc/binance do not import one another.

Using the library

Each node that may participate in the threshold key generation phase or in the threshold signature phase must instantiate an instance of threshold.Scheme. The instance can be obtained either by explicitly instantiating it and filling out its various public fields:

s := &threshold.Scheme{
   RBF: ...
   SyncFactory: ...
   ...
}

or alternatively by using the default construction method: threshold.DefaultScheme(...).

By initializing the threshold.Scheme instance explicitly, it is possible to pass different implementations of its underlying dependencies (rbc, disc, etc.).

It is the responsibility of the consumer to ensure that messages that arrive to the party are dispatched by the threshold.Scheme instance by calling HandleMessage(msg *IncMessage). The IncMessage struct appears below, and it is also the responsibility of the consumer to provide a secure, secret, and authenticated communication layer. The net package provides an implementation that fits the requirements, but it is not mandatory to use it.

type IncMessage struct {
	Data    []byte
	Source  uint16
	MsgType uint8
	Topic   []byte
}

What are universal identifiers and party identifiers?

In a decentralized setting, two or more nodes may belong to the same company, institution, or just be administered by the same entities.

One of the dependencies of the threshold.Scheme is a function that returns the membership:

func() map[UniversalID]PartyID

A UniversalID uniquely defines a node within a network. The library expects that nodes with different universal identifiers will have different endpoints, or even be different hosts.

On the other hand, a PartyID is a unique identifier of one or more universal identifiers that belong to the same organization. In every occurrence of a threshold protocol (either a key generation or a signature), each universal identifier may only correspond to a unique party identifier. See the following example for motivation and clarity:

Consider a scenario where we have a threshold key that is secret shared across three different companies: A, B, C. Each company runs a node and has its own universal identifier, and it receives a single (secret) share of the private key. Next, company A decides to replicate its secret share across three servers for high availability and load balancing. Since company A now has several servers, but only one server at a time may participate in signing (or key generation), the library needs to distinguish the servers by their identifiers.

However, in order for the mpc module to be simple, it shouldn't be aware of the fact that several identifiers may correspond to the same party. To that end, the mpc module uses PartyID for its membership, and the rest of the modules (rbc, disc) use UniversalID. The threshold package performs translation between the two spaces in a way that is transparent to the mpc module.

Note: Since the mpc module is pluggable and decoupled from this library, it does not reference the library. As a result, its API uses uint16 instead of PartyID:

Init(parties []uint16, threshold int, sendMsg func(msg []byte, isBroadcast bool, to uint16))

OnMsg(msgBytes []byte, from uint16, broadcast bool)

Generating a threshold public key

Before signing, the threshold public key needs to be generated by having the threshold.Scheme instance in each party invoke:

The totalParties is the number of shares this key needs to be split into, and threshold is the lowest number of shares for which reconstruction of the secret is not possible.

secretData, err := KeyGen(ctx context.Context, totalParties, threshold int) ([]byte, error)

Then, the returned byte slice secretData needs to be stored in a secure and reliable place.

Signing a message

After generating a public key, the threshold.Scheme instance needs to be initialized with the secret data. To do that, simply assign the StoredData field:

// Initialize the instance, either explicitly, or using the a constructor function as below
s := threshold.LoudScheme(...) // Or, threshold.SilentScheme(...)
// Assign the secret data returned from KeyGen()
s.SetStoredData(secretData)

Then, two operations are available:

  • ThresholdPK() ([]byte, error): Returns the serialized threshold public key, encoded by the mpc dependency injected.

  • Sign(c context.Context, msg []byte, topic string) ([]byte, error): Signs msg in the context of given topic. Returns the signature encoded by the mpc dependency injected. To avoid denial of service by malicious parties that haven't received msg, topic must be unpredictable.

Bootstrapping membership without communication

Before an instance of the TSS library can sign a message or generate a threshold key, it needs to discover who are the other parties that will participate in the protocol. Furthermore, the library needs to be instantiated at each node running the protocol, else messages will be lost. There are two ways of achieving this:

  1. Running a membership establishment protocol to have the parties discover each other and wait for each other to start

  2. Decide deterministically who are the parties that will sign a message given its topic, and temporarily store protocol messages in memory and then insert them once the instance has started its execution.

The first approach is implemented by the LoudScheme constructor method, while the second approach is implemented by the SilentScheme method.

Using threshold BLS:

When using BLS, the threshold.Scheme only orchestrates the key generation, but not the signing.

After generating the threshold key, persist the secret share of the party:

secretShareData, err := p.KeyGen(ctx, partynum, threshold)
saveToSafePlace(secretShareData)

Next, initialize a bls.TBLS instance and initialize it:

signer := &bls.TBLS{
	Logger: logger,
	Party:  uint16(id),
}

parties := []uint16{1, 2, 3}
signer.Init(parties, threshold, nil)
signer.SetShareData(secretShareData)

Get the public key from the initialized signer:

pk, err := signer.ThresholdPK()

And proceed to sign the message:

sig, err := signer.Sign(nil, msgHash)

Next, aggregate multiple signatures signed by distinct signers, but make sure each party identifier corresponds to the correct signature, and verify the threshold signature returned:

var v bls.Verifier
err = v.Init(pk)
sig, err := v.AggregateSignatures([][]byte{sig1, sig3, []uint16{1, 3})
err = v.Verify(msgHash, sig)

tss's People

Contributors

mbrandenburger avatar yacovm avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar

tss's Issues

go get github.com/IBM/TSS/mpc/binance/ecdsa facing issue.

When I try to go get the github.com/IBM/TSS/mpc/binance/ecdsa package, the following issue occurs

github.com/IBM/TSS/mpc/binance/ecdsa imports
github.com/bnb-chain/tss-lib/tss imports
github.com/btcsuite/btcd/btcec: cannot find module providing package github.com/btcsuite/btcd/btcec

Cannot recover the ethereum wallet address from the generated signature!!

Although, signature is verified successfully within the Libaray. I'ld like to verify inside my SignatureVerifiableContract.sol in which all the required functions for wallet recovery process is implemented. And I tested the signature generated using web3.js . Everything is working as expected. But the signature generated from the ecdsa protocol of this lib is not working. Even when I try to recover with web3.js, it throw exception with the message "invalid point".
Whats wrong?

DKG takes too long

DKG time consumption seems to increase exponentially as the number of nodes increases. DKG time consumption seems to increase exponentially as the number of nodes increases. When the total number of nodes is 20, the DKG time is 2 minutes, and when the number is 25, the time is one hour.

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.